High-resolution quantitative and functional MRI indicate lower myelination of thin and thick stripes in human secondary visual cortex

Daniel Haenelt; Robert Trampel; Shahin Nasr; Jonathan R Polimeni; Roger BH Tootell; Martin I Sereno; Kerrin J Pine; Luke J Edwards; Saskia Helbling; Nikolaus Weiskopf

doi:10.7554/eLife.78756

. 2023 Mar 8;12:e78756. doi: 10.7554/eLife.78756

High-resolution quantitative and functional MRI indicate lower myelination of thin and thick stripes in human secondary visual cortex

Daniel Haenelt ^1,^2,^✉, Robert Trampel ¹, Shahin Nasr ^3,⁴, Jonathan R Polimeni ^3,^4,⁵, Roger BH Tootell ^3,⁴, Martin I Sereno ⁶, Kerrin J Pine ¹, Luke J Edwards ¹, Saskia Helbling ^1,⁷, Nikolaus Weiskopf ^1,⁸

Editors: Kristine Krug⁹, Floris P de Lange¹⁰

PMCID: PMC9995117 PMID: 36888685

Abstract

The characterization of cortical myelination is essential for the study of structure-function relationships in the human brain. However, knowledge about cortical myelination is largely based on post-mortem histology, which generally renders direct comparison to function impossible. The repeating pattern of pale-thin-pale-thick stripes of cytochrome oxidase (CO) activity in the primate secondary visual cortex (V2) is a prominent columnar system, in which histology also indicates different myelination of thin/thick versus pale stripes. We used quantitative magnetic resonance imaging (qMRI) in conjunction with functional magnetic resonance imaging (fMRI) at ultra-high field strength (7 T) to localize and study myelination of stripes in four human participants at sub-millimeter resolution in vivo. Thin and thick stripes were functionally localized by exploiting their sensitivity to color and binocular disparity, respectively. Resulting functional activation maps showed robust stripe patterns in V2 which enabled further comparison of quantitative relaxation parameters between stripe types. Thereby, we found lower longitudinal relaxation rates (R₁) of thin and thick stripes compared to surrounding gray matter in the order of 1–2%, indicating higher myelination of pale stripes. No consistent differences were found for effective transverse relaxation rates (R₂*). The study demonstrates the feasibility to investigate structure-function relationships in living humans within one cortical area at the level of columnar systems using qMRI.

Research organism: Human

Introduction

In primates, visual information sent from the primary visual cortex (V1) to the secondary visual cortex (V2) is segregated into distinct modules known as thin, thick, and pale stripes (Hubel and Livingstone, 1987; Livingstone and Hubel, 1987). These stripes form a columnar system in the sense that their functional properties extend roughly through cortical depth (Tootell and Hamilton, 1989). Functional properties include the sensitivity to different visual features like color, orientation, binocular disparity, and motion, which are largely processed in different stripe types and sent to distinct cortical areas. For example, thin stripes are sensitive to color content and project to functional area V4, whereas thick stripes are more sensitive to binocular disparity and project to area MT (V5) (Shipp and Zeki, 1985; Hubel and Livingstone, 1987; Livingstone and Hubel, 1987; Ts’o et al., 2001).

Using cytochrome oxidase (CO) staining, these stripes were found first in squirrel monkeys and macaques as dark and pale patches organized in repeating pale-thin-pale-thick cycles, running through V2 and oriented approximately perpendicular to the V1/V2 border (Livingstone and Hubel, 1982; Tootell et al., 1983). In macaques, stripes of the same type have a center-to-center distance of around 4.0 mm and a width ranging from 0.7 to 1.3 mm (Shipp and Zeki, 1985; Tootell and Hamilton, 1989). In humans, these widths are approximately doubled in size (Hockfield et al., 1990; Tootell and Taylor, 1995; Adams et al., 2007).

Histological studies also showed a stripe pattern in V2 of post-mortem brain specimens when techniques for the staining of myelin were used (Tootell et al., 1983; Krubitzer and Kaas, 1989; Horton and Hocking, 1997). However, these studies gave an inconsistent picture of the correspondence between stripes defined by CO activity and myelin density. Staining with Luxol fast blue indicated stronger myelination in thin/thick (Tootell et al., 1983) stripes, while Gallyas silver staining showed pale (Krubitzer and Kaas, 1989) stripes being more myelinated. This discrepancy between myelin staining methods was replicated in another study in which several methods were compared to each other (Horton and Hocking, 1997). In addition to inconsistencies across staining methods, all standard histochemical methods are highly sensitive to the condition of the brain specimen (e.g. post-mortem delay time), variations in fixation and staining procedures, and exposure time (Savaskan et al., 2009).

Magnetic resonance imaging (MRI) provides an alternative view to study tissue microstructure in living humans with the possibility to generate a multitude of image contrasts which can specifically be sensitized to myelin (Edwards et al., 2018; Weiskopf et al., 2021). Quantitative MRI (qMRI) is a collective term of techniques that aims to isolate one source of image contrast and represent it as a single quantitative parameter map, for example, maps of longitudinal relaxation rate (R₁), effective transverse relaxation rate (R₂*), proton density (PD), or magnetic susceptibility. Parameter maps are created by combining multiple ‘weighted’ MR images in a model-based fashion yielding reproducible and standardized measures in physical units, which are less dependent on the acquisition (Edwards et al., 2018; Trampel et al., 2019; Weiskopf et al., 2021). Furthermore, by separating different sources of image contrast, qMRI provides a closer relationship to specific microstructural components, for example, myelin and iron (Edwards et al., 2018; Weiskopf et al., 2021). In this regard, it should be kept in mind that single qMRI parameter maps do not have the specificity to directly infer the abundance of single microstructural tissue components by being sensitive to multiple tissue components to a varying extent and therefore can only serve as a surrogate, for example, in the study of cortical myelination (Lazari and Lipp, 2021). Nevertheless, concerning R₁, it is known that cortical R₁ is largely affected by myelin content (Stüber et al., 2014; Leuze et al., 2017) and, hence, was often used as surrogate to study the macroscopic distribution of cortical myelination (Sigalovsky et al., 2006; Dick et al., 2012; Sereno et al., 2013; Lutti et al., 2014). In addition, it might be possible to exploit the complementary multi-modal information from different qMRI parameter maps using appropriate biophysical models to directly quantify the underlying tissue microstructure with higher specificity, which would open the way to in vivo histology (Weiskopf et al., 2021).

Functional MRI (fMRI) allows in vivo localization of functional architecture. Recent developments in ultra-high field MRI enabled the functional localization of thin and thick stripes using high-resolution fMRI (Nasr et al., 2016; Dumoulin et al., 2017; Navarro et al., 2021) by, for example, exploiting their different sensitivity to color (Tootell et al., 1983; Tootell et al., 2004) and binocular disparity (Peterhans and von der Heydt, 1993; Chen et al., 2008), respectively (Nasr et al., 2016). This allows MR-based investigations of mesoscale structure-function relationships in the same living participant, that is, at the spatial scale of cortical columns and layers.

We combined the localization of V2 stripes using high-resolution fMRI with qMRI measurements to infer myelination differences between stripe types. We robustly show lower R₁ values in color-selective thin and disparity-selective thick stripes in comparison to locations containing pale stripe contributions, which points toward higher myelin density in pale stripes. Whereas recent studies have explored cortical myelination in V2 in macaques (Li et al., 2019) and humans (Dumoulin et al., 2017) using non-quantitative, weighted MR images, to the best of our knowledge, we showed for the first time myelination differences using MRI on a quantitative basis at the spatial scale of columnar systems. This shows the feasibility to use high-resolution qMRI in conjunction with high-resolution fMRI to study the relationship between functional and structural properties of the brain in living humans, which is a fundamental goal in neuroscience.

Results

Participants (n = 4) were invited for multiple fMRI and qMRI sessions at 7 T (see Figure 1). On different days, we measured high-resolution (0.8 mm isotropic) fMRI responses to stimuli varying in color and binocular disparity content, respectively, to locate color-selective thin stripes (color stripes) and disparity-selective thick stripes (disparity stripes) in V2 (Nasr et al., 2016). We use the terms color-selective thin and disparity-selective thick stripes acknowledging the close relationship between color processing and CO thin stripes (Tootell et al., 2004) and disparity processing and CO thick stripes (Chen et al., 2008), respectively, despite not having directly measured CO content in this study. In a separate session, we used the multi-parameter mapping (MPM) protocol (Vaculčiaková et al., 2022) to acquire high-resolution anatomical images with 0.5 mm isotropic resolution from which quantitative parameter maps (R₁, R₂*, PD) were derived.

Functional mapping of color-selective and disparity-selective stripes

Color- and disparity-selective stripes were identified in each individual in separate scanning sessions. Figure 2 shows activation maps averaged over two sessions and sampled at mid-cortical depth from one representative participant (see Appendix 1 for activation maps from all participants).

Figure 2—figure supplement 1. — Thin stripes (contrast: color > luminance) and thick stripes (contrast: depth > no depth) are shown as thresholded activation maps in (a) and (b), respectively. Both maps were averaged across sessions, sampled at mid-cortical depth, and are illustrated on the flattened surface of the right hemisphere for one representative participant (subject 3). Surfaces were flattened using FreeSurfer (6.0.0, http://surfer.nmr.mgh.harvard.edu/) after cutting out a region on the surface mesh which included all stimulated portions of V1 and V2. Data from all participants can be found in Appendix 1. In V2, patchy stripes can be identified, which run through V2 oriented perpendicular to the V1/V2 border. Borders were manually defined on the basis of a separate retinotopy measurement, which is illustrated in Figure 2—figure supplement 1 (see Appendix 1 for retinotopy data from all participants). The white border illustrates a region around the foveal representation with missing activation. This region was excluded in the supporting analysis shown in Figure 5—figure supplement 2. For the main analysis, all data points in V2 were used. Manually drawn cyan dots mark activated regions in (a) to illustrate the alternating activation pattern between (a) and (b). Cortical curvature is shown in gray (sulcal cortex is dark gray and gyral cortex light gray). Another example can be found in Figure 2—figure supplement 2. Scale bar: 1 cm.

Color-selective thin stripes can be identified in Figure 2a with expected topography (Tootell et al., 1983; Nasr et al., 2016), that is, they start at the V1/V2 border, radiate outward in parallel, and are confined to area V2. Figure 2b shows locations selective for binocular disparity. Activation maps for binocular disparity showed a less pronounced stripe pattern in V2. It should be noted that color-selective stripes are known to be largely confined to CO thin stripes (Xiao et al., 2003; Tootell et al., 2004), whereas selectivity for binocular disparity is found in all stripe types but most frequently in CO thick stripes (Peterhans and von der Heydt, 1993; Chen et al., 2008). In V1, no activation was found for binocular disparity in Figure 2b which is consistent with findings by Tsao et al., 2003; Nasr et al., 2016, while large V1 activation was found for color contrast in Figure 2a as also shown by Nasr et al., 2016.

Cyan dots were added in Figure 2 to qualitatively illustrate the alternation of activation clusters between stripe types as expected from the thin/thick stripe arrangement. We note that, as Figure 2a and Figure 2b show results from two independent experiments, the alternating stripe pattern is not an intrinsic outcome of the experimental design.

Each stripe type was localized in two independent scanning sessions and activation maps were consistent between sessions of color and disparity stripe measurements, respectively. This is illustrated in Figure 3, which shows statistically significant correlations of activation maps between sessions for one representative participant.

Figure 3. — Scatter plots with kernel density estimation illustrate the consistency of activation maps across scanning sessions for one representative participant (subject 3). Sessions were carried out on different days and activation maps were sampled at mid-cortical depth. (a) shows correspondences of $z$ -scores in the secondary visual cortex (V2) between single color-selective thin stripe mapping sessions (contrast: color > luminance). (b) shows the same for single disparity-selective thick stripe mapping sessions (contrast: depth > no depth). In (c), correspondences of average $z$ -scores (across sessions) between thin and thick stripe sessions are shown. Regression lines are indicated as red lines. Spearman’s rank correlation coefficients $r$ and $p$ -values determined by permutation analysis (see Materials and methods) are annotated inside the plots and demonstrate high repeatability of color-selective thin and disparity-selective thick stripe scanning sessions. Note that the comparison between thin and thick stripe sessions shows no statistically significant correlation as expected from the interdigitated nature of both stripe types. Plots for all participants can be found in Figure 3—figure supplement 1 and Figure 3—figure supplement 2.

Figure 3—figure supplement 1. — Scatter plots with kernel density estimation illustrate the consistency of activation maps across scanning sessions for one representative participant (subject 3). Sessions were carried out on different days and activation maps were sampled at mid-cortical depth. (a) shows correspondences of $z$ -scores in the secondary visual cortex (V2) between single color-selective thin stripe mapping sessions (contrast: color > luminance). (b) shows the same for single disparity-selective thick stripe mapping sessions (contrast: depth > no depth). In (c), correspondences of average $z$ -scores (across sessions) between thin and thick stripe sessions are shown. Regression lines are indicated as red lines. Spearman’s rank correlation coefficients $r$ and $p$ -values determined by permutation analysis (see Materials and methods) are annotated inside the plots and demonstrate high repeatability of color-selective thin and disparity-selective thick stripe scanning sessions. Note that the comparison between thin and thick stripe sessions shows no statistically significant correlation as expected from the interdigitated nature of both stripe types. Plots for all participants can be found in Figure 3—figure supplement 1 and Figure 3—figure supplement 2.

Consistent qMRI maps across cortical regions and cortical depth

Figure 4a–b shows an R₁ map sampled at mid-cortical depth for one representative participant. Primary motor and primary sensory cortical areas have higher R₁ values, congruent with higher myelin density in these areas (Flechsig, 1920; Glasser and Van Essen, 2011; Sereno et al., 2013). To further check the consistency of our data with literature, we qualitatively compared cortical mean R₁ parameters between several cortical regions of interest (ROIs) with known myelination differences. ROIs were defined by probabilistic FreeSurfer (6.0.0, http://surfer.nmr.mgh.harvard.edu/) labels for each participant. First, we used the FreeSurfer Brodmann area maps of V1, V2, and MT (V1_exvivo.thresh.label, V2_exvivo.thresh.label and MT_exvivo.thresh.label) (Fischl et al., 2008; Hinds et al., 2008). Second, we defined an angular gyrus label from the FreeSurfer parcellation (Destrieux et al., 2010).

Cortical R₁ values are shown at mid-cortical depth of the left hemisphere on an inflated surface from a representative participant (subject 3) in lateral (a) and posterior (b) view. Higher R₁ values can be qualitatively identified in primary motor and sensory areas, which reflects known cortical myeloarchitecture (Flechsig, 1920; Glasser and Van Essen, 2011). The arrow in (a) points to an artifact outside of V2 caused by magnetic field inhomogeneities. Data from all participants can be found in Appendix 1. (c) Mean R₁ values are shown for different cortical regions (angular gyrus, MT, V2, V1) defined by corresponding FreeSurfer labels (Fischl et al., 2008; Hinds et al., 2008; Destrieux et al., 2010) of each participant (similar to Fig. 1(b) in Sereno et al., 2013). All participants show increased R₁ values in V1. Corresponding source data is given in Figure 4—source data 1. A corresponding illustration for quantitative R₂* can be found in Figure 4—figure supplement 1. Figure 4—figure supplement 2 shows mean R₁ (based on a separate whole-brain MP2RAGE acquisition) and PD values across cortical areas. Higher R₁ values in V1 as shown in (c) could be confirmed with the independent R₁ estimate from the MP2RAGE acquisition. Mean across participants is shown in gray. Vertical error bars indicate 1 standard deviation across participants.

Figure 4—source data 1. Source data for mean quantitative magnetic resonance imaging (qMRI) values (R₁, R₂*, PD) across cortical areas (angular gyrus, MT, V2, V1).

elife-78756-fig4-data1.xlsx^{(6KB, xlsx)}

Figure 4c shows systematic R₁ variations with highest values in V1 for each participant, which is in line with Fig. 1(b) in Sereno et al., 2013. Figure 4—figure supplement 1 and Figure 4—figure supplement 2 illustrate the same comparison for R₂* and PD values. Whereas R₂* values showed similar results, PD lacked a consistent trend across participants. This might be due to remaining receiver bias in final PD maps, which is challenging to remove especially at high magnetic field strengths (Volz et al., 2012). We therefore did not consider PD parameter maps for the main analysis. We also checked cortical profiles of mean parameters in V2 by sampling data on surfaces defined at different cortical depths (see Appendix 2). In all participants, we confirmed the expected decrease of R₁, R₂*, and MTVF = 100% − PD (macromolecular tissue volume fraction [Mezer et al., 2013]) values toward the pial surface since all three parameters are sensitive to myelin (Marques et al., 2017; Carey et al., 2018; Kirilina et al., 2020).

Higher myelination of pale stripes

Using R₁ and R₂* as surrogates for cortical myelination, we tested whether different stripe types are differentially myelinated by comparing R₁ and R₂* parameter values between stripe types following a similar procedure described in Li et al., 2019. In brief, color-selective thin and disparity-selective thick stripe ROIs were demarcated by applying a $z$ -score threshold to the corresponding functional contrasts (see Appendix 3 for a depiction of ROIs for each participant defined at $z = 1.96$ ). Mean R₁ and R₂* from one stripe type were then tested against the mean value within V2 excluding data belonging to the other stripe type (see Materials and methods). This enabled us to indirectly demarcate pale stripes assuming a strict tripartite stripe division of V2. Since the definition of ROIs solely based on $z$ -score thresholds is inevitably arbitrary, we performed the above analysis for several thresholds. Figure 5 shows the pooled R₁ and R₂* for $z \in {0, 0.5, \dots, 4.5}$ across participants. Quantitative parameter values are shown as deviation from the mean within V2 after regressing out variations due to local curvature. For each $z$ -score threshold level, we tested the difference for statistical significance using permutation testing. Figure 5a–b shows statistically significant differences of R₁ between thin or thick stripes and mean of V2 excluding the other stripe type, which points toward higher myelin density in pale stripes. These results were confirmed by an independent data set using R₁ values estimated from the MP2RAGE sequence (Marques et al., 2010), which is shown in Figure 5—figure supplement 2. The maximum $z$ -score threshold was chosen arbitrarily and was limited by the resulting ROI size. ROI sizes for all threshold levels and participants are illustrated in Appendix 4. Note that higher thresholds lead to an expansion of the pale stripe ROI and contamination from other stripe types. In Figure 5, shaded areas denote the standard deviation of the generated null distribution used for permutation testing. This illustrates the enlargement of pale stripe ROIs at high threshold levels since larger ROIs lead to less variation across permutations. For an intermediate threshold level of $z = 1.96$ ( $p < 0.05$ , two-sided), R₁ values in thin and thick stripes differ from pale stripes by 0.005 s⁻¹ and 0.014 s⁻¹, respectively, which corresponds to a deviation of around 1–2% assuming a mean longitudinal relaxation rate of 0.58 s⁻¹ in V2 (see Figure 4c). No statistically significant effects were found for R₂* as shown in Figure 5c–d.

Figure 5. — Cortical R₁ (a–b) and R₂* (c–d) values in thin stripes (red), thick stripes (blue), and whole V2 excluding the other stripe type (gray; and therefore containing contributions from pale stripes) are shown for various $z$ -score threshold levels, which were used to define thin and thick stripe regions of interest (ROIs). Quantitative values are illustrated as deviation from the mean within V2 after removing variance from local curvature. Values were pooled across participants and hemispheres. Differences between data in thin/thick stripes and whole V2 without thick/thin stripes were tested for statistical significance at $z \in {0, 0.5, \dots, 4.5}$ . Statistical significance was assessed by permutation testing (see Materials and methods). R₁ in both thin and thick stripes is lower than surrounding gray matter, which points toward higher myelination of pale stripes assuming a strict tripartite stripe division in human V2. No effects were found for R₂*. A supporting analysis confirmed these results when excluding the region around the foveal representation from the analysis. This addresses a potential bias due to missing activation around the foveal representation in functional activation maps, which were used to define stripe ROIs. This is shown in Figure 5—figure supplement 1. The results for R₁ values were further confirmed using an independent estimate of cortical R₁ based on separately acquired whole-brain MP2RAGE scans, which is shown in Figure 5—figure supplement 2 and Figure 5—figure supplement 3. Corresponding source data is given in Figure 5—source data 1. Statistically significant differences are marked by asterisks, *: $p < 0.05$ , **: $p < 0.01$ , ***: $p < 0.001$ . Shaded areas indicate 1 standard deviation of the generated null distribution used for permutation testing.

Figure 5—source data 1. Source data for comparisons of R₁ and R₂* between V2 stripe types.

elife-78756-fig5-data1.xlsx^{(17.4KB, xlsx)}

Cortical R₁ (a–b) and R₂* (c–d) values in thin stripes (red), thick stripes (blue), and whole V2 excluding the other stripe type (gray; and therefore containing contributions from pale stripes) are shown for various $z$ -score threshold levels, which were used to define thin and thick stripe regions of interest (ROIs). Quantitative values are illustrated as deviation from the mean within V2 after removing variance from local curvature. Values were pooled across participants and hemispheres. Differences between data in thin/thick stripes and whole V2 without thick/thin stripes were tested for statistical significance at $z \in {0, 0.5, \dots, 4.5}$ . Statistical significance was assessed by permutation testing (see Materials and methods). R₁ in both thin and thick stripes is lower than surrounding gray matter, which points toward higher myelination of pale stripes assuming a strict tripartite stripe division in human V2. No effects were found for R₂*. A supporting analysis confirmed these results when excluding the region around the foveal representation from the analysis. This addresses a potential bias due to missing activation around the foveal representation in functional activation maps, which were used to define stripe ROIs. This is shown in Figure 5—figure supplement 1. The results for R₁ values were further confirmed using an independent estimate of cortical R₁ based on separately acquired whole-brain MP2RAGE scans, which is shown in Figure 5—figure supplement 2 and Figure 5—figure supplement 3. Corresponding source data is given in Figure 5—source data 1. Statistically significant differences are marked by asterisks, *: $p < 0.05$ , **: $p < 0.01$ , ***: $p < 0.001$ . Shaded areas indicate 1 standard deviation of the generated null distribution used for permutation testing.

Figure 5—source data 1. Source data for comparisons of R₁ and R₂* between V2 stripe types.

elife-78756-fig5-data1.xlsx^{(17.4KB, xlsx)}

From functional contrasts shown in Figure 2 (see also Appendix 1 and Appendix 3), it is evident that the region around the foveal representation of V2 largely did not show functional activation. Since we used all portions of V2 in the analysis which were covered by retinotopy measurements, this potentially could have biased our results. To check if missing activation around the foveal representation affected our results, we performed an additional analysis, in which we excluded that foveal region. More precisely, we used the phase responses from retinotopy measurements (eccentricity) and constructed masks to exclude the inner one-third of the full phase cycle in individual activation maps (see white lines in Figure 2 and Appendix 1). The excluded eccentricity range was a compromise between masking out a sufficient region around the foveal representation with no functional activation and leaving enough data points for analysis. Comparison of qMRI parameters yielded basically the same results both for MPM and for MP2RAGE data shown in Figure 5—figure supplement 1 and Figure 5—figure supplement 3, respectively. Note that Figure 5—figure supplement 1 also shows significant differences for R₂* which also points into the direction of higher myelin content of pale stripes without being as conclusive as R₁.

Discussion

The secondary visual cortex of the primate contains a repeating pale-thin-pale-thick stripe pattern of CO activity. It is known that components of visual information like color, orientation, and binocular disparity are largely segregated into separate pathways and processed in different stripe types (Hubel and Livingstone, 1987; Livingstone and Hubel, 1987). We robustly mapped color-selective thin and disparity-selective thick stripes in humans using high-resolution fMRI. Furthermore, we demonstrated that R₁ parameter maps reflect the expected macroscopic features of cortical myelination in primary motor and sensory areas and across cortical depth. By combining data from fMRI and qMRI, we then showed that locations in V2 have higher R₁ values which neither correspond to functionally defined thin nor thick stripes. Because myelin content is a major contrast mechanism for cortical R₁ (Stüber et al., 2014) and we expect a tripartite stripe architecture, we interpret this as an indication of higher myelination of pale stripes than thin and thick stripes in V2.

It is important to note that R₁ is used as a surrogate for cortical myelination and histological findings of myelin content in V2 are not conclusive so far. On the one hand, Tootell et al., 1983, showed higher myelination in dark CO stripes in monkey V2 using the Luxol fast blue staining technique and using wet unstained brain tissue sections. On the other hand, Krubitzer and Kaas, 1989 showed higher myelination in pale CO stripes using the Gallyas staining technique. These conflicting findings were reproduced in a later study by Horton and Hocking, 1997. This study also examined myelination in monkey V1 after monocular enucleation. After monocular enucleation, CO activity is down-regulated in layer 4c neurons located in ocular dominance columns (ODCs) of the deafferented eye. This enables the visualization of ODCs of the missing eye, which appear as light columns after CO staining. Monocular enucleation further leads to atrophy of the optic nerve and loss of myelin through the process of Wallerian degeneration. Unexpectedly, Horton and Hocking, 1997, showed higher myelination in ODCs of the missing eye in wet unstained tissue and using myelin basic protein immunohistochemistry. Furthermore, wet unstained sections showed higher myelination in V2 dark CO stripes of normal monkeys. Both findings would be in conflict with each other if a relationship between patches/stripes of CO activity and myelination is assumed. Both classical myelin stains (Luxol fast blue and Gallyas) showed the expected higher myelination in ODCs of the intact eye. However, it is known that Luxol fast blue, which stained CO-rich regions in both V1 and V2, is not only sensitive to myelin content but also shows an affinity to mitochondria where CO is mostly found. This dual character might have influenced the staining results in CO-rich cortical regions of V1 and V2. The Gallyas staining showed the expected higher myelination in ODCs of the normal eye after monocular enucleation and higher myelination in V2 pale stripes. While this does not reveal a complete and conclusive picture, we reproduced the results obtained with the Gallyas staining method using qMRI. Our results were further corroborated by a recent MRI study in macaques (Li et al., 2019), which used a similar approach to define thin and thick stripes as in the present study.

However, our results do not align with a recent human MRI study by Dumoulin et al., 2017, which found higher myelin density in thick stripes using a T₁-weighted imaging sequence (MPRAGE) to infer myelination differences. Although the reason for this discrepancy cannot be conclusively determined, two aspects that differ between studies are worth mentioning. First, the functional localization of stripes was different. In Dumoulin et al., 2017, parvo- and magnocellular dominated pathways were targeted by exploiting known differences in the processing of slow and fast temporal frequencies in the visual stimulus, respectively. However, the assignment of parvo- and magnocellular streams to particular stripes in V2 is still controversial (Sincich and Horton, 2005). Thus, their relation to the tripartite stripe architecture is less clear than for color content and binocular disparity as used in our study (Hubel and Livingstone, 1987; Tootell et al., 2004; Chen et al., 2008). Second, in contrast to our study, myelin density was inferred from weighted MR images, which are known to be more affected by technical biases.

The observed differences of myelin density between stripe types were based on R₁ estimates using the MPM protocol in the main analysis. We further confirmed these results with an independent data set using R₁ estimates from the MP2RAGE acquisition (see Figure 5—figure supplement 2). This further demonstrates the generalizability of our results across acquisition methods. Nevertheless, although previous studies show the validity of R₁ as a myelin marker in this type of healthy participant group with no known pathologies (Stüber et al., 2014; Leuze et al., 2017) and differences were found with two different R₁ mapping methods, we cannot fully exclude that some unknown factors may systematically bias the indirect measures.

We did not find any significant differences of R₂* between stripe types in the main analysis as shown in Figure 5c–d. Whereas R₁ in the normal cortex is largely influenced by myelination levels, R₂* is sensitive to paramagnetic iron and diamagnetic myelin (Stüber et al., 2014; Kirilina et al., 2020). Other factors like vasculature and the orientation of the cortex to the static magnetic field of the MR system have an influence on R₂*, which might have obscured the underlying dependency on myelin content (Cohen-Adad et al., 2012). The dark appearance of thin and thick stripes after CO staining is a marker for increased oxidative metabolism compared to pale stripes. This favors the hypothesis of richer vascularization in thin and thick stripes, which potentially could counteract reductions in R₂* due to lower myelination. Indeed, higher vessel densities were found in blobs (another CO-rich structure in V1) and stripes of squirrel and macaque monkeys (Zheng et al., 1991; Keller et al., 2011). However, this was later disputed by another study, which showed no differences in vascular supply between blobs and inter-blobs in V1 (Adams et al., 2015).

Figure 5a–b shows that R₁ differences are in the range of around 1–2%. This is smaller but comparable to R₁ differences between cortical areas, which are in the range of a few percent at 7 T (Marques et al., 2017). The required high spatial resolution and the measured small effect sizes limited the contrast-to-noise ratio in our measurements. This most probably hindered a direct visualization of stripes at the voxel level in R₁ maps as illustrated in Figure 4 (see Appendix 1 for visualization of cortical R₁ and R₂* in V1 and V2 for each participant) and required pooling of data within stripe types defined by fMRI followed by a statistical analysis. The coefficient of variation in R₁ maps was 11.3 ± 0.7 (mean ± standard deviation across participants) in V2. However, when data were pooled across participants, we could robustly detect significant differences between stripe types. We believe that the usage of qMRI parameters at this spatial scale in living humans is favorable as these parameters are less affected by technical biases and are in physical units and thus more accessible for biophysical modeling, which will facilitate studies of tissue microstructure with higher specificity.

The following considerations are related to the definition of the stripe ROIs. We could only functionally delineate thin and thick stripes but not pale stripes. Thus, the definition of pale stripes was indirect and relied on the assumption of a tripartite architecture.

In the analysis, the whole stimulated part of V2, which could be defined by retinotopy, was considered. However, the paradigms used for localization of color and disparity stripes did not show pronounced activation at the representation of the central fovea (see Figure 2 and Appendix 1). First, the color stimulus with red/blue gratings (see Figure 1a) is expected to have a different effect in the central fovea than in parafoveal regions due to the macula lutea (yellow pigmented spot of the retina) and absence of blue cones in the central fovea, which might have hindered the detection of color stripes there (Nasr and Tootell, 2018). Second, missing activation at the representation of the central fovea for the disparity stimulus could be due to eccentricity dependence of disparity tuning. Other possible explanations might be a smaller fusion limit in central fovea or a global attention mechanism, which might preferentially activate peripheral representations (Tsao et al., 2003). Using conventional fMRI with lower resolution, Tsao et al., 2003, found an overall similar eccentricity-dependent activation pattern for a stimulation with the same maximal disparity (±0.22°). To address the possibility that our results were biased by eccentricity, we repeated the analysis shown in Figure 5 but excluded the foveal representation of V2, which yielded basically the same results (see Figure 5—figure supplement 1 and Figure 5—figure supplement 3).

Figure 2 shows that activation maps for color-selective thin and disparity-selective thick stripes partly overlap, which might complicate the definition of separate ROIs for thin and thick stripes. It should be kept in mind that spatial overlap is expected to some degree and mainly driven by the limiting physiological point spread function of the measured blood oxygenation level-dependent signal in fMRI (Polimeni et al., 2010; Chaimow et al., 2018). This did not interfere with our analysis, since all data points with overlapping activation were excluded in ROI definitions. On the one hand, it is expected that the degree of overlap depends on the chosen $z$ -score threshold level (Nasr et al., 2016) assuming higher thresholds to increase the probability of solely sampling within one stripe type. On the other hand, high $z$ -score thresholds bear the risk to predominantly sample from large veins (Boxerman et al., 1995), which degrades the accuracy of the ROI due to blurring and displacement of the functional signal (Olman et al., 2007). We based the ROI definition on activation maps from differential contrasts between two experimental conditions as illustrated in Figure 1a–b, which are known to be less affected by unspecific macrovascular contributions and draining veins. Furthermore, we would have expected any venous bias to be reflected in R₂* maps (Yacoub et al., 2001; Peters et al., 2007), for example, by uneven sampling of veins between stripe types which is not the case. For these reasons, we conclude that venous bias did not drive our results.

The regular compartmentalization of V2 into distinct stripe types leads to the expectation of specific coverage of cortical area by thin, thick, and pale stripes. For example, it is expected that thick stripes are slightly larger than thin stripes as their name suggests, and that pale stripes cover around 50% of V2 (Shipp and Zeki, 1985; Tootell and Hamilton, 1989). Using fMRI for ROI definitions, the coverage depends on the chosen $z$ -score threshold as stated further above. For $z = 1.96$ ( $p < 0.05$ , two-sided), the relative V2 coverage of non-overlapping portions of thin and thick stripes is 14.1% ± 3.4% and 24.5% ± 6.9% (mean ± standard deviation across participants and hemispheres; see Appendix 4 for absolute coverage of stripe ROIs at different threshold levels). This sums up to a pale stripes coverage of 61.4%.

An alternative hypothesis for different myelination of V2 stripes is that only the borders between pale and dark CO stripes are more strongly myelinated. Pistorio et al., 2006, showed in monkey V1 that the edges of blobs are more myelinated rather than blobs or inter-blob regions itself using a modified Gallyas stain. We note that the used methods in the present study neither allow a functional determination of a definite border between stripe types nor are able to spatially resolve the border region between stripes. Therefore, we cannot rule out this alternative hypothesis. However, to our knowledge, no previous study showed evidence for higher myelination of stripe borders in V2.

Measurements with high resolution are vulnerable to head movements during image acquisition, especially for the long anatomical scans. Therefore, we used an optical tracking system to prospectively correct head movements during anatomical scans (see Materials and methods). With this system, head movements could be robustly detected and corrected for at the length scale of movements induced by respiration and heart beat. Exemplary motion traces are shown in Appendix 5.

The packing density of myelinated fibers in the cerebral cortex varies with cortical depth (Flechsig, 1920; Glasser and Van Essen, 2011) and is also dependent on the cortical folding (Smart and McSherry, 1986). The correct and consistent sampling of data within cortex is therefore critical for our study. We used the equi-volume model to sample at mid-cortical depth. This model has been shown to be less affected by curvature biases than other models (e.g. equi-distant sampling) (Waehnert et al., 2014). The validity of the depth model also depends on accurate cortex segmentation. We visually inspected the cortical segmentation carefully in each participant (see Figure 6). Remaining curvature contributions were regressed out as in other studies (Glasser and Van Essen, 2011; Sereno et al., 2013; Dumoulin et al., 2017).

Figure 6. — (a) Posterior part of the MP2RAGE UNI image in sagittal orientation of one representative participant (subject 3), which was used for segmentation of the cerebral cortex. Overlaid contour lines show the reconstructed white matter/gray matter boundary surface (yellow), the pial boundary surface (red), and a surface at mid-cortical depth (blue). The computed R₁ map from the multi-parameter mapping (MPM) acquisition and the temporal mean from a representative functional magnetic resonance imaging (fMRI) session are shown in (b) and (c), respectively. Overlaid contour lines are identical to (a) to visualize the segmentation and registration quality. Other illustrations of registration quality are provided for each participant in video format, see Figure 6—video 1, Figure 6—video 2, Figure 6—video 3, and Figure 6—video 4. A more quantitative analysis of the achieved registration performance is shown in Appendix 7.

Our findings point toward higher myelination of pale stripes which exhibit lower oxidative metabolic activity according to staining with CO. V2 receives most of its input from V1 and the pulvinar (Tootell et al., 1983; Sincich and Horton, 2005). Pulvinar projections, however, only terminate in layers 3 and 5, whereas the alternating myelin pattern is most obvious in layer 4, which receives input almost exclusively from V1 (Tootell et al., 1983; Sincich and Horton, 2005). An anterograde tracer study in macaques by Sincich and Horton, 2002, showed that [³H]proline injections into V1 preferentially targeted V2 pale stripes. Although we cannot exclude that systematic differences in terminal axonal arborization between stripe types could explain this observation, we speculate that the results of that study correspond to higher axonal density of V1 to V2 projections in layer 4 of pale stripes. This would lead to higher myelination in pale stripes, which is in line with our measurements.

By comparing myelin-sensitive longitudinal relaxation rates (R₁) between stripe types in V2 defined by high-resolution fMRI, we found significant quantitative evidence for myelination differences in living humans at the level of columnar systems. This shows the feasibility to use high-resolution quantitative R₁ values to study cortical myelination, which is known to be less biased by technical artifacts and are thus better comparable among participants and scanner sites. Moreover, it is well known that the myelination of cortical areas and structures affects their functional properties, that is, the propagation of action potentials (Sanders and Whitteridge, 1946), and correlates with postnatal development (Glasser and Van Essen, 2011). Therefore, the estimation of myelin content of specific structures in the human brain in vivo may increase our knowledge about its relationship to functional properties of the brain in particular and the ontogeny of the human brain in general. Our study shows that with the use of qMRI at ultra-high magnetic field strength, this is possible at the spatial scale of thin, thick, and pale stripes. We therefore believe that the current study shows the applicability of qMRI to further advance our knowledge of cortical myelination and tissue microstructure for exploration of structure-function relationships in the living human brain at mesoscopic scale.

Materials and methods

Participants

Four healthy participants (1 female, age = 27.50 ± 4.39, mean ± standard deviation) gave written informed consent to participate in this study. The study was approved by the local ethics committee of the University of Leipzig (reference number: 273–14). All participants had normal or corrected-to-normal visual acuity, normal color vision (based on Ishihara and Farnsworth D15 tests), and normal stereoscopic vision (based on Lang I test). For conducting the optometric tests, separate approval was granted by the ethics committee (reference number: 006–19).

General procedures

Each participant was scanned multiple times on different days in an ultra-high field MR scanner (7 T). The first session was used to acquire a high-resolution anatomical reference scan and retinotopy data (Sereno et al., 1995; Engel et al., 1997) to functionally locate area V2 in each individual. Additionally, a baseline fMRI scan without task was acquired to aid between-session registrations (see below). Color-selective thin stripes (two sessions) and disparity-selective thick stripes (two sessions) were mapped in subsequent scanning sessions. For two participants, we had time to acquire a third thin and thick stripe session, respectively. However, we restricted the data analysis to the use of data from two sessions for consistency. Furthermore, high-resolution anatomical scans (one session) were acquired in a separate scanning session to estimate whole-brain quantitative MR relaxation parameters. A subset of acquired quantitative MR and fMRI retinotopy data was already used in other experiments (Movahedian Attar et al., 2020; McColgan et al., 2021) but was independently processed for this study.

Visual stimulation

For the presentation of visual stimuli, we used an LCD projector (Sanyo PLC-XT20L with custom-built focusing objective, refresh rate: 60 Hz, pixel resolution: 1024 × 768), which was positioned inside the magnet room. To suppress interference with the MR scanner, the projector was placed inside a custom-built Faraday cage. Stimuli were projected onto a rear-projection screen mounted above the participants’ chest inside the bore and viewed through a mirror attached to the head coil. This setup allowed the visual stimulation of around 22° × 13° visual angle. Black felt was put around the screen and all lights were turned off during experiments to mitigate scattered light reaching the participants’ eyes. Experimental stimuli were written in GNU Octave (4.0.0, http://www.gnu.org/software/octave/) using the Psychophysics Toolbox (Brainard, 1997; Pelli, 1997; Kleiner et al., 2007) (3.0.14). A block design consisting of two experimental conditions was used for mapping color-selective thin stripes and disparity-selective thick stripes in V2, which was reported in detail previously (Nasr et al., 2016) and was only changed marginally for this experiment.

Experiment 1: Color-selective thin stripes

Stimuli consisted of isoluminant sinusoidal color-varying (red/blue) or luminance-varying (black/white) gratings as illustrated in Figure 1a. Gratings moved perpendicular to one of four orientations (0°, 45°, 90°, 135°) with direction reversals every 5 s and a drift velocity of 5°/s. Orientations were pseudorandomized between blocks. A low spatial frequency (0.4 cpd) was used to mitigate linear chromatic aberration at color borders and exploit the relatively higher selectivity to color relative to luminance at this spatial scale (Tootell and Nasr, 2017). We point out that the appropriateness to use red and blue colors to stimulate color-selective thin stripes has already been demonstrated for macaques (Tootell et al., 2004; Li et al., 2019) and humans (Nasr et al., 2016). In one run, color and luminance stimuli were both shown four times in separate blocks with a length of 30 s. Each run started and ended with 15 s of uniform gray. Ten runs were conducted in one session. During runs, participants were asked to fix their gaze on a central point (0.1° × 0.1°) and respond on a keypad when the fixation point changed its color (light green, dark green). To measure functional activation related to color, it is important to control for luminance variations across stimuli. Furthermore, isoluminance points between colors are known to change with eccentricity (Livingstone and Hubel, 1987; Bilodeau and Faubert, 1997). We used a flicker photometry (Ives, 1907; Bone and Landrum, 2004) paradigm to get isoluminance ratios between stimuli for each participant. In brief, the luminance of blue was set to 17.3 cd/m² (cf. Li et al., 2019). Before scanning, each participant performed a behavioral task inside the scanner in which they viewed a uniform blue flickering in temporal counter-phase with gray (30 Hz). Participants were asked to adjust the luminance of gray so that the perceived flickering was minimized using a keypad. This procedure was repeated to adjust the luminance for red and conducted at three different eccentricities (0°–1.7°, 1.7°–4.1°, 4.1°–8.3°). As expected, isoluminance ratios changed with eccentricity, which is illustrated in Appendix 6.

Experiment 2: Disparity-selective thick stripes

Stimuli consisted of two overlaid random dot stereograms (RDSs) (Blakemore and Julesz, 1971) made of red and green dots on a black background (dot size: 0.1°, dot density: ∼17%), respectively. Participants viewed stimuli through custom-built anaglyph spectacles using Kodak Wratten filters No. 25 (red) and 44A (cyan). In one condition, red and green RDSs performed a horizontal sinusoidal movement with temporal frequency of 0.25 Hz. Phases of red and green dots were 180° out of phase and initialized to create the perception of a 8 × 6 checkerboard moving periodically in depth (away and toward the participant), which is schematically illustrated in Figure 1b. Maximal disparity was set to ±0.22° (Tsao et al., 2003). In the other condition, static dots were presented, which were perceived as a plane at depth of the fixation point. In one run, both conditions were shown four times in separate blocks with a length of 30 s. Each run started and ended with 15 s of black background. Ten runs were conducted in one session. During runs, participants were asked to fix their gaze on a central point (0.2° × 0.2°) and respond on a keypad when the fixation point changed its form (square, circle). The luminance of red and green dots was kept low to decrease cross-talk between eyes (red dots through red filter: 3.1 cd/m², red dots through cyan filter: 0.07 cd/m², green dots through green filter: 5.7 cd/m², green dots through cyan filter: 0.09 cd/m²). Luminance of green dots was doubled to approximately excite the same amount of cone photoreceptors with both colors (Dobkins et al., 2000).

Retinotopic mapping

A standard phase-encoded paradigm (Sereno et al., 1995; Engel et al., 1997) was used to locate the stimulated portion of V2. Stimuli consisted of a flickering (4 Hz) black-and-white radial checkerboard restricted to a clockwise/anticlockwise rotating wedge (angle: 30°, period: 64 s) or expanding/contracting ring (period: 32 s) presented in separate runs to reveal polar angle and eccentricity maps, respectively; 8.25 cycles were shown in each run. Each run started and ended with 12 s of uniform gray background. Mean luminance was set to 44 cd/m². Participants were asked to fix their gaze on a central point during visual stimulation. No explicit task was given.

Imaging

All experiments were performed on a 7 T whole-body MR scanner (MAGNETOM 7 T, Siemens Healthineers, Erlangen, Germany) equipped with SC72 body gradients (maximum gradient strength: 70 mT/m; maximum slew rate: 200 mT/m/s). For radio frequency (RF) signal transmission and reception, a single-channel transmit/32-channel receive head coil (Nova Medical, Wilmington, DE, USA) was used. At the beginning of each scanning session, a low-resolution transmit field map was acquired to optimize the transmit voltage over the occipital lobe.

Functional data was acquired with a 2D single-shot gradient-echo echo-planar imaging (EPI) sequence (Feinberg et al., 2010; Moeller et al., 2010). A coronal-oblique slab was imaged, which covered all stimulated portions of V2. The following parameters were used for the mapping of color-selective thin stripes, disparity-selective thick stripes, and the baseline fMRI scan without task: nominal voxel size = 0.8 mm isotropic, repetition time (TR) = 3000 ms, echo time (TE) = 24 ms, excitation flip angle (FA) = 77°, field of view (FOV) = 148 × 148 mm², 50 slices, readout bandwidth (rBW) = 1182 Hz/px, echo spacing = 1 ms, partial Fourier = 6/8, and generalized autocalibrating partially parallel acquisition (GRAPPA) (Griswold et al., 2002) = 3. A slightly modified protocol was used for retinotopy measurements with the following parameter changes: voxel size = 1.0 mm isotropic, TR = 2000 ms, TE = 21 ms, FA = 68°, 40 slices, and rBW = 1164 Hz/px.

MR relaxation parameters (R₁, R₂*, PD) were measured with a multi-echo variable flip angle protocol for multi-parameter mapping (MPM) (Vaculčiaková et al., 2022). The protocol was adapted for whole-brain coverage with 0.5 mm isotropic voxel size and consisted of two multi-echo 3D fast low angle shot (FLASH) scans with T₁- and PD-weighting (T1w, PDw) plus maps of B1⁺ and B0. For T1w and PDw, the following parameters were used: TR = 25 ms, TE = 2.8–16.1 ms (6 equidistant echoes with bipolar readout), FA(PDw/T1w) = 5°/24°, FOV = 248 × 217 × 176 mm³ (read × phase × partition), rBW = 420 Hz/px, and GRAPPA = 2 × 2 in both phase-encoding directions. Head movements during the scan were corrected prospectively using an optical tracking system (Kineticor, USA). For motion detection, a mouth guard assembly with attached markers was manufactured for each participant by the Department of Cariology, Endodontology and Periodontology of the University of Leipzig Medical Center (Vaculčiaková et al., 2022). No prospective motion correction was used during functional scans because the camera system and the projection screen did not fit together in the bore. Note that functional scans are also less sensitive to motion due to the short acquisition time per volume.

For the correction of RF transmit field (B1⁺) inhomogeneities in relaxation parameter maps (R₁, PD), we followed the procedure detailed in Lutti et al., 2010; Lutti et al., 2012, acquiring spin-echo and stimulated echo images with a 3D EPI readout. The total scanning time of the MPM protocol was approximately 45 min.

For cortex segmentation and image registration, a whole-brain anatomy was acquired using a 3D T1-weighted MP2RAGE sequence (Marques et al., 2010) with the following parameters: voxel size = 0.7 mm isotropic, TR = 5000 ms, TE = 2.45 ms, inversion times (TI1/TI2) = 900 ms/2750 ms with FA = 5°/3° for T1w and PDw images, respectively, FOV = 224 × 224 × 168 mm³ (read × phase × partition), rBW = 250 Hz/px, partial Fourier = 6/8, and GRAPPA = 2 (primary phase-encoding direction; outer loop). From both inversion times, a uniform T₁-weighted image (UNI) and a T₁-map were created in the online image reconstruction on the scanner.

Data analysis

Functional time series from color-selective and disparity-selective stripe mapping sessions were corrected for within-run and between-run motion using SPM12 (v6906, https://www.fil.ion.ucl.ac.uk/spm/) with Matlab R2019b (MathWorks, Natick, MA, USA). Motion corrected time series were high-pass filtered (cutoff frequency: 1/270 Hz) and voxel-wise statistical analyses were performed for each session using a general linear model as implemented in SPM12 with both experimental conditions as regressors.

For retinotopy measurements, time series were first converted to percent signal change by division by their temporal mean. Furthermore, slice timing correction was added before motion correction by voxel-wise temporal interpolation to a common time grid using 3drefit from Analysis of Function NeuroImages software (Cox, 1996) (AFNI, 19.1.05). Motion corrected time series were high-pass filtered (cutoff frequency: 1/(3 × stimulus cycle period) Hz) and data from the first quarter of the stimulus cycle was discarded from further processing. A voxel-wise Fourier transform was computed and real and imaginary parts at stimulus frequency were averaged from runs with opposite stimulus direction to compensate for the hemodynamic lag. A phase map from averaged polar angle real and imaginary parts was computed to delineate the borders of V2.

For one participant (subject 3), we acquired a second set of retinotopy data in a separate session due to low functional responses in comparison to other participants. Motion corrected time series from the second session were registered nonlinearly to the first session using ANTs (2.3.1, http://stnava.github.io/ANTs/; Avants et al., 2019). Time series from both session were then averaged before further processing.

Quantitative parameter maps (R₁, R₂*, PD) were computed using the hMRI toolbox (Tabelow et al., 2019) (0.2.2, http://hmri.info) implemented in SPM12 (v7487). In brief, T1w and PDw images from the MPM protocol were averaged across echoes and used to compute a registration between both contrasts using SPM12. All available echoes from both contrasts were then used to compute an R₂* map by ordinary least squares regression using the ESTATICS model (Weiskopf et al., 2014). For the calculation of R₁ and PD maps, the extrapolation of T1w and PDw to TE = 0 (to remove any R₂*-weighting bias from resulting maps) was fit to an approximation of the Ernst equation for short-TR dual flip angle measurements using the FLASH signal (Helms et al., 2008; Edwards et al., 2021). The B1⁺ field map was corrected for off-resonance effects using the acquired B0 map. A flip angle map was computed from the resulting B1⁺ map to correct the apparent flip angles for inhomogeneities of the RF transmit field in the fitting procedure. For PD map calculations, the resulting map was corrected for the receiver coil sensitivity profile using the adapted data-driven UNICORT method, which applies the bias field correction implemented in the segmentation module of SPM12 (Weiskopf et al., 2011), and calibrated such that the mean PD over a white matter mask PD(WM) = 69 percent units (pu) (Tofts, 2018). Final maps (R₁, R₂*, PD) were corrected for spatial gradient nonlinearity distortions using the gradunwarp toolbox (Glasser et al., 2013; 1.0.2, https://github.com/Washington-University/gradunwarp; Subramaniam and Brown, 2014) and spherical harmonic coefficients provided by the manufacturer.

Cortex segmentation was based on the MP2RAGE UNI image. First, the UNI image was corrected for gradient nonlinearities with the gradunwarp toolbox and remaining bias fields using SPM12. The resulting image was input to the recon-all pipeline in FreeSurfer (Dale et al., 1999; Fischl et al., 1999) (6.0.0, http://surfer.nmr.mgh.harvard.edu/) with the hires flag to segment at the original voxel resolution (Zaretskaya et al., 2018). The brain mask used during segmentation was computed from the second inversion image of the MP2RAGE using SPM12 and was defined by excluding all voxels that exceeded the tissue class threshold of 10% in non-WM and non-GM tissue classes. Final gray matter/white matter and pial boundary surfaces were corrected manually. Extra care was applied to correct the pial surface around the sagittal sinus. The resulting gray matter/white matter surface was shifted 0.5 mm inward to counteract a potential segmentation bias using FreeSurfer with MP2RAGE (Fujimoto et al., 2014). Final surface meshes were slightly smoothed and upsampled to an average edge length of around 0.3 mm. A surface mesh at mid-cortical depth was computed using the equi-volume model (Waehnert et al., 2014; Wagstyl et al., 2018).

All images were registered to the space of the qMRI maps. For the registration of MP2RAGE and MPM, we used R₁ maps from both acquisitions. Just for the purpose of registration, both images were corrected for potentially remaining bias fields (SPM12) and a brain mask was applied. Images were then transformed into the same space via the scanner coordinate system and a rigid registration was computed using flirt (Jenkinson et al., 2002) (6.0) from the FMRIB Software Library (5.0.11; https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/). A nonlinear transformation was computed to register activation maps and qMRI data in several steps. First, the baseline fMRI scan from the the first session was registered to the MP2RAGE using the Symmetric Normalization (SyN) algorithm (Avants et al., 2008) from Advanced Normalization Tools (ANTs, 2.3.1, http://stnava.github.io/ANTs/; Avants et al., 2019). A nonlinear registration was chosen to account for geometric distortions in functional images resulting from the low bandwidth in phase-encoding direction. Since both images were acquired in the same session, the registration between modalities was robust. Both images were prepared by removing any bias fields (Tustison et al., 2010) and applying a brain mask. Functional data from other sessions were then registered nonlinearly to the baseline EPI using the same procedure. The final transform was computed by concatenating transforms from all steps (EPI → baseline EPI → MP2RAGE → MPM). An exemplary illustration of the registration and segmentation quality can be seen in Figure 6. A more detailed analysis on achieved registration accuracy is given in Appendix 7.

Generated surfaces from cortex segmentation were transformed to MPM space using linear interpolation. For data sampling, images were transformed to MPM space using linear interpolation before sampling onto the surface mesh using nearest neighbor interpolation.

Reliability analysis of fMRI sessions

The consistency of activation maps was analyzed by computing the vertex-wise correlation of activities within V2 between sessions acquired on different days. Spearman’s rank correlation coefficient $r$ was computed. A $p$ -value was determined by permutation testing. A null distribution was created by computing correlation coefficients between data from the first session and spatially shuffled data from the second session $n$ times ( $n = 10, 000$ ). We paid attention to preserve the spatial autocorrelation in spatially shuffled maps using the BrainSMASH package (Burt et al., 2020) (0.10.0) to consider the non-independence of data from neighboring locations. The $p$ -value was then defined as the fraction of the null distribution which is greater or smaller than $r$ . We corrected the estimate of the $p$ -value for the variability resulting from the finite sample size of the null distribution. The variability was described by the variance of the binomial distribution $σ^{2} = n p (1 - p)$ . Here, we used an upper bound of $3 σ$ , which was added to the number of samples exceeding the test statistics (Burt et al., 2020). A $p$ -value of <0.05 was considered as statistically significant.

Quantitative comparison of qMRI parameters between stripe types

We tested the hypothesis that pale stripes are differentially myelinated in comparison to color-selective thin and disparity-selective thick stripes. Activation maps from color and disparity stripe measurements were averaged across sessions, respectively. For participants with more than two acquired sessions, we chose to use the two sessions with highest between-session correlation of activities within V2. Color and disparity stripes were demarcated by thresholding activation maps at a selected threshold level $z \in {0, 0.5, \dots, 4.5}$ . Data points that did not exclusively belong to one stripe type were discarded. Similar to a procedure described in Li et al., 2019, mean qMRI parameter values across participants sampled in color/disparity stripes were tested against the mean throughout V2 excluding values sampled in disparity/color stripes to correct for effects from the other stripe type. This allowed us to indirectly infer effects in pale stripes assuming a tripartite stripe division of V2. For each participant, we subtracted the mean within V2 from qMRI parameter values to account for inter-subject variability (e.g. see variability between participants in Appendix 2). We considered the covariance of qMRI parameter values with local curvature of the cortical sheet (Sereno et al., 2013) by regressing out any linear curvature dependencies. Note that partial volume effects induced by cortical folding are themselves linear, which justifies the use of linear regression. The mean was computed across participants and statistical significance was determined by permutation testing. A null distribution was created by repeating the same procedure $n$ times ( $n = 10, 000$ ) with ROIs generated from spatially shuffled activation maps. The spatial autocorrelation in shuffled maps was preserved using the BrainSMASH package (Burt et al., 2020) and the $p$ -value was computed as stated further above for the fMRI reliability analysis.

Acknowledgements

The research leading to these results has received funding from the European Research Council under the European Union’s Seventh Framework Program (FP7/2007-2013)/ERC grant agreement no. 616905. Nikolaus Weiskopf has received funding from the European Union’s Horizon 2020 research and innovation program under the grant agreement no. 681094 and from the BMBF (01EW1711A and B) in the framework of ERA-NET NEURON. We thank the University of Minnesota Center for Magnetic Resonance Research for the provision of the multiband EPI sequence software. We thank Roland Mueller for the help with building the anaglyph spectacles.

Appendix 1

Appendix 1—figure 3. — The first row shows activation maps for color-selective thin stripe measurements (contrast: color > luminance) (**a, c**) and disparity-thick-stripe measurements (contrast: depth > no depth) (**b, d**). The second row shows eccentricity (**e, g**) and polar angle (**f, h**) phase responses from the retinotopy measurement, which were used to define the positions of V1/V2 and V2/V3 borders. The third row shows qMRI maps of R₁ (i, l) and R₂* (**k, m**) from the multi-parameter mapping (MPM) acquisition, which were used to compare myelin content between stripe types. Other details as in Figure 2. Scale bar: 1 cm.

Appendix 1—figure 4. — The first row shows activation maps for color-selective thin stripe measurements (contrast: color > luminance) (**a, c**) and disparity-thick-stripe measurements (contrast: depth > no depth) (**b, d**). The second row shows eccentricity (**e, g**) and polar angle (**f, h**) phase responses from the retinotopy measurement, which were used to define the positions of V1/V2 and V2/V3 borders. The third row shows qMRI maps of R₁ (i, l) and R₂* (**k, m**) from the multi-parameter mapping (MPM) acquisition, which were used to compare myelin content between stripe types. Other details as in Figure 2. Scale bar: 1 cm.

Appendix 2

Appendix 3

Appendix 4

Appendix 5

Appendix 6

Appendix 7

Analysis of registration accuracy

Accurate registration between qMRI and fMRI data is of utmost importance for the current study. However, knowledge about the quality of registration beyond visual inspection, which is subjective, qualitative, and not usable for large quantities of data, is difficult to achieve (Pluim et al., 2016).

In the following analysis, we quantitatively assessed the quality of registration by exploiting that both qMRI and fMRI data were registered to the same whole-brain anatomy (MP2RAGE). Knowing that in all acquired MR images, gray matter (GM) and white matter (WM) can easily be differentiated, we estimated registration accuracy by the analysis of local image contrast of the registered image around the GM/WM border, which was defined by cortex segmentation based on the MP2RAGE. We expect that the local image contrast is higher at the WM/GM border than in nearby locations in the registered image if registration between whole-brain anatomy and qMRI and fMRI is accurate, respectively. We restricted the analysis to vertices $v$ in V2. For each vertex $v$ , we computed a normal vector and defined 100 points along the normal in equidistant steps around the GM/WM border (±10 mm). Let $λ$ be the parameter defining the position on the normal line, then $v (λ)$ is the coordinate of vertex $v$ shifted along the normal direction by $λ$ . For each $v (λ)$ , we sampled data from two points with distance along the normal vector in both positive $v_{+}$ and negative $v_{-}$ direction using trilinear interpolation. To evaluate the local image contrast, we used the cost function $J$ , which was introduced for the boundary-based registration (BBR) method (Greve and Fischl, 2009). $J$ has the property to approach zero if local image contrast increases. Note that we did not use the BBR method for actual registration. To derive $J$ , a contrast measure $Q (λ)$ is first computed for each $v (λ)$ in V2.

Q (λ) = 100 \frac{v_{+} (λ) - v_{-} (λ)}{0.5 \cdot (v_{+} (λ) + v_{-} (λ))}

(1)

The sign of the contrast depends on the imaging modality, that is, if WM is darker or brighter than GM. This was considered by choosing the direction of the normal vector to point toward the brighter tissue component. $J$ was then computed based on all contrast measures in V2.

J (λ) = \frac{1}{N} \sum_{v (λ) \in V 2} [1 + \tanh (0.5 \cdot Q (λ))]

(2)

where $N$ is the number of vertices in V2. Appendix 7—figure 1 shows for each participant and imaging modality the size of $J$ around the GM/WM border. It can be seen that the position of minimum $J$ generally coincides with the GM/WM border for each data set. This enables us to conclude that accurate registration could be achieved for both qMRI and fMRI data, respectively. However, a small shift was also observed in all data sets relative to the GM/WM border. While partial volume effects could have contributed to this bias, we note that we applied an inward shift (0.5 mm) to the GM/WM boundary surface during segmentation (see Materials and methods). This was done to counteract a potential bias in the segmentation pipeline using the flat image from the MP2RAGE acquisition which yielded visually better results. The current analysis reveals that this step was probably not necessary. However, note that this did not affect any conclusions drawn in further analyses, since the shift is negligible small when compared to changes in local image contrast as shown in Appendix 7—figure 1.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. Open access funding provided by Max Planck Society.

Contributor Information

Daniel Haenelt, Email: haenelt@cbs.mpg.de.

Kristine Krug, Otto-von-Guericke University Magdeburg, Germany.

Floris P de Lange, Radboud University Nijmegen, Netherlands.

Funding Information

This paper was supported by the following grants:

European Research Council ERC grant agreement no 616905 to Nikolaus Weiskopf.
Horizon 2020 - Research and Innovation Framework Programme Grant agreement no 681094 to Nikolaus Weiskopf.
Bundesministerium für Bildung und Forschung 01EW1711A & B to Nikolaus Weiskopf.

Additional information

Competing interests

No competing interests declared.

The Max Planck Institute for Human Cognitive and Brain Sciences has an institutional research agreement with Siemens Healthcare. Nikolaus Weiskopf holds a patent on MRI data acquisition during spoiler gradients (United States Patent 10,401,453). Nikolaus Weiskopf was a speaker at an event organized by Siemens Healthcare and was reimbursed for the travel expenses.

Author contributions

Conceptualization, Data curation, Software, Formal analysis, Investigation, Visualization, Methodology, Writing – original draft, Writing – review and editing.

Supervision, Investigation, Writing – review and editing.

Software, Methodology, Writing – review and editing.

Investigation, Methodology, Writing – review and editing.

Methodology, Writing – review and editing.

Formal analysis, Writing – review and editing.

Conceptualization, Resources, Supervision, Funding acquisition, Project administration, Writing – review and editing.

Ethics

Human subjects: Participants gave written informed consent to participate in this study. The study was approved by the local ethics committee of the University of Leipzig (reference number: 273-14). For conducting the optometric tests, separate approval was granted by the ethics committee (reference number: 006-19).

Additional files

MDAR checklist

elife-78756-mdarchecklist1.docx^{(78.1KB, docx)}

Data availability

Anonymized and defaced MRI data used in the present study are openly accessible at: https://osf.io/624cz/.

The following dataset was generated:

Haenelt D, Trampel R, Nasr S, Polimeni JR, Tootell RBH, Sereno MI, Pine KJ, Edwards LJ, Helbling S, Weiskopf N. 2022. High resolution quantitative and functional MRI indicate lower myelination of thin and thick stripes in human secondary visual cortex. Open Science Framework.

References

Adams DL, Sincich LC, Horton JC. Complete pattern of ocular dominance columns in human primary visual cortex. The Journal of Neuroscience. 2007;27:10391–10403. doi: 10.1523/JNEUROSCI.2923-07.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
Adams DL, Piserchia V, Economides JR, Horton JC. Vascular supply of the cerebral cortex is specialized for cell layers but not columns. Cerebral Cortex. 2015;25:3673–3681. doi: 10.1093/cercor/bhu221. [DOI] [PMC free article] [PubMed] [Google Scholar]
Avants BB, Epstein CL, Grossman M, Gee JC. Symmetric diffeomorphic image registration with cross-correlation: evaluating automated labeling of elderly and neurodegenerative brain. Medical Image Analysis. 2008;12:26–41. doi: 10.1016/j.media.2007.06.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
Avants BB, Tustison NJ, Johnson HJ. Advanced normalization tools. 2.3.1GitHub. 2019 http://stnava.github.io/ANTs/
Bilodeau L, Faubert J. Isoluminance and chromatic motion perception throughout the visual field. Vision Research. 1997;37:2073–2081. doi: 10.1016/s0042-6989(97)00012-6. [DOI] [PubMed] [Google Scholar]
Blakemore C, Julesz B. Stereoscopic depth aftereffect produced without monocular cues. Science. 1971;171:286–288. doi: 10.1126/science.171.3968.286. [DOI] [PubMed] [Google Scholar]
Bone RA, Landrum JT. Heterochromatic flicker photometry. Archives of Biochemistry and Biophysics. 2004;430:137–142. doi: 10.1016/j.abb.2004.04.003. [DOI] [PubMed] [Google Scholar]
Boxerman JL, Hamberg LM, Rosen BR, Weisskoff RM. MR contrast due to intravascular magnetic susceptibility perturbations. Magnetic Resonance in Medicine. 1995;34:555–566. doi: 10.1002/mrm.1910340412. [DOI] [PubMed] [Google Scholar]
Brainard DH. The psychophysics toolbox. Spatial Vision. 1997;10:433–436. doi: 10.1163/156856897X00357. [DOI] [PubMed] [Google Scholar]
Burt JB, Helmer M, Shinn M, Anticevic A, Murray JD. Generative modeling of brain maps with spatial autocorrelation. NeuroImage. 2020;220:117038. doi: 10.1016/j.neuroimage.2020.117038. [DOI] [PubMed] [Google Scholar]
Carey D, Caprini F, Allen M, Lutti A, Weiskopf N, Rees G, Callaghan MF, Dick F. Quantitative MRI provides markers of intra-, inter-regional, and age-related differences in young adult cortical microstructure. NeuroImage. 2018;182:429–440. doi: 10.1016/j.neuroimage.2017.11.066. [DOI] [PMC free article] [PubMed] [Google Scholar]
Chaimow D, Yacoub E, Uğurbil K, Shmuel A. Spatial specificity of the functional MRI blood oxygenation response relative to neuronal activity. NeuroImage. 2018;164:32–47. doi: 10.1016/j.neuroimage.2017.08.077. [DOI] [PubMed] [Google Scholar]
Chen G, Lu HD, Roe AW. A map for horizontal disparity in monkey V2. Neuron. 2008;58:442–450. doi: 10.1016/j.neuron.2008.02.032. [DOI] [PMC free article] [PubMed] [Google Scholar]
Cohen-Adad J, Polimeni JR, Helmer KG, Benner T, McNab JA, Wald LL, Rosen BR, Mainero C. T₂* mapping and B₀ orientation-dependence at 7 T reveal cyto- and myeloarchitecture organization of the human cortex. NeuroImage. 2012;60:1006–1014. doi: 10.1016/j.neuroimage.2012.01.053. [DOI] [PMC free article] [PubMed] [Google Scholar]
Cox RW. AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Computers and Biomedical Research, an International Journal. 1996;29:162–173. doi: 10.1006/cbmr.1996.0014. [DOI] [PubMed] [Google Scholar]
Dale AM, Fischl B, Sereno MI. Cortical surface-based analysis. I. segmentation and surface reconstruction. NeuroImage. 1999;9:179–194. doi: 10.1006/nimg.1998.0395. [DOI] [PubMed] [Google Scholar]
Destrieux C, Fischl B, Dale A, Halgren E. Automatic parcellation of human cortical gyri and sulci using standard anatomical nomenclature. NeuroImage. 2010;53:1–15. doi: 10.1016/j.neuroimage.2010.06.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dick F, Tierney AT, Lutti A, Josephs O, Sereno MI, Weiskopf N. In vivo functional and myeloarchitectonic mapping of human primary auditory areas. The Journal of Neuroscience. 2012;32:16095–16105. doi: 10.1523/JNEUROSCI.1712-12.2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dobkins KR, Thiele A, Albright TD. Comparison of red-green equiluminance points in humans and macaques: evidence for different L: M cone ratios between species. Journal of the Optical Society of America. A, Optics, Image Science, and Vision. 2000;17:545–556. doi: 10.1364/josaa.17.000545. [DOI] [PubMed] [Google Scholar]
Dumoulin SO, Harvey BM, Fracasso A, Zuiderbaan W, Luijten PR, Wandell BA, Petridou N. In vivo evidence of functional and anatomical stripe-based subdivisions in human V2 and V3. Scientific Reports. 2017;7:733. doi: 10.1038/s41598-017-00634-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
Edwards LJ, Kirilina E, Mohammadi S, Weiskopf N. Microstructural imaging of human neocortex in vivo. NeuroImage. 2018;182:184–206. doi: 10.1016/j.neuroimage.2018.02.055. [DOI] [PubMed] [Google Scholar]
Edwards LJ, Pine KJ, Helms G, Weiskopf N. Rational approximation of the Ernst equation for dual angle R1 mapping revisited: beyond the small flip-angle assumption. In Book of Abstracts ESMRMB 2021 Online 38th Annual Scientific Meeting 7–9 October 2021. Magn Reson Mater Phy.2021. [Google Scholar]
Engel SA, Glover GH, Wandell BA. Retinotopic organization in human visual cortex and the spatial precision of functional MRI. Cerebral Cortex. 1997;7:181–192. doi: 10.1093/cercor/7.2.181. [DOI] [PubMed] [Google Scholar]
Feinberg DA, Moeller S, Smith SM, Auerbach E, Ramanna S, Gunther M, Glasser MF, Miller KL, Ugurbil K, Yacoub E. Multiplexed echo planar imaging for sub-second whole brain fMRI and fast diffusion imaging. PLOS ONE. 2010;5:e15710. doi: 10.1371/journal.pone.0015710. [DOI] [PMC free article] [PubMed] [Google Scholar]
Fischl B, Sereno MI, Dale AM. Cortical surface-based analysis. II: inflation, flattening, and a surface-based coordinate system. NeuroImage. 1999;9:195–207. doi: 10.1006/nimg.1998.0396. [DOI] [PubMed] [Google Scholar]
Fischl B, Rajendran N, Busa E, Augustinack J, Hinds O, Yeo BTT, Mohlberg H, Amunts K, Zilles K. Cortical folding patterns and predicting cytoarchitecture. Cerebral Cortex. 2008;18:1973–1980. doi: 10.1093/cercor/bhm225. [DOI] [PMC free article] [PubMed] [Google Scholar]
Flechsig P. Anatomie des menschlichen Gehirns und Rückenmarks auf myelogenetischer Grundlage. Leipzig: Georg Thieme; 1920. [Google Scholar]
Fujimoto K, Polimeni JR, van der Kouwe AJW, Reuter M, Kober T, Benner T, Fischl B, Wald LL. Quantitative comparison of cortical surface reconstructions from MP2RAGE and multi-echo MPRAGE data at 3 and 7 T. NeuroImage. 2014;90:60–73. doi: 10.1016/j.neuroimage.2013.12.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
Glasser MF, Van Essen DC. Mapping human cortical areas in vivo based on myelin content as revealed by T1- and T2-weighted MRI. The Journal of Neuroscience. 2011;31:11597–11616. doi: 10.1523/JNEUROSCI.2180-11.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
Glasser MF, Sotiropoulos SN, Wilson JA, Coalson TS, Fischl B, Andersson JL, Xu J, Jbabdi S, Webster M, Polimeni JR, Van Essen DC, Jenkinson M, WU-Minn HCP Consortium The minimal preprocessing pipelines for the human connectome project. NeuroImage. 2013;80:105–124. doi: 10.1016/j.neuroimage.2013.04.127. [DOI] [PMC free article] [PubMed] [Google Scholar]
Greve DN, Fischl B. Accurate and robust brain image alignment using boundary-based registration. NeuroImage. 2009;48:63–72. doi: 10.1016/j.neuroimage.2009.06.060. [DOI] [PMC free article] [PubMed] [Google Scholar]
Griswold MA, Jakob PM, Heidemann RM, Nittka M, Jellus V, Wang J, Kiefer B, Haase A. Generalized autocalibrating partially parallel acquisitions (GRAPPA) Magnetic Resonance in Medicine. 2002;47:1202–1210. doi: 10.1002/mrm.10171. [DOI] [PubMed] [Google Scholar]
Helms G, Dathe H, Dechent P. Quantitative FLASH MRI at 3T using a rational approximation of the Ernst equation. Magnetic Resonance in Medicine. 2008;59:667–672. doi: 10.1002/mrm.21542. [DOI] [PubMed] [Google Scholar]
Hinds OP, Rajendran N, Polimeni JR, Augustinack JC, Wiggins G, Wald LL, Diana Rosas H, Potthast A, Schwartz EL, Fischl B. Accurate prediction of V1 location from cortical folds in a surface coordinate system. NeuroImage. 2008;39:1585–1599. doi: 10.1016/j.neuroimage.2007.10.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hockfield S, Tootell RBH, Zaremba S. Molecular differences among neurons reveal an organization of human visual cortex. PNAS. 1990;87:3027–3031. doi: 10.1073/pnas.87.8.3027. [DOI] [PMC free article] [PubMed] [Google Scholar]
Horton JC, Hocking DR. Myelin patterns in V1 and V2 of normal and monocularly enucleated monkeys. Cerebral Cortex. 1997;7:166–177. doi: 10.1093/cercor/7.2.166. [DOI] [PubMed] [Google Scholar]
Hubel DH, Livingstone MS. Segregation of form, color, and stereopsis in primate area 18. The Journal of Neuroscience. 1987;7:3378–3415. doi: 10.1523/JNEUROSCI.07-11-03378.1987. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ives FE. A new color meter. Journal of the Franklin Institute. 1907;164:47–56. doi: 10.1016/S0016-0032(07)90164-7. [DOI] [Google Scholar]
Jenkinson M, Bannister P, Brady M, Smith S. Improved optimization for the robust and accurate linear registration and motion correction of brain images. NeuroImage. 2002;17:825–841. doi: 10.1016/s1053-8119(02)91132-8. [DOI] [PubMed] [Google Scholar]
Keller AL, Schüz A, Logothetis NK, Weber B. Vascularization of cytochrome oxidase-rich blobs in the primary visual cortex of squirrel and macaque monkeys. The Journal of Neuroscience. 2011;31:1246–1253. doi: 10.1523/JNEUROSCI.2765-10.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kirilina E, Helbling S, Morawski M, Pine K, Reimann K, Jankuhn S, Dinse J, Deistung A, Reichenbach JR, Trampel R, Geyer S, Müller L, Jakubowski N, Arendt T, Bazin PL, Weiskopf N. Superficial white matter imaging: contrast mechanisms and whole-brain in vivo mapping. Science Advances. 2020;6:eaaz9281. doi: 10.1126/sciadv.aaz9281. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kleiner M, Brainard DH, Pelli D, Ingling A, Murray R, Broussard C. What’s new in psychtoolbox-3. Perception. 2007;36:1–16. [Google Scholar]
Krubitzer LA, Kaas JH. Cortical integration of parallel pathways in the visual system of primates. Brain Research. 1989;478:161–165. doi: 10.1016/0006-8993(89)91490-x. [DOI] [PubMed] [Google Scholar]
Lazari A, Lipp I. Can MRI measure myelin? Systematic review, qualitative assessment, and meta-analysis of studies validating microstructural imaging with myelin histology. NeuroImage. 2021;230:117744. doi: 10.1016/j.neuroimage.2021.117744. [DOI] [PMC free article] [PubMed] [Google Scholar]
Leuze C, Aswendt M, Ferenczi E, Liu CW, Hsueh B, Goubran M, Tian Q, Steinberg G, Zeineh MM, Deisseroth K, McNab JA. The separate effects of lipids and proteins on brain MRI contrast revealed through tissue clearing. NeuroImage. 2017;156:412–422. doi: 10.1016/j.neuroimage.2017.04.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
Li X, Zhu Q, Janssens T, Arsenault JT, Vanduffel W. In vivo identification of thick, thin, and pale stripes of macaque area V2 using submillimeter resolution (f)MRI at 3 T. Cerebral Cortex. 2019;29:544–560. doi: 10.1093/cercor/bhx337. [DOI] [PubMed] [Google Scholar]
Livingstone MS, Hubel DH. Thalamic inputs to cytochrome oxidase-rich regions in monkey visual cortex. PNAS. 1982;79:6098–6101. doi: 10.1073/pnas.79.19.6098. [DOI] [PMC free article] [PubMed] [Google Scholar]
Livingstone MS, Hubel DH. Psychophysical evidence for separate channels for the perception of form, color, movement, and depth. The Journal of Neuroscience. 1987;7:3416–3468. doi: 10.1523/JNEUROSCI.07-11-03416.1987. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lutti A, Hutton C, Finsterbusch J, Helms G, Weiskopf N. Optimization and validation of methods for mapping of the radiofrequency transmit field at 3T. Magnetic Resonance in Medicine. 2010;64:229–238. doi: 10.1002/mrm.22421. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lutti A, Stadler J, Josephs O, Windischberger C, Speck O, Bernarding J, Hutton C, Weiskopf N. Robust and fast whole brain mapping of the RF transmit field B1 at 7T. PLOS ONE. 2012;7:e32379. doi: 10.1371/journal.pone.0032379. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lutti A, Dick F, Sereno MI, Weiskopf N. Using high-resolution quantitative mapping of R1 as an index of cortical myelination. NeuroImage. 2014;93:176–188. doi: 10.1016/j.neuroimage.2013.06.005. [DOI] [PubMed] [Google Scholar]
Marques JP, Kober T, Krueger G, van der Zwaag W, Van de Moortele PF, Gruetter R. MP2RAGE, a self bias-field corrected sequence for improved segmentation and T1-mapping at high field. NeuroImage. 2010;49:1271–1281. doi: 10.1016/j.neuroimage.2009.10.002. [DOI] [PubMed] [Google Scholar]
Marques JP, Khabipova D, Gruetter R. Studying cyto and myeloarchitecture of the human cortex at ultra-high field with quantitative imaging: R1, R2* and magnetic susceptibility. NeuroImage. 2017;147:152–163. doi: 10.1016/j.neuroimage.2016.12.009. [DOI] [PubMed] [Google Scholar]
McColgan P, Helbling S, Vaculčiaková L, Pine K, Wagstyl K, Attar FM, Edwards L, Papoutsi M, Wei Y, Van den Heuvel MP, Tabrizi SJ, Rees G, Weiskopf N. Relating quantitative 7T MRI across cortical depths to cytoarchitectonics, gene expression and connectomics. Human Brain Mapping. 2021;42:4996–5009. doi: 10.1002/hbm.25595. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mezer A, Yeatman JD, Stikov N, Kay KN, Cho NJ, Dougherty RF, Perry ML, Parvizi J, Hua LH, Butts-Pauly K, Wandell BA. Quantifying the local tissue volume and composition in individual brains with magnetic resonance imaging. Nature Medicine. 2013;19:1667–1672. doi: 10.1038/nm.3390. [DOI] [PMC free article] [PubMed] [Google Scholar]
Moeller S, Yacoub E, Olman CA, Auerbach E, Strupp J, Harel N, Uğurbil K. Multiband multislice GE-EPI at 7 Tesla, with 16-fold acceleration using partial parallel imaging with application to high spatial and temporal whole-brain fMRI. Magnetic Resonance in Medicine. 2010;63:1144–1153. doi: 10.1002/mrm.22361. [DOI] [PMC free article] [PubMed] [Google Scholar]
Movahedian Attar F, Kirilina E, Haenelt D, Pine KJ, Trampel R, Edwards LJ, Weiskopf N. Mapping short association fibers in the early cortical visual processing stream using in vivo diffusion tractography. Cerebral Cortex. 2020;30:4496–4514. doi: 10.1093/cercor/bhaa049. [DOI] [PMC free article] [PubMed] [Google Scholar]
Nasr S, Polimeni JR, Tootell RBH. Interdigitated color- and disparity-selective columns within human visual cortical areas V2 and V3. The Journal of Neuroscience. 2016;36:1841–1857. doi: 10.1523/JNEUROSCI.3518-15.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
Nasr S, Tootell RBH. Columnar organization of mid-spectral and end-spectral hue preferences in human visual cortex. NeuroImage. 2018;181:748–759. doi: 10.1016/j.neuroimage.2018.07.053. [DOI] [PMC free article] [PubMed] [Google Scholar]
Navarro KT, Sanchez MJ, Engel SA, Olman CA, Weldon KB. Depth-dependent functional MRI responses to chromatic and achromatic stimuli throughout V1 and V2. NeuroImage. 2021;226:117520. doi: 10.1016/j.neuroimage.2020.117520. [DOI] [PMC free article] [PubMed] [Google Scholar]
Olman CA, Inati S, Heeger DJ. The effect of large veins on spatial localization with GE BOLD at 3 T: displacement, not blurring. NeuroImage. 2007;34:1126–1135. doi: 10.1016/j.neuroimage.2006.08.045. [DOI] [PubMed] [Google Scholar]
Pelli DG. The videotoolbox software for visual psychophysics: transforming numbers into movies. Spatial Vision. 1997;10:437–442. [PubMed] [Google Scholar]
Peterhans E, von der Heydt R. Functional organization of area V2 in the alert macaque. The European Journal of Neuroscience. 1993;5:509–524. doi: 10.1111/j.1460-9568.1993.tb00517.x. [DOI] [PubMed] [Google Scholar]
Peters AM, Brookes MJ, Hoogenraad FG, Gowland PA, Francis ST, Morris PG, Bowtell R. T2* measurements in human brain at 1.5, 3 and 7 T. Magnetic Resonance Imaging. 2007;25:748–753. doi: 10.1016/j.mri.2007.02.014. [DOI] [PubMed] [Google Scholar]
Pistorio AL, Hendry SH, Wang X. A modified technique for high-resolution staining of myelin. Journal of Neuroscience Methods. 2006;153:135–146. doi: 10.1016/j.jneumeth.2005.10.014. [DOI] [PubMed] [Google Scholar]
Pluim JPW, Muenzing SEA, Eppenhof KAJ, Murphy K. The truth is hard to make: Validation of medical image registration. 2016 23rd International Conference on Pattern Recognition (ICPR; Cancun. 2016. [DOI] [Google Scholar]
Polimeni JR, Fischl B, Greve DN, Wald LL. Laminar analysis of 7T BOLD using an imposed spatial activation pattern in human V1. NeuroImage. 2010;52:1334–1346. doi: 10.1016/j.neuroimage.2010.05.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sanders FK, Whitteridge D. Conduction velocity and myelin thickness in regenerating nerve fibres. The Journal of Physiology. 1946;105:152–174. doi: 10.1113/jphysiol.1946.sp004160. [DOI] [PubMed] [Google Scholar]
Savaskan NE, Weinmann O, Heimrich B, Eyupoglu IY. High resolution neurochemical gold staining method for myelin in peripheral and central nervous system at the light- and electron-microscopic level. Cell and Tissue Research. 2009;337:213–221. doi: 10.1007/s00441-009-0815-9. [DOI] [PubMed] [Google Scholar]
Sereno MI, Dale AM, Reppas JB, Kwong KK, Belliveau JW, Brady TJ, Rosen BR, Tootell RBH. Borders of multiple visual areas in humans revealed by functional magnetic resonance imaging. Science. 1995;268:889–893. doi: 10.1126/science.7754376. [DOI] [PubMed] [Google Scholar]
Sereno MI, Lutti A, Weiskopf N, Dick F. Mapping the human cortical surface by combining quantitative T1 with retinotopy. Cerebral Cortex. 2013;23:2261–2268. doi: 10.1093/cercor/bhs213. [DOI] [PMC free article] [PubMed] [Google Scholar]
Shipp S, Zeki S. Segregation of pathways leading from area V2 to areas V4 and V5 of macaque monkey visual cortex. Nature. 1985;315:322–325. doi: 10.1038/315322a0. [DOI] [PubMed] [Google Scholar]
Sigalovsky IS, Fischl B, Melcher JR. Mapping an intrinsic MR property of gray matter in auditory cortex of living humans: a possible marker for primary cortex and hemispheric differences. NeuroImage. 2006;32:1524–1537. doi: 10.1016/j.neuroimage.2006.05.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sincich LC, Horton JC. Pale cytochrome oxidase stripes in V2 receive the richest projection from macaque striate cortex. The Journal of Comparative Neurology. 2002;447:18–33. doi: 10.1002/cne.10174. [DOI] [PubMed] [Google Scholar]
Sincich LC, Horton JC. The circuitry of V1 and V2: integration of color, form, and motion. Annual Review of Neuroscience. 2005;28:303–326. doi: 10.1146/annurev.neuro.28.061604.135731. [DOI] [PubMed] [Google Scholar]
Smart IHM, McSherry GM. Gyrus formation in the cerebral cortex of the ferret. II. description of the internal histological changes. Journal of Anatomy. 1986;147:27–43. [PMC free article] [PubMed] [Google Scholar]
Stüber C, Morawski M, Schäfer A, Labadie C, Wähnert M, Leuze C, Streicher M, Barapatre N, Reimann K, Geyer S, Spemann D, Turner R. Myelin and iron concentration in the human brain: a quantitative study of MRI contrast. NeuroImage. 2014;93 Pt 1:95–106. doi: 10.1016/j.neuroimage.2014.02.026. [DOI] [PubMed] [Google Scholar]
Subramaniam K, Brown TB. Gradunwarp. 1.0.2GitHub. 2014 https://github.com/Washington-University/gradunwarp
Tabelow K, Balteau E, Ashburner J, Callaghan MF, Draganski B, Helms G, Kherif F, Leutritz T, Lutti A, Phillips C, Reimer E, Ruthotto L, Seif M, Weiskopf N, Ziegler G, Mohammadi S. hMRI – a toolbox for quantitative MRI in neuroscience and clinical research. NeuroImage. 2019;194:191–210. doi: 10.1016/j.neuroimage.2019.01.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
Tofts PS. In: Quantitative MRI of the Brain: Principles of Physical Measurement. Tofts PS, editor. Boca Raton: CRC Press; 2018. PD: proton density of tissue water; pp. 55–71. [DOI] [Google Scholar]
Tootell RBH, Silverman MS, De Valois RL, Jacobs GH. Functional organization of the second cortical visual area in primates. Science. 1983;220:737–739. doi: 10.1126/science.6301017. [DOI] [PubMed] [Google Scholar]
Tootell RBH, Hamilton SL. Functional anatomy of the second visual area (V2) in the macaque. The Journal of Neuroscience. 1989;9:2620–2644. doi: 10.1523/JNEUROSCI.09-08-02620.1989. [DOI] [PMC free article] [PubMed] [Google Scholar]
Tootell RBH, Taylor JB. Anatomical evidence for MT and additional cortical visual areas in humans. Cerebral Cortex. 1995;5:39–55. doi: 10.1093/cercor/5.1.39. [DOI] [PubMed] [Google Scholar]
Tootell RBH, Nelissen K, Vanduffel W, Orban GA. Search for color ‘center(s)’ in macaque visual cortex. Cerebral Cortex. 2004;14:353–363. doi: 10.1093/cercor/bhh001. [DOI] [PubMed] [Google Scholar]
Tootell RBH, Nasr S. Columnar segregation of magnocellular and parvocellular streams in human extrastriate cortex. The Journal of Neuroscience. 2017;37:8014–8032. doi: 10.1523/JNEUROSCI.0690-17.2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
Trampel R, Bazin PL, Pine K, Weiskopf N. In-vivo magnetic resonance imaging (MRI) of laminae in the human cortex. NeuroImage. 2019;197:707–715. doi: 10.1016/j.neuroimage.2017.09.037. [DOI] [PubMed] [Google Scholar]
Tsao DY, Vanduffel W, Sasaki Y, Fize D, Knutsen TA, Mandeville JB, Wald LL, Dale AM, Rosen BR, Van Essen DC, Livingstone MS, Orban GA, Tootell RBH. Stereopsis activates V3A and caudal intraparietal areas in macaques and humans. Neuron. 2003;39:555–568. doi: 10.1016/s0896-6273(03)00459-8. [DOI] [PubMed] [Google Scholar]
Ts’o DY, Roe AW, Gilbert CD. A hierarchy of the functional organization for color, form and disparity in primate visual area V2. Vision Research. 2001;41:1333–1349. doi: 10.1016/s0042-6989(01)00076-1. [DOI] [PubMed] [Google Scholar]
Tustison NJ, Avants BB, Cook PA, Zheng Y, Egan A, Yushkevich PA, Gee JC. N4ITK: improved N3 bias correction. IEEE Transactions on Medical Imaging. 2010;29:1310–1320. doi: 10.1109/TMI.2010.2046908. [DOI] [PMC free article] [PubMed] [Google Scholar]
Vaculčiaková L, Podranski K, Edwards LJ, Ocal D, Veale T, Fox NC, Haak R, Ehses P, Callaghan MF, Pine KJ, Weiskopf N. Combining navigator and optical prospective motion correction for high-quality 500 μm resolution quantitative multi-parameter mapping at 7T. Magnetic Resonance in Medicine. 2022;88:787–801. doi: 10.1002/mrm.29253. [DOI] [PubMed] [Google Scholar]
Volz S, Nöth U, Deichmann R. Correction of systematic errors in quantitative proton density mapping. Magnetic Resonance in Medicine. 2012;68:74–85. doi: 10.1002/mrm.23206. [DOI] [PubMed] [Google Scholar]
Waehnert MD, Dinse J, Weiss M, Streicher MN, Waehnert P, Geyer S, Turner R, Bazin PL. Anatomically motivated modeling of cortical laminae. NeuroImage. 2014;93 Pt 2:210–220. doi: 10.1016/j.neuroimage.2013.03.078. [DOI] [PubMed] [Google Scholar]
Wagstyl K, Paquola C, Bethlehem R, Evans AC, Huth A. Equivolumetric layering for mesh surfaces. Zenodo. 2018 doi: 10.5281/zenodo.1442584. [DOI]
Weiskopf N, Lutti A, Helms G, Novak M, Ashburner J, Hutton C. Unified segmentation based correction of R1 brain maps for RF transmit field inhomogeneities (UNICORT) NeuroImage. 2011;54:2116–2124. doi: 10.1016/j.neuroimage.2010.10.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
Weiskopf N, Callaghan MF, Josephs O, Lutti A, Mohammadi S. Estimating the apparent transverse relaxation time (R2(*)) from images with different contrasts (ESTATICS) reduces motion artifacts. Frontiers in Neuroscience. 2014;8:278. doi: 10.3389/fnins.2014.00278. [DOI] [PMC free article] [PubMed] [Google Scholar]
Weiskopf N, Edwards LJ, Helms G, Mohammadi S, Kirilina E. Quantitative magnetic resonance imaging of brain anatomy and in vivo histology. Nature Reviews Physics. 2021;3:570–588. doi: 10.1038/s42254-021-00326-1. [DOI] [Google Scholar]
Xiao Y, Wang Y, Felleman DJ. A spatially organized representation of colour in macaque cortical area V2. Nature. 2003;421:535–539. doi: 10.1038/nature01372. [DOI] [PubMed] [Google Scholar]
Yacoub E, Shmuel A, Pfeuffer J, Van De Moortele PF, Adriany G, Andersen P, Vaughan JT, Merkle H, Ugurbil K, Hu X. Imaging brain function in humans at 7 Tesla. Magnetic Resonance in Medicine. 2001;45:588–594. doi: 10.1002/mrm.1080. [DOI] [PubMed] [Google Scholar]
Zaretskaya N, Fischl B, Reuter M, Renvall V, Polimeni JR. Advantages of cortical surface reconstruction using submillimeter 7 T MEMPRAGE. NeuroImage. 2018;165:11–26. doi: 10.1016/j.neuroimage.2017.09.060. [DOI] [PMC free article] [PubMed] [Google Scholar]
Zheng D, LaMantia AS, Purves D. Specialized vascularization of the primate visual cortex. The Journal of Neuroscience. 1991;11:2622–2629. doi: 10.1523/JNEUROSCI.11-08-02622.1991. [DOI] [PMC free article] [PubMed] [Google Scholar]

eLife. doi: 10.7554/eLife.78756.sa0

Editor's evaluation

Kristine Krug ¹

Using a combination of cutting-edge high-resolution magnetic resonance protocols, this important study investigates the structure-function relationship of specialised compartments in the human visual cortex in vivo. The use of quantitative MRI provides convincing evidence that color- and disparity-selective functional compartments (thin and thick stripes) of visual area V2 have different MR relaxation properties than the pale stripes of V2. While these results indicate different patterns of myelination across the "stripes" of V2, the MR signals will require independent validation to be considered specific to myelin. This study will be of interest to a wide range of neuroscientists and clinicians employing imaging methods to study cortical organization.

eLife. doi: 10.7554/eLife.78756.sa1

Decision letter

Editor: Kristine Krug¹

Reviewed by: Wim Van Duffel²

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "High resolution quantitative and functional MRI indicate lower myelination of thin and thick stripes in human secondary visual cortex" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Kristine Krug as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Floris de Lange as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Wim Van Duffel (Reviewer #3).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) Specificity of the used MR signals to myelin:

– Please temper claims to the specificity of the MR protocols used throughout the manuscript and discuss the evidence critically.

– Where possible, add a quantitative comparison of different MR signals for myelin. We were also wondering whether the authors had Tw1/Tw2 maps available from the same brains.

– Discuss the literature with regard to quantitative MRI, myelin-related MR signals and histology in a more differentiated manner.

2) Visualise ROIs used for the functional stripe compartments and relate them directly to flat maps of R1 and z-scores:

– What do ROIs created from myelin-related MR measures look like? Can they be used to define stripes?

– The reviewers would like to see the three-way ROIs for all hemispheres included.

– Provide additional controls that any irregular distribution of functional ROIs throughout the visual field map has not affected the results.

– Can you exclude the possibility that the high myelination might be located at stripe borders?

3) Include supplements for all subjects showing (i) goodness of registration and (ii) how the borders of V2 were defined based on retinotopy.

Reviewer #1 (Recommendations for the authors):

1) One critical question is whether the direct association of the pale stripe region with higher density of myelination is safe. Ideally, the pale stripes would have been defined functionally as well and the interdigitating pattern of the three stripe compartments would have been similarly distributed throughout the imaged parts of V2. The authors should at least add the following information/analysis, please:

– Show the detailed ROI patterns (thick, thin _and_ inter stripe) used for the analysis of myelin in individual hemispheres for all subjects. This should show the regular distribution of inter stripe regions through the imaged and analysed visual field representations.

– Show that the exclusion of foveal regions from the analysis (mostly without significant activation for binocular disparity (thick stripes, all subjects) or for colour (thin stripes, subject 2)) does not affect the main conclusions. Explanation in lines 233-236 is not sufficient to exclude this.

– Discuss whether you can exclude the alternative interpretation that myelination is particularly prominent as the borders of certain stripes, as it as been indicated for V1 between patches and inter-patches (Pistorio et al. 2006). This might also explain contradictory results between papers.

– A brief discussion should be included on the adequacy of all the chosen visual stimuli, specifically for

(i) activating (or not) foveal representations of V2 (add spatial scale to the discussion of absence of blue cones in the fovea), especially for binocular disparity.

(ii) the artefactual luminance responses that red-blue gratings can generate even at isoluminance (because of chromatic aberrations: red and blue light cannot be in focus on the retina at the same time) and whether this could have activated for instance pale stripes,

and hence (iii) whether one could reasonably expect that the chosen stimuli will not activate pale stripes or whether there might be overlap (with specific reference also to the chosen binocular disparity stimulus).

2) Reconsider the easy, unnecessary straw-man of "all histological results are conflicting" in abstract and introduction for a more considered analysis of the discussed papers and of what is the most likely "ground truth" that you would expect to find with MRI. For instance, Luxol not only stains phospholipids of the neuronal myelin sheath but is also associated with staining mitochondria (Takaya, 1967), which might lead to differential results in cortical staining for CO-rich compartments. Furthermore, the contradictory results for myelin staining "of" CO compartments in the literature might also have to do with more recent carefully analysis showing that myelin might be particularly associated with borders of CO blobs and inter blobs due to their relationship with OD columns (Pistorio et al. 2006). On close inspection, some of the earlier papers (e.g. Tootle et al. 1983) present little clear data on the alignment of the different maps in the papers; even with the more recent, careful studies, one needs to take a close look at the conditions tested (e.g. monocular enucleation, Horton & Hocking 1997). The discussion (e.g. lines 183 -185) seems to be in parts more considered.

Reviewer #2 (Recommendations for the authors):

The Introduction strongly gives the impression that quantitative MRI emerged in the mid-2010s, and entirely from the co-authors laboratory.

On p10, the statement "To detect signal changes with magnitudes of a few percent, qMRI is advantageous since it is less biased by inhomogeneities in the radiofrequency transmit and receive fields, and therefore allows better comparison of differences between ROIs and across participants" doesn't make sense to me. I gather the authors intend this to be taken as an argument for qMRI rather than non-quantitative "weighted" imaging. There are a few oddities of logic here. First, the fact that these differences are a few percent in units of R1 does not necessarily mean they are a few percent in terms of weighted contrast; second, even if the signals are of similar magnitude of percentage, the process of quantification "introduces" noise (e.g., involves propagation of error that amplifies noise from the weighted images) making it harder to detect those small percent signals; third, this statement conflates effect size (% signal) with sources of bias. I can believe that avoiding sources of bias is a major argument for qMRI, but this statement needs editing.

On p10, the authors note that the small effect sizes "hindered a direct visualization of stripes at the voxel level in R1 maps" and refer to Figure 4. But this cannot be seen in Figure 4. I recommend adding a direct comparison of flat maps of R1 and z-score so readers can see what the data looks like. Ideally this would superimpose some of the structure from fMRI (potentially the blue dots from Figure 2, although these may obscure any structure that is present).

eLife. 2023 Mar 8;12:e78756. doi: 10.7554/eLife.78756.sa2

Author response

Essential revisions:

(1) Specificity of the used MR signals to myelin:

– Please temper claims to the specificity of the MR protocols used throughout the manuscript and discuss the evidence critically.

Thank you for this comment. We agree that we have inadvertently overstated the specificity of R1 at several occasions in the text and modified the statements in the revised manuscript accordingly (p. 2, lines 68–88; p. 6, line 163; p. 8, line 216; p. 9, lines. 257–260). Please note that we also discuss this point in the author response to reviewer #2 (public review) in more detail.

– Where possible, add a quantitative comparison of different MR signals for myelin. We were also wondering whether the authors had Tw1/Tw2 maps available from the same brains.

In the manuscript, we performed the analysis with two independent approaches (MPM and MP2RAGE) which yielded basically the same results (see Figure 5 and Figure 5-Supplement 2). Both MPM and MP2RAGE are well suited for measurements at 7 T. Unfortunately, we do not have T1w/T2w maps available for any of our participants. We also would like to point out that the acquisition of high-resolution T2-weighted images or T2 maps are very challenging at high magnetic fields and to the best of our knowledge, at 7 T, neither T2w nor T2 maps have been successfully acquired with an isotropic resolution of (0.5 mm)³ as used in the present study. Therefore, we were unfortunately not able to acquire T1w/T2w data at 7 T with a comparable resolution. Please also see the author response to reviewer #3 (recommendations for the authors; comment 1), which provides an extended discussion, including a discussion of different MR contrasts for myelin mapping.

– Discuss the literature with regard to quantitative MRI, myelin-related MR signals and histology in a more differentiated manner.

In the revised manuscript, we removed the phrasing that gave the impression that MRI can resolve the conflicting results of histological studies. In the introduction, we changed the corresponding paragraph by emphasizing the alternative view, which can be obtained from MRI by the possibility to investigate structure-function relationships in the living human brain (see p. 2, lines 68–70; p. 3, lines 93–95). We also included references in the introduction which show the increasing body of literature using R1 as surrogate for cortical myelination while still pointing out the limitations concerning the specificity of R1 (p. 2, lines 78–85). We also included an extended discussion about current knowledge of V2 stripes myelination in the Discussion section (pp. 8–9, lines 216–240).

(2) Visualise ROIs used for the functional stripe compartments and relate them directly to flat maps of R1 and z-scores:

We added Appendix 1, which shows figures of thresholded functional activation maps (color-selective thin stripes and disparity-selective thick stripes), retinotopy data (polar angle and eccentricity maps) and qMRI parameter maps (R1 and R2*) for each participant. Within a single hemisphere of each participant, all figures are aligned to each other, which allows direct visual comparison between maps.

– What do ROIs created from myelin-related MR measures look like? Can they be used to define stripes?

Appendix 1 of the revised manuscript shows quantitative parameter maps (R1 and R2*) for each participant in V2. Those maps do not allow direct identification of thin, thick and pale stripes. This is in line with the rather small (but reliable) R1 differences found in the current study, which demanded a statistical analysis across participants.

– The reviewers would like to see the three-way ROIs for all hemispheres included.

Since ROIs were based on functional activation maps, actual ROI sizes depend on the chosen z-score threshold level. This was the reason for performing the main analysis at different threshold levels. For visualization purposes we now illustrate the ROIs for a specific intermediate threshold level (z=1.96) in Appendix 3.

– Provide additional controls that any irregular distribution of functional ROIs throughout the visual field map has not affected the results.

As pointed out by the reviewers and editors, a consistent pattern in functional activation maps across participants was the general absence of activation around the foveal representation. Possible explanations were added to the Discussion section (pp. 9–10, lines 289–303). See also the extended discussion in the authors responses to reviewer #1 (recommendations for the authors). Since we used activation maps for ROI definitions, this could potentially have introduced a visual field dependent bias in our analysis. To exclude this possibility, we included a supporting analysis, in which we excluded the region around the foveal representation from the analysis – see further details in the revised manuscript (p. 8, lines 189–202). Excluded regions were delineated in Figure 2, Figure 2-Supplement 1 and Appendix 1 using white lines. In Figure 5-Supplement 1 and Figure 5-Supplement 3, results from this supporting analysis are shown which reproduced the primary findings from the main analysis, particularly the relatively higher myelination of pale stripes, and thus corroborate the main findings of the current study.

– Can you exclude the possibility that the high myelination might be located at stripe borders?

Thank you for this comment. This is an intriguing alternative hypothesis. Due to the limited spatial resolution of our methods, we cannot exclude this possibility. We added a paragraph in the Discussion section (p. 10, lines 330–337) to inform the reader about this alternative hypothesis. However, since the recommended article by the reviewers (Pistorio et al., 2006) only considers the CO-rich blob structure in V1 and no evidence for this hypothesis is known to the authors for area V2, we consider this hypothesis as rather speculative.

(3) Include supplements for all subjects showing (i) goodness of registration and (ii) how the borders of V2 were defined based on retinotopy.

(i) To further demonstrate the high quality of registration in the current study, we included video files showing the registration results for each participant (see Figure 6-video 1, Figure 6-video 2, Figure 6-video 3, Figure 6-video 4). We additionally included an analysis in Appendix 7, which examined the achieved goodness-of-registration.

(ii) In Figure 2—figure supplement 1, we show retinotopy data to illustrate the definition of retinotopic borders in Figure 2. In Appendix 1, retinotopy data is shown (next to the illustration of qMRI parameter maps and functional activation maps) for each participant.

Reviewer #1 (Recommendations for the authors):

(1) One critical question is whether the direct association of the pale stripe region with higher density of myelination is safe. Ideally, the pale stripes would have been defined functionally as well and the interdigitating pattern of the three stripe compartments would have been similarly distributed throughout the imaged parts of V2. The authors should at least add the following information/analysis, please:

– Show the detailed ROI patterns (thick, thin _and_ inter stripe) used for the analysis of myelin in individual hemispheres for all subjects. This should show the regular distribution of inter stripe regions through the imaged and analysed visual field representations.

We agree that illustration of the actual ROIs will be beneficial for the reader. We included Appendix 3, which presents ROIs at an intermediate threshold level of z=1.96 where clear differences between thick/thin and pale stripes were found (Figure 5). Please note that ROIs were defined at different threshold levels in the analysis to avoid bias due to an arbitrary choice of threshold and demonstrate robustness of the results. In the present study, activation for disparity contrasts was overall higher than for color contrasts (see e.g. that disparity contrast was shown with different color bars in Figure 2 and Appendix 1). This is also reflected in the functionally delineated ROIs, which were based on using the same threshold level for both color and disparity contrast resulting in larger areas covered by the disparity-selective ROIs in V2. Despite this, color-selective ROIs were visible throughout the entire stimulated region of V2.

– Show that the exclusion of foveal regions from the analysis (mostly without significant activation for binocular disparity (thick stripes, all subjects) or for colour (thin stripes, subject 2)) does not affect the main conclusions. Explanation in lines 233-236 is not sufficient to exclude this.

Thank you for raising this very important point. To identify and exclude this potential bias in our results, we performed a supporting analysis where we excluded the region encompassing the foveal representation in V2 (see p. 8, lines 189–202; p. 10, lines 301–303). To have a consistent exclusion criterion across participants, we used the eccentricity maps from retinotopy measurements for the definition of exclusion masks. The excluded area is illustrated by a white border in Figure 2, Figure 2-Supplement 1 and Appendix 1. Repeating the comparison of R1 and R2* between stripe types after exclusion of the foveal region corroborated the main results, i.e., significant R1 differences between thick/thin and pale stripes. This can be seen in Figure 5-Supplement 1 and Figure 5-Supplement 3.

– Discuss whether you can exclude the alternative interpretation that myelination is particularly prominent as the borders of certain stripes, as it as been indicated for V1 between patches and inter-patches (Pistorio et al. 2006). This might also explain contradictory results between papers.

Thank you for the insightful recommendation pointing us to this article. This is an intriguing alternative hypothesis which we were not aware of. Due to the limited spatial resolution of our methods, we cannot exclude this possibility. We therefore included it in the Discussion section (p. 10, lines 330–337). It is tempting to argue that, since we defined stripe ROIs at different z-score threshold levels (and thus including more or less border volume), we might have seen a particular discontinuity in R1 differences if only stripe borders were more myelinated, which we did not. However, this is rather speculative and presumably out of reach considering the resolution and data quality of our study.

– A brief discussion should be included on the adequacy of all the chosen visual stimuli, specifically for

(i) activating (or not) foveal representations of V2 (add spatial scale to the discussion of absence of blue cones in the fovea), especially for binocular disparity.

(ii) the artefactual luminance responses that red-blue gratings can generate even at isoluminance (because of chromatic aberrations: red and blue light cannot be in focus on the retina at the same time) and whether this could have activated for instance pale stripes,

and hence (iii) whether one could reasonably expect that the chosen stimuli will not activate pale stripes or whether there might be overlap (with specific reference also to the chosen binocular disparity stimulus).

Regarding point (i): In the original Discussion section, we already shortly discussed the absence of activation in the region around the foveal representation (see p. 10, lines 290–303) and its potential causes including absence of blue cones and eccentricity dependence of disparity tuning. Concerning the color contrast, it is known that blue cones are absent in the central 0.5° of the retina and the macula lutea covers the central 5° of the retina (Tootell and Nasr, 2015).

It would be insightful to mark this range of the visual field in activation maps. One possibility to demarcate this region would be to use the acquired retinotopy data and use a population receptive field (pRF) model to estimate for each voxel the visual space it is most sensitive. However, the acquired retinotopy data was primarily optimized for localization of V1/V2 and V2/V3 borders and more data would be beneficial for robust pRF modeling. Furthermore, a retinotopic response of the central fovea is difficult to achieve. Therefore, we did not attempt to use our retinotopy data to get a precise mapping (spatial scale) of the central region. However, we have used it to approximate a region including the fovea and performed the data analyses excluding this region (see above), in order to identify potential biases.

Concerning the disparity contrast, it is known that many cells are clustered in disparity zones and only respond to stimuli over a narrow range (Ts’o et al., 2001). However, it remains speculative if the covered eccentricity range in the stimulus can explain absence of activation in the foveal representation.

In this regard, it might be insightful to compare our activation maps with a previous human fMRI study, which studied cortical regions selective for stereoscopic processing (Tsao et al., 2003). The study was conducted at 3T at conventional resolution (isotropic 3.1 mm voxel size, note that this is a 58-fold larger nominal voxel volume compared to our study). Critically, this study used a very similar disparity stimulus and examined the same depth range (±0.22°). This study revealed a very similar topography (due to the larger voxel size, no disparity-selective thick stripes could be resolved) with absence of activation around the foveal representation. We now added other possible explanation for the missing of disparity-selective activation around the foveal representation in the revised manuscript (p. 10, lines 295–301).

We cannot pinpoint any of these explanations. However, since we could confirm our findings after exclusion of the V2 regions around the central fovea, we conclude that missing activation around the central fovea did not bias the results in the present study in a relevant way.

Regarding (ii): For the color experiment, we used isoluminant red/blue gratings. As pointed out by the reviewer, chromatic aberration means that different colors from the same object in the visual field stimulate the retina at different locations. Thus, blue and red stimuli may not have been well matched, compromising the fMRI subtraction design. To address this issue, our stimulus included several properties to minimize chromatic aberration: (1) We used a low spatial frequency for gratings to tap the relative higher sensitivity to color (relative to luminance) at lower spatial frequencies and minimize chromatic aberration at color borders (Tootell and Nasr, 2017). (2) Additionally, the grating had a sinusoidal shape which further decreased chromatic aberration at color borders. However, it was shown in a previous study that waveform of red/blue gratings had only a marginal influence on color-selective thin stripe activation (Tootell and Nasr, 2017).

Moreover, it is known from optical imaging that color-selective thin stripes in V2 possess a systematic topography of hue representation (Xiao et al., 2003). This means that, in principle, other hues than red/blue might be used to activate color-selective thin stripes. A recent high-resolution fMRI study in humans investigated the processing of different hues (red, yellow, green, blue) in color-selective thin stripes of V2 (Nasr and Tootell, 2018). Consistent with Xiao et al. (2003), they found that thin stripes were responsive to all tested hues. However, the response of mid-spectral hues (green, yellow) compared to end-spectral hues (red, blue) showed a more uniform response across V2. That means that in contrast to mid-spectral hues, red/blue colors are suited to selectively activate V2 thin stripes.

An impressive example is the study by Tootell et al. (2004), which used the double-label deoxyglucose technique to map thin stripes in macaques while stimulating with very similar red/blue gratings compared to the present study. Derived activation maps were compared to histological sections after staining with CO. It can be seen that color-selective activation clusters are exclusively located within CO dark thin stripes (see Figure 3 in Tootell et al., 2004). We added a statement in the methods section to underline the applicability of the used colors for the stimulation of color-selective thin stripes (p. 12, lines 419–421).

Regarding (iii): Indeed, it is expected that color-selective and disparity-selective activation clusters will overlap to some extent. As pointed out in the previous paragraph, color-selective activation is confined to CO dark stripes. However, disparity-selective clusters are found to some extent in all stripes (but mostly in dark CO thick stripes). We stated this on p. 4 (lines 125–128). Furthermore, the effective spatial resolution of the BOLD response is limited not only due to technical but also physiological mechanisms, leading to potential overlap between stripes. This is discussed on p. 10 (lines 304–320). This was one of the reasons to use different z-score threshold levels in the analysis. Furthermore, all vertices with overlapping activation from color-selective thin and disparity-selective thick stripe measurements were excluded in thin and thick stripe ROIs, to dichotomize the data set.

(2) Reconsider the easy, unnecessary straw-man of "all histological results are conflicting" in abstract and introduction for a more considered analysis of the discussed papers and of what is the most likely "ground truth" that you would expect to find with MRI. For instance, Luxol not only stains phospholipids of the neuronal myelin sheath but is also associated with staining mitochondria (Takaya, 1967), which might lead to differential results in cortical staining for CO-rich compartments. Furthermore, the contradictory results for myelin staining "of" CO compartments in the literature might also have to do with more recent carefully analysis showing that myelin might be particularly associated with borders of CO blobs and inter blobs due to their relationship with OD columns (Pistorio et al. 2006). On close inspection, some of the earlier papers (e.g. Tootle et al. 1983) present little clear data on the alignment of the different maps in the papers; even with the more recent, careful studies, one needs to take a close look at the conditions tested (e.g. monocular enucleation, Horton & Hocking 1997). The discussion (e.g. lines 183 -185) seems to be in parts more considered.

Thank you very much for these comments and the insightful article recommendations. We toned down the claims about the specificity of R1 values towards myelin throughout the manuscript. We also removed the unfortunate formulation that gave the impression that MRI can resolve the conflicting results found in histological studies. In the introduction, we changed the corresponding paragraph by emphasizing the alternative and complementary view which MRI offers by the possibility to investigate structure-function relationships in the living human brain (see p. 2, lines 68–70; p. 3, lines 93–95).

Because of the contradictory findings of histological studies, we could not further finesse the hypothesis beyond that we expect differences in the myelin sensitive MRI metrics between the thin/thick versus pale stripes. If the reviewer has a particular refined hypothesis or statement in mind, which we may have overlooked, we would be grateful for their feedback. To improve the contextual understanding, we added a paragraph in the Discussion section covering in more depth how the MRI results relate to known histological findings (see pp. 8–9, lines 216–240).

Reviewer #2 (Recommendations for the authors):

The Introduction strongly gives the impression that quantitative MRI emerged in the mid-2010s, and entirely from the co-authors laboratory.

We did not intend to give the impression that we founded the research field of qMRI or that the concept is as recent as the 2010s. We changed the introduction accordingly, e.g., by stating that qMRI is a collective term of techniques (see p. 2, lines 70–76) which includes the concept of physical modeling of MRI data. At the same time, we tried to stay as general as possible to accommodate the diverse readership of eLife.

On p10, the statement "To detect signal changes with magnitudes of a few percent, qMRI is advantageous since it is less biased by inhomogeneities in the radiofrequency transmit and receive fields, and therefore allows better comparison of differences between ROIs and across participants" doesn't make sense to me. I gather the authors intend this to be taken as an argument for qMRI rather than non-quantitative "weighted" imaging. There are a few oddities of logic here. First, the fact that these differences are a few percent in units of R1 does not necessarily mean they are a few percent in terms of weighted contrast; second, even if the signals are of similar magnitude of percentage, the process of quantification "introduces" noise (e.g., involves propagation of error that amplifies noise from the weighted images) making it harder to detect those small percent signals; third, this statement conflates effect size (% signal) with sources of bias. I can believe that avoiding sources of bias is a major argument for qMRI, but this statement needs editing.

Thank you for this comment. For brevity, we had shortened the discussion clearly too much at this point. We now elaborate further on the differences between qMRI and weighted MRI. Since a comprehensive discussion would distract from the main points of the paper we also refer to our recent review paper on qMRI (Weiskopf et al., 2021). In the following, we comment on the points raised by the reviewer:

(1) In our study, the contrast (i.e. relative percent difference between thick/thin and pale stripes) is comparable for the R1, R2* metrics and the respective weighted gradient echo images. Since the observed relative differences in R1 and R2* are in the range of a few percent, the Ernst equation and free induction decay, will lead to relative weighted signal differences (i.e. contrast) that are also in the range of a few percent and no relevant contrast increase over qMRI parameters can be expected. For example, assuming a R1 of 0.58 s^-1 for cortical gray matter and a ΔR1 of about 0.014 s^-1, which corresponds to our findings (see p.7, lines 185–188), the maximum achievable relative signal change is on the order of about 2% (see Author response image 1), irrespective of the chosen TR and flip angle. This percent change is comparable to the observed change of 2.4% in R1.

Author response image 1. — Based on the Ernst equation, the set of curves shows the relative signal change in percent unit between the signal at R1 and the signal at R1+ΔR1 for different repetition times (TR) and flip angles. T1-weighting in gradient echo sequences increases for small TR and large flip angles. We set the R1 = 0.58 s^-1 (which is the mean cortical gray matter R1 of our cohort in V2) and ΔR1 = 0.014 s^-1 (see p. 7, lines 185–188), which corresponds to a relative R1 change of about 2.4%. In the graphic, it can be seen that this change in R1 translates to a similar maximal relative signal change of ca. 2% in the weighted gradient echo images.

(2) The estimation if and how the signal-to-noise ratio (SNR) or contrast-to-noise ratio (CNR) is decreased in qMRI versus weighted imaging is a highly technical and complex issue and beyond the scope of our paper. However, we briefly summarize the main aspects in the following. Assuming constant contrast (see previous point), the different noise sources need to be distinguished, particularly thermal, physiological and other sources of instrumental noise.

We agree that the quantification of R1, which requires a non-linear operation combining two differently weighted source images, leads to a reduction of signal-to-thermal noise ratio when compared against a weighted image acquired with optimal parameters. However, at 7 T, the optimal acquisition parameters are highly location dependent, since the RF transmit field varies across the brain. Thus, it is a rather academic argument that does not translate into practice.

When other sources of noise such as differences in signal due to RF transmit/receive field differences are considered, qMRI clearly outperforms weighted imaging (see e.g. Figure 1b in Weiskopf et al., 2021). At 7 T this also applies to small regions such as V2.

Acquisition protocols are frequently different between qMRI and weighted imaging. For example, the standard weighted gradient echo sequences usually acquire one image at a single echo time. The MPM protocol acquired 6 echoes, i.e. 6x the amount of data. This significantly increases the SNR compared to the conventional approach while maintaining accuracy due to appropriate modeling such as ESTATICS (Weiskopf et al., 2014). Similar considerations apply in the case of physiological noise impact, which was significantly reduced in our study due to the use of prospective motion correction (Vačulciaková et al., 2022). We modified the statements in the Discussion section accordingly, but we hope that the reviewer agrees that a comprehensive discussion would be beyond the scope of the paper (p. 9, lines 273–285).

(3) We are not sure what the “conflates effect size (% signal) with sources of bias“ statement refers to, but we presume it may refer to the use of CNR, which takes the ratio of the effect of interest (i.e. signal/parameter difference between thick/thin and pale stripes) divided by the relevant noise metric. This metric is useful for estimating the statistical power and detection probability in a measurement and in our opinion actually the most relevant metric. Note that we also stated the coefficient of variation on p. 9 (lines 279–280). Regarding the noise metric we point out that for the purpose of a ROI analysis across a large brain area such as V2, the bias is a relevant noise source in the statistical analysis, since it cannot be reliably removed. To clarify our point, we removed this sentence and revised the paragraph (p. 9, lines 273–285). We hope that the revised phrasing could clarify this point.

On p10, the authors note that the small effect sizes "hindered a direct visualization of stripes at the voxel level in R1 maps" and refer to Figure 4. But this cannot be seen in Figure 4. I recommend adding a direct comparison of flat maps of R1 and z-score so readers can see what the data looks like. Ideally this would superimpose some of the structure from fMRI (potentially the blue dots from Figure 2, although these may obscure any structure that is present).

Thank you for pointing this out. In addition to the R1 maps on inflated surfaces in the original version of the manuscript, we have included Figure 4-Supplement 1, which shows the whole-brain topography of cortical R2* for one representative participant. Furthermore, we have included now Appendix 1, which presents functional activation maps for color- and disparity-selective stripe measurements, retinotopy data (eccentricity, polar angle), and qMRI parameters maps (R1, R2*) on flattened surfaces covering V1 and V2 for each participant and both hemispheres.

References

Haast RAM, Ivanov D, Formisano E, Uludag K. Reproducibility and Reliability of Quantitative and Weighted T1 and T2* Mapping for Myelin-Based Cortical Parcellation at 7 Tesla. Front Neuroanat 2016; 10:112.

Marques JP, Kober T, Krueger G, Van der Zwaag W, Van de Moortele P-F, Gruetter R. MP2RAGE, a self bias-field corrected sequence for improved segmentation and T1-mapping at high field. Neuroimage 2010; 49:1271–1281.

Nasr S, Polimeni JR, Tootell RBH. Interdigitated Color- and Disparity-Selective Columns within Human Visual Cortical Areas V2 and V3. J Neurosci 2016; 36:1841–1857.

Nasr S, Tootell RBH. Columnar organization of mid-spectral and end-spectral hue preferences in human visual cortex. Neuroimage 2018; 181:748–759.Petracca M, El Mendili MM, Moro M, Cocozza S, Podranski K, Fleysher L, Inglese M. Laminar analysis of the cortical T1/T2-weighted ratio at 7T. Neurol-Neuroimmunol 2020; 7:e900.

Schmidt J, Radunsky D, Scheibe P, Ben-Eliezer N, Weiskopf N, Trampel R. Quantification of transverse relaxation times in vivo at 7T field-strength. Proceedings International Society for Magnetic Resonance in Medicine 2021; Abstract #3049.

Tootell RBH, Nelissen K, Vanduffel W, Orban GA. Search for Color 'Center(s)' in Macaque Visual Cortex. Cereb 2004; 14:353–363.Tootell RBH, Nasr S. Primate Color Vision. In: Brain Mapping. An Encyclopedic Reference. Elsevier 2015; 2:489–506.

Tootell RBH, Nasr S. Columnar Segregation of Magnocellular and Parvocellular Streams in Human Extrastriate Cortex. J Neurosci 2017; 37:8014–8032.

Tsao DY, Vanduffel W, Sasaki Y, Fize D, Knutsen TA, Mandeville JB, Wald LL, Dale AM, Rosen BR, Van Essen DC, Livingstone MS, Orban GA, Tootell RBH. Stereopsis Activates V3A and Caudal Intraparietal Areas in Macaques and Humans. Neuron 2003; 39:555–568.

Ts’o DY, Roe AW, Gilbert CD. A hierarchy of the functional organization for color, form and disparity in primate visual area V2. Vis Res 2001; 41:1333–1349.

Vačulciaková L, Podranski K, Edwards LJ, Ocal D, Veale T, Fox NC, Haak R, Ehses P, Callaghan MF, Pine KJ, Weiskopf N. Combining navigator and optical prospective motion correction for high-quality 500 μm resolution quantitative multi-parameter mapping at 7T. MRM 2022; 88:787–801.

Weiskopf N, Callaghan MF, Josephs O, Lutti A, Mohammadi S. Estimating the apparent transverse relaxation time (R2*) from images with different contrasts (ESTATICS) reduces motion artifacts. Front Neurosci 2014; 8:278.

Weiskopf N, Edwards LJ, Helms G, Mohammadi S, Kirilina E. Quantitative magnetic resonance imaging of brain anatomy and in vivo histology. Nat Rev Phys 2021; 3:570–588.

Xiao Y, Wang Y, Felleman DJ. A spatially organized representation of colour in macaque cortical area V2. Nature 2003; 421:535–539.

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Citations

Haenelt D, Trampel R, Nasr S, Polimeni JR, Tootell RBH, Sereno MI, Pine KJ, Edwards LJ, Helbling S, Weiskopf N. 2022. High resolution quantitative and functional MRI indicate lower myelination of thin and thick stripes in human secondary visual cortex. Open Science Framework. [DOI] [PMC free article] [PubMed]

Supplementary Materials

Figure 4—source data 1. Source data for mean quantitative magnetic resonance imaging (qMRI) values (R₁, R₂*, PD) across cortical areas (angular gyrus, MT, V2, V1).

elife-78756-fig4-data1.xlsx^{(6KB, xlsx)}

Figure 5—source data 1. Source data for comparisons of R₁ and R₂* between V2 stripe types.

elife-78756-fig5-data1.xlsx^{(17.4KB, xlsx)}

MDAR checklist

elife-78756-mdarchecklist1.docx^{(78.1KB, docx)}

Data Availability Statement

Anonymized and defaced MRI data used in the present study are openly accessible at: https://osf.io/624cz/.

The following dataset was generated:

[bib1] Adams DL, Sincich LC, Horton JC. Complete pattern of ocular dominance columns in human primary visual cortex. The Journal of Neuroscience. 2007;27:10391–10403. doi: 10.1523/JNEUROSCI.2923-07.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib2] Adams DL, Piserchia V, Economides JR, Horton JC. Vascular supply of the cerebral cortex is specialized for cell layers but not columns. Cerebral Cortex. 2015;25:3673–3681. doi: 10.1093/cercor/bhu221. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib3] Avants BB, Epstein CL, Grossman M, Gee JC. Symmetric diffeomorphic image registration with cross-correlation: evaluating automated labeling of elderly and neurodegenerative brain. Medical Image Analysis. 2008;12:26–41. doi: 10.1016/j.media.2007.06.004. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib4] Avants BB, Tustison NJ, Johnson HJ. Advanced normalization tools. 2.3.1GitHub. 2019 http://stnava.github.io/ANTs/

[bib5] Bilodeau L, Faubert J. Isoluminance and chromatic motion perception throughout the visual field. Vision Research. 1997;37:2073–2081. doi: 10.1016/s0042-6989(97)00012-6. [DOI] [PubMed] [Google Scholar]

[bib6] Blakemore C, Julesz B. Stereoscopic depth aftereffect produced without monocular cues. Science. 1971;171:286–288. doi: 10.1126/science.171.3968.286. [DOI] [PubMed] [Google Scholar]

[bib7] Bone RA, Landrum JT. Heterochromatic flicker photometry. Archives of Biochemistry and Biophysics. 2004;430:137–142. doi: 10.1016/j.abb.2004.04.003. [DOI] [PubMed] [Google Scholar]

[bib8] Boxerman JL, Hamberg LM, Rosen BR, Weisskoff RM. MR contrast due to intravascular magnetic susceptibility perturbations. Magnetic Resonance in Medicine. 1995;34:555–566. doi: 10.1002/mrm.1910340412. [DOI] [PubMed] [Google Scholar]

[bib9] Brainard DH. The psychophysics toolbox. Spatial Vision. 1997;10:433–436. doi: 10.1163/156856897X00357. [DOI] [PubMed] [Google Scholar]

[bib10] Burt JB, Helmer M, Shinn M, Anticevic A, Murray JD. Generative modeling of brain maps with spatial autocorrelation. NeuroImage. 2020;220:117038. doi: 10.1016/j.neuroimage.2020.117038. [DOI] [PubMed] [Google Scholar]

[bib11] Carey D, Caprini F, Allen M, Lutti A, Weiskopf N, Rees G, Callaghan MF, Dick F. Quantitative MRI provides markers of intra-, inter-regional, and age-related differences in young adult cortical microstructure. NeuroImage. 2018;182:429–440. doi: 10.1016/j.neuroimage.2017.11.066. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib12] Chaimow D, Yacoub E, Uğurbil K, Shmuel A. Spatial specificity of the functional MRI blood oxygenation response relative to neuronal activity. NeuroImage. 2018;164:32–47. doi: 10.1016/j.neuroimage.2017.08.077. [DOI] [PubMed] [Google Scholar]

[bib13] Chen G, Lu HD, Roe AW. A map for horizontal disparity in monkey V2. Neuron. 2008;58:442–450. doi: 10.1016/j.neuron.2008.02.032. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib14] Cohen-Adad J, Polimeni JR, Helmer KG, Benner T, McNab JA, Wald LL, Rosen BR, Mainero C. T₂* mapping and B₀ orientation-dependence at 7 T reveal cyto- and myeloarchitecture organization of the human cortex. NeuroImage. 2012;60:1006–1014. doi: 10.1016/j.neuroimage.2012.01.053. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib15] Cox RW. AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Computers and Biomedical Research, an International Journal. 1996;29:162–173. doi: 10.1006/cbmr.1996.0014. [DOI] [PubMed] [Google Scholar]

[bib16] Dale AM, Fischl B, Sereno MI. Cortical surface-based analysis. I. segmentation and surface reconstruction. NeuroImage. 1999;9:179–194. doi: 10.1006/nimg.1998.0395. [DOI] [PubMed] [Google Scholar]

[bib17] Destrieux C, Fischl B, Dale A, Halgren E. Automatic parcellation of human cortical gyri and sulci using standard anatomical nomenclature. NeuroImage. 2010;53:1–15. doi: 10.1016/j.neuroimage.2010.06.010. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib18] Dick F, Tierney AT, Lutti A, Josephs O, Sereno MI, Weiskopf N. In vivo functional and myeloarchitectonic mapping of human primary auditory areas. The Journal of Neuroscience. 2012;32:16095–16105. doi: 10.1523/JNEUROSCI.1712-12.2012. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib19] Dobkins KR, Thiele A, Albright TD. Comparison of red-green equiluminance points in humans and macaques: evidence for different L: M cone ratios between species. Journal of the Optical Society of America. A, Optics, Image Science, and Vision. 2000;17:545–556. doi: 10.1364/josaa.17.000545. [DOI] [PubMed] [Google Scholar]

[bib20] Dumoulin SO, Harvey BM, Fracasso A, Zuiderbaan W, Luijten PR, Wandell BA, Petridou N. In vivo evidence of functional and anatomical stripe-based subdivisions in human V2 and V3. Scientific Reports. 2017;7:733. doi: 10.1038/s41598-017-00634-6. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib21] Edwards LJ, Kirilina E, Mohammadi S, Weiskopf N. Microstructural imaging of human neocortex in vivo. NeuroImage. 2018;182:184–206. doi: 10.1016/j.neuroimage.2018.02.055. [DOI] [PubMed] [Google Scholar]

[bib22] Edwards LJ, Pine KJ, Helms G, Weiskopf N. Rational approximation of the Ernst equation for dual angle R1 mapping revisited: beyond the small flip-angle assumption. In Book of Abstracts ESMRMB 2021 Online 38th Annual Scientific Meeting 7–9 October 2021. Magn Reson Mater Phy.2021. [Google Scholar]

[bib23] Engel SA, Glover GH, Wandell BA. Retinotopic organization in human visual cortex and the spatial precision of functional MRI. Cerebral Cortex. 1997;7:181–192. doi: 10.1093/cercor/7.2.181. [DOI] [PubMed] [Google Scholar]

[bib24] Feinberg DA, Moeller S, Smith SM, Auerbach E, Ramanna S, Gunther M, Glasser MF, Miller KL, Ugurbil K, Yacoub E. Multiplexed echo planar imaging for sub-second whole brain fMRI and fast diffusion imaging. PLOS ONE. 2010;5:e15710. doi: 10.1371/journal.pone.0015710. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib25] Fischl B, Sereno MI, Dale AM. Cortical surface-based analysis. II: inflation, flattening, and a surface-based coordinate system. NeuroImage. 1999;9:195–207. doi: 10.1006/nimg.1998.0396. [DOI] [PubMed] [Google Scholar]

[bib26] Fischl B, Rajendran N, Busa E, Augustinack J, Hinds O, Yeo BTT, Mohlberg H, Amunts K, Zilles K. Cortical folding patterns and predicting cytoarchitecture. Cerebral Cortex. 2008;18:1973–1980. doi: 10.1093/cercor/bhm225. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib27] Flechsig P. Anatomie des menschlichen Gehirns und Rückenmarks auf myelogenetischer Grundlage. Leipzig: Georg Thieme; 1920. [Google Scholar]

[bib28] Fujimoto K, Polimeni JR, van der Kouwe AJW, Reuter M, Kober T, Benner T, Fischl B, Wald LL. Quantitative comparison of cortical surface reconstructions from MP2RAGE and multi-echo MPRAGE data at 3 and 7 T. NeuroImage. 2014;90:60–73. doi: 10.1016/j.neuroimage.2013.12.012. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib29] Glasser MF, Van Essen DC. Mapping human cortical areas in vivo based on myelin content as revealed by T1- and T2-weighted MRI. The Journal of Neuroscience. 2011;31:11597–11616. doi: 10.1523/JNEUROSCI.2180-11.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib30] Glasser MF, Sotiropoulos SN, Wilson JA, Coalson TS, Fischl B, Andersson JL, Xu J, Jbabdi S, Webster M, Polimeni JR, Van Essen DC, Jenkinson M, WU-Minn HCP Consortium The minimal preprocessing pipelines for the human connectome project. NeuroImage. 2013;80:105–124. doi: 10.1016/j.neuroimage.2013.04.127. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib31] Greve DN, Fischl B. Accurate and robust brain image alignment using boundary-based registration. NeuroImage. 2009;48:63–72. doi: 10.1016/j.neuroimage.2009.06.060. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib32] Griswold MA, Jakob PM, Heidemann RM, Nittka M, Jellus V, Wang J, Kiefer B, Haase A. Generalized autocalibrating partially parallel acquisitions (GRAPPA) Magnetic Resonance in Medicine. 2002;47:1202–1210. doi: 10.1002/mrm.10171. [DOI] [PubMed] [Google Scholar]

[bib33] Helms G, Dathe H, Dechent P. Quantitative FLASH MRI at 3T using a rational approximation of the Ernst equation. Magnetic Resonance in Medicine. 2008;59:667–672. doi: 10.1002/mrm.21542. [DOI] [PubMed] [Google Scholar]

[bib34] Hinds OP, Rajendran N, Polimeni JR, Augustinack JC, Wiggins G, Wald LL, Diana Rosas H, Potthast A, Schwartz EL, Fischl B. Accurate prediction of V1 location from cortical folds in a surface coordinate system. NeuroImage. 2008;39:1585–1599. doi: 10.1016/j.neuroimage.2007.10.033. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib35] Hockfield S, Tootell RBH, Zaremba S. Molecular differences among neurons reveal an organization of human visual cortex. PNAS. 1990;87:3027–3031. doi: 10.1073/pnas.87.8.3027. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib36] Horton JC, Hocking DR. Myelin patterns in V1 and V2 of normal and monocularly enucleated monkeys. Cerebral Cortex. 1997;7:166–177. doi: 10.1093/cercor/7.2.166. [DOI] [PubMed] [Google Scholar]

[bib37] Hubel DH, Livingstone MS. Segregation of form, color, and stereopsis in primate area 18. The Journal of Neuroscience. 1987;7:3378–3415. doi: 10.1523/JNEUROSCI.07-11-03378.1987. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib38] Ives FE. A new color meter. Journal of the Franklin Institute. 1907;164:47–56. doi: 10.1016/S0016-0032(07)90164-7. [DOI] [Google Scholar]

[bib39] Jenkinson M, Bannister P, Brady M, Smith S. Improved optimization for the robust and accurate linear registration and motion correction of brain images. NeuroImage. 2002;17:825–841. doi: 10.1016/s1053-8119(02)91132-8. [DOI] [PubMed] [Google Scholar]

[bib40] Keller AL, Schüz A, Logothetis NK, Weber B. Vascularization of cytochrome oxidase-rich blobs in the primary visual cortex of squirrel and macaque monkeys. The Journal of Neuroscience. 2011;31:1246–1253. doi: 10.1523/JNEUROSCI.2765-10.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib41] Kirilina E, Helbling S, Morawski M, Pine K, Reimann K, Jankuhn S, Dinse J, Deistung A, Reichenbach JR, Trampel R, Geyer S, Müller L, Jakubowski N, Arendt T, Bazin PL, Weiskopf N. Superficial white matter imaging: contrast mechanisms and whole-brain in vivo mapping. Science Advances. 2020;6:eaaz9281. doi: 10.1126/sciadv.aaz9281. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib42] Kleiner M, Brainard DH, Pelli D, Ingling A, Murray R, Broussard C. What’s new in psychtoolbox-3. Perception. 2007;36:1–16. [Google Scholar]

[bib43] Krubitzer LA, Kaas JH. Cortical integration of parallel pathways in the visual system of primates. Brain Research. 1989;478:161–165. doi: 10.1016/0006-8993(89)91490-x. [DOI] [PubMed] [Google Scholar]

[bib44] Lazari A, Lipp I. Can MRI measure myelin? Systematic review, qualitative assessment, and meta-analysis of studies validating microstructural imaging with myelin histology. NeuroImage. 2021;230:117744. doi: 10.1016/j.neuroimage.2021.117744. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib45] Leuze C, Aswendt M, Ferenczi E, Liu CW, Hsueh B, Goubran M, Tian Q, Steinberg G, Zeineh MM, Deisseroth K, McNab JA. The separate effects of lipids and proteins on brain MRI contrast revealed through tissue clearing. NeuroImage. 2017;156:412–422. doi: 10.1016/j.neuroimage.2017.04.021. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib46] Li X, Zhu Q, Janssens T, Arsenault JT, Vanduffel W. In vivo identification of thick, thin, and pale stripes of macaque area V2 using submillimeter resolution (f)MRI at 3 T. Cerebral Cortex. 2019;29:544–560. doi: 10.1093/cercor/bhx337. [DOI] [PubMed] [Google Scholar]

[bib47] Livingstone MS, Hubel DH. Thalamic inputs to cytochrome oxidase-rich regions in monkey visual cortex. PNAS. 1982;79:6098–6101. doi: 10.1073/pnas.79.19.6098. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib48] Livingstone MS, Hubel DH. Psychophysical evidence for separate channels for the perception of form, color, movement, and depth. The Journal of Neuroscience. 1987;7:3416–3468. doi: 10.1523/JNEUROSCI.07-11-03416.1987. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib49] Lutti A, Hutton C, Finsterbusch J, Helms G, Weiskopf N. Optimization and validation of methods for mapping of the radiofrequency transmit field at 3T. Magnetic Resonance in Medicine. 2010;64:229–238. doi: 10.1002/mrm.22421. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib50] Lutti A, Stadler J, Josephs O, Windischberger C, Speck O, Bernarding J, Hutton C, Weiskopf N. Robust and fast whole brain mapping of the RF transmit field B1 at 7T. PLOS ONE. 2012;7:e32379. doi: 10.1371/journal.pone.0032379. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib51] Lutti A, Dick F, Sereno MI, Weiskopf N. Using high-resolution quantitative mapping of R1 as an index of cortical myelination. NeuroImage. 2014;93:176–188. doi: 10.1016/j.neuroimage.2013.06.005. [DOI] [PubMed] [Google Scholar]

[bib52] Marques JP, Kober T, Krueger G, van der Zwaag W, Van de Moortele PF, Gruetter R. MP2RAGE, a self bias-field corrected sequence for improved segmentation and T1-mapping at high field. NeuroImage. 2010;49:1271–1281. doi: 10.1016/j.neuroimage.2009.10.002. [DOI] [PubMed] [Google Scholar]

[bib53] Marques JP, Khabipova D, Gruetter R. Studying cyto and myeloarchitecture of the human cortex at ultra-high field with quantitative imaging: R1, R2* and magnetic susceptibility. NeuroImage. 2017;147:152–163. doi: 10.1016/j.neuroimage.2016.12.009. [DOI] [PubMed] [Google Scholar]

[bib54] McColgan P, Helbling S, Vaculčiaková L, Pine K, Wagstyl K, Attar FM, Edwards L, Papoutsi M, Wei Y, Van den Heuvel MP, Tabrizi SJ, Rees G, Weiskopf N. Relating quantitative 7T MRI across cortical depths to cytoarchitectonics, gene expression and connectomics. Human Brain Mapping. 2021;42:4996–5009. doi: 10.1002/hbm.25595. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib55] Mezer A, Yeatman JD, Stikov N, Kay KN, Cho NJ, Dougherty RF, Perry ML, Parvizi J, Hua LH, Butts-Pauly K, Wandell BA. Quantifying the local tissue volume and composition in individual brains with magnetic resonance imaging. Nature Medicine. 2013;19:1667–1672. doi: 10.1038/nm.3390. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib56] Moeller S, Yacoub E, Olman CA, Auerbach E, Strupp J, Harel N, Uğurbil K. Multiband multislice GE-EPI at 7 Tesla, with 16-fold acceleration using partial parallel imaging with application to high spatial and temporal whole-brain fMRI. Magnetic Resonance in Medicine. 2010;63:1144–1153. doi: 10.1002/mrm.22361. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib57] Movahedian Attar F, Kirilina E, Haenelt D, Pine KJ, Trampel R, Edwards LJ, Weiskopf N. Mapping short association fibers in the early cortical visual processing stream using in vivo diffusion tractography. Cerebral Cortex. 2020;30:4496–4514. doi: 10.1093/cercor/bhaa049. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib58] Nasr S, Polimeni JR, Tootell RBH. Interdigitated color- and disparity-selective columns within human visual cortical areas V2 and V3. The Journal of Neuroscience. 2016;36:1841–1857. doi: 10.1523/JNEUROSCI.3518-15.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib59] Nasr S, Tootell RBH. Columnar organization of mid-spectral and end-spectral hue preferences in human visual cortex. NeuroImage. 2018;181:748–759. doi: 10.1016/j.neuroimage.2018.07.053. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib60] Navarro KT, Sanchez MJ, Engel SA, Olman CA, Weldon KB. Depth-dependent functional MRI responses to chromatic and achromatic stimuli throughout V1 and V2. NeuroImage. 2021;226:117520. doi: 10.1016/j.neuroimage.2020.117520. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib61] Olman CA, Inati S, Heeger DJ. The effect of large veins on spatial localization with GE BOLD at 3 T: displacement, not blurring. NeuroImage. 2007;34:1126–1135. doi: 10.1016/j.neuroimage.2006.08.045. [DOI] [PubMed] [Google Scholar]

[bib62] Pelli DG. The videotoolbox software for visual psychophysics: transforming numbers into movies. Spatial Vision. 1997;10:437–442. [PubMed] [Google Scholar]

[bib63] Peterhans E, von der Heydt R. Functional organization of area V2 in the alert macaque. The European Journal of Neuroscience. 1993;5:509–524. doi: 10.1111/j.1460-9568.1993.tb00517.x. [DOI] [PubMed] [Google Scholar]

[bib64] Peters AM, Brookes MJ, Hoogenraad FG, Gowland PA, Francis ST, Morris PG, Bowtell R. T2* measurements in human brain at 1.5, 3 and 7 T. Magnetic Resonance Imaging. 2007;25:748–753. doi: 10.1016/j.mri.2007.02.014. [DOI] [PubMed] [Google Scholar]

[bib65] Pistorio AL, Hendry SH, Wang X. A modified technique for high-resolution staining of myelin. Journal of Neuroscience Methods. 2006;153:135–146. doi: 10.1016/j.jneumeth.2005.10.014. [DOI] [PubMed] [Google Scholar]

[bib66] Pluim JPW, Muenzing SEA, Eppenhof KAJ, Murphy K. The truth is hard to make: Validation of medical image registration. 2016 23rd International Conference on Pattern Recognition (ICPR; Cancun. 2016. [DOI] [Google Scholar]

[bib67] Polimeni JR, Fischl B, Greve DN, Wald LL. Laminar analysis of 7T BOLD using an imposed spatial activation pattern in human V1. NeuroImage. 2010;52:1334–1346. doi: 10.1016/j.neuroimage.2010.05.005. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib68] Sanders FK, Whitteridge D. Conduction velocity and myelin thickness in regenerating nerve fibres. The Journal of Physiology. 1946;105:152–174. doi: 10.1113/jphysiol.1946.sp004160. [DOI] [PubMed] [Google Scholar]

[bib69] Savaskan NE, Weinmann O, Heimrich B, Eyupoglu IY. High resolution neurochemical gold staining method for myelin in peripheral and central nervous system at the light- and electron-microscopic level. Cell and Tissue Research. 2009;337:213–221. doi: 10.1007/s00441-009-0815-9. [DOI] [PubMed] [Google Scholar]

[bib70] Sereno MI, Dale AM, Reppas JB, Kwong KK, Belliveau JW, Brady TJ, Rosen BR, Tootell RBH. Borders of multiple visual areas in humans revealed by functional magnetic resonance imaging. Science. 1995;268:889–893. doi: 10.1126/science.7754376. [DOI] [PubMed] [Google Scholar]

[bib71] Sereno MI, Lutti A, Weiskopf N, Dick F. Mapping the human cortical surface by combining quantitative T1 with retinotopy. Cerebral Cortex. 2013;23:2261–2268. doi: 10.1093/cercor/bhs213. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib72] Shipp S, Zeki S. Segregation of pathways leading from area V2 to areas V4 and V5 of macaque monkey visual cortex. Nature. 1985;315:322–325. doi: 10.1038/315322a0. [DOI] [PubMed] [Google Scholar]

[bib73] Sigalovsky IS, Fischl B, Melcher JR. Mapping an intrinsic MR property of gray matter in auditory cortex of living humans: a possible marker for primary cortex and hemispheric differences. NeuroImage. 2006;32:1524–1537. doi: 10.1016/j.neuroimage.2006.05.023. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib74] Sincich LC, Horton JC. Pale cytochrome oxidase stripes in V2 receive the richest projection from macaque striate cortex. The Journal of Comparative Neurology. 2002;447:18–33. doi: 10.1002/cne.10174. [DOI] [PubMed] [Google Scholar]

[bib75] Sincich LC, Horton JC. The circuitry of V1 and V2: integration of color, form, and motion. Annual Review of Neuroscience. 2005;28:303–326. doi: 10.1146/annurev.neuro.28.061604.135731. [DOI] [PubMed] [Google Scholar]

[bib76] Smart IHM, McSherry GM. Gyrus formation in the cerebral cortex of the ferret. II. description of the internal histological changes. Journal of Anatomy. 1986;147:27–43. [PMC free article] [PubMed] [Google Scholar]

[bib77] Stüber C, Morawski M, Schäfer A, Labadie C, Wähnert M, Leuze C, Streicher M, Barapatre N, Reimann K, Geyer S, Spemann D, Turner R. Myelin and iron concentration in the human brain: a quantitative study of MRI contrast. NeuroImage. 2014;93 Pt 1:95–106. doi: 10.1016/j.neuroimage.2014.02.026. [DOI] [PubMed] [Google Scholar]

[bib78] Subramaniam K, Brown TB. Gradunwarp. 1.0.2GitHub. 2014 https://github.com/Washington-University/gradunwarp

[bib79] Tabelow K, Balteau E, Ashburner J, Callaghan MF, Draganski B, Helms G, Kherif F, Leutritz T, Lutti A, Phillips C, Reimer E, Ruthotto L, Seif M, Weiskopf N, Ziegler G, Mohammadi S. hMRI – a toolbox for quantitative MRI in neuroscience and clinical research. NeuroImage. 2019;194:191–210. doi: 10.1016/j.neuroimage.2019.01.029. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib80] Tofts PS. In: Quantitative MRI of the Brain: Principles of Physical Measurement. Tofts PS, editor. Boca Raton: CRC Press; 2018. PD: proton density of tissue water; pp. 55–71. [DOI] [Google Scholar]

[bib81] Tootell RBH, Silverman MS, De Valois RL, Jacobs GH. Functional organization of the second cortical visual area in primates. Science. 1983;220:737–739. doi: 10.1126/science.6301017. [DOI] [PubMed] [Google Scholar]

[bib82] Tootell RBH, Hamilton SL. Functional anatomy of the second visual area (V2) in the macaque. The Journal of Neuroscience. 1989;9:2620–2644. doi: 10.1523/JNEUROSCI.09-08-02620.1989. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib83] Tootell RBH, Taylor JB. Anatomical evidence for MT and additional cortical visual areas in humans. Cerebral Cortex. 1995;5:39–55. doi: 10.1093/cercor/5.1.39. [DOI] [PubMed] [Google Scholar]

[bib84] Tootell RBH, Nelissen K, Vanduffel W, Orban GA. Search for color ‘center(s)’ in macaque visual cortex. Cerebral Cortex. 2004;14:353–363. doi: 10.1093/cercor/bhh001. [DOI] [PubMed] [Google Scholar]

[bib85] Tootell RBH, Nasr S. Columnar segregation of magnocellular and parvocellular streams in human extrastriate cortex. The Journal of Neuroscience. 2017;37:8014–8032. doi: 10.1523/JNEUROSCI.0690-17.2017. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib86] Trampel R, Bazin PL, Pine K, Weiskopf N. In-vivo magnetic resonance imaging (MRI) of laminae in the human cortex. NeuroImage. 2019;197:707–715. doi: 10.1016/j.neuroimage.2017.09.037. [DOI] [PubMed] [Google Scholar]

[bib87] Tsao DY, Vanduffel W, Sasaki Y, Fize D, Knutsen TA, Mandeville JB, Wald LL, Dale AM, Rosen BR, Van Essen DC, Livingstone MS, Orban GA, Tootell RBH. Stereopsis activates V3A and caudal intraparietal areas in macaques and humans. Neuron. 2003;39:555–568. doi: 10.1016/s0896-6273(03)00459-8. [DOI] [PubMed] [Google Scholar]

[bib88] Ts’o DY, Roe AW, Gilbert CD. A hierarchy of the functional organization for color, form and disparity in primate visual area V2. Vision Research. 2001;41:1333–1349. doi: 10.1016/s0042-6989(01)00076-1. [DOI] [PubMed] [Google Scholar]

[bib89] Tustison NJ, Avants BB, Cook PA, Zheng Y, Egan A, Yushkevich PA, Gee JC. N4ITK: improved N3 bias correction. IEEE Transactions on Medical Imaging. 2010;29:1310–1320. doi: 10.1109/TMI.2010.2046908. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib90] Vaculčiaková L, Podranski K, Edwards LJ, Ocal D, Veale T, Fox NC, Haak R, Ehses P, Callaghan MF, Pine KJ, Weiskopf N. Combining navigator and optical prospective motion correction for high-quality 500 μm resolution quantitative multi-parameter mapping at 7T. Magnetic Resonance in Medicine. 2022;88:787–801. doi: 10.1002/mrm.29253. [DOI] [PubMed] [Google Scholar]

[bib91] Volz S, Nöth U, Deichmann R. Correction of systematic errors in quantitative proton density mapping. Magnetic Resonance in Medicine. 2012;68:74–85. doi: 10.1002/mrm.23206. [DOI] [PubMed] [Google Scholar]

[bib92] Waehnert MD, Dinse J, Weiss M, Streicher MN, Waehnert P, Geyer S, Turner R, Bazin PL. Anatomically motivated modeling of cortical laminae. NeuroImage. 2014;93 Pt 2:210–220. doi: 10.1016/j.neuroimage.2013.03.078. [DOI] [PubMed] [Google Scholar]

[bib93] Wagstyl K, Paquola C, Bethlehem R, Evans AC, Huth A. Equivolumetric layering for mesh surfaces. Zenodo. 2018 doi: 10.5281/zenodo.1442584. [DOI]

[bib94] Weiskopf N, Lutti A, Helms G, Novak M, Ashburner J, Hutton C. Unified segmentation based correction of R1 brain maps for RF transmit field inhomogeneities (UNICORT) NeuroImage. 2011;54:2116–2124. doi: 10.1016/j.neuroimage.2010.10.023. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib95] Weiskopf N, Callaghan MF, Josephs O, Lutti A, Mohammadi S. Estimating the apparent transverse relaxation time (R2(*)) from images with different contrasts (ESTATICS) reduces motion artifacts. Frontiers in Neuroscience. 2014;8:278. doi: 10.3389/fnins.2014.00278. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib96] Weiskopf N, Edwards LJ, Helms G, Mohammadi S, Kirilina E. Quantitative magnetic resonance imaging of brain anatomy and in vivo histology. Nature Reviews Physics. 2021;3:570–588. doi: 10.1038/s42254-021-00326-1. [DOI] [Google Scholar]

[bib97] Xiao Y, Wang Y, Felleman DJ. A spatially organized representation of colour in macaque cortical area V2. Nature. 2003;421:535–539. doi: 10.1038/nature01372. [DOI] [PubMed] [Google Scholar]

[bib98] Yacoub E, Shmuel A, Pfeuffer J, Van De Moortele PF, Adriany G, Andersen P, Vaughan JT, Merkle H, Ugurbil K, Hu X. Imaging brain function in humans at 7 Tesla. Magnetic Resonance in Medicine. 2001;45:588–594. doi: 10.1002/mrm.1080. [DOI] [PubMed] [Google Scholar]

[bib99] Zaretskaya N, Fischl B, Reuter M, Renvall V, Polimeni JR. Advantages of cortical surface reconstruction using submillimeter 7 T MEMPRAGE. NeuroImage. 2018;165:11–26. doi: 10.1016/j.neuroimage.2017.09.060. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib100] Zheng D, LaMantia AS, Purves D. Specialized vascularization of the primate visual cortex. The Journal of Neuroscience. 1991;11:2622–2629. doi: 10.1523/JNEUROSCI.11-08-02622.1991. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

High-resolution quantitative and functional MRI indicate lower myelination of thin and thick stripes in human secondary visual cortex

Daniel Haenelt

Robert Trampel

Shahin Nasr

Jonathan R Polimeni

Roger BH Tootell

Martin I Sereno

Kerrin J Pine

Luke J Edwards

Saskia Helbling

Nikolaus Weiskopf

Roles

Abstract

Introduction

Results

Figure 1. General overview of acquired magnetic resonance (MR) data and their use in the analysis.

Functional mapping of color-selective and disparity-selective stripes

Figure 2. Activation maps for color-selective thin and disparity-selective thick stripes.

Figure 2—figure supplement 1. Eccentricity and polar angle maps.

Figure 2—figure supplement 2. Exemplary activation maps from another participant.

Figure 3. Repeatability of functional magnetic resonance imaging (fMRI) activation maps across scanning sessions.

Figure 3—figure supplement 1. Correlation plots for single participants (subjects 1–2).

Figure 3—figure supplement 2. Correlation plots for single participants (subjects 3–4).

Consistent qMRI maps across cortical regions and cortical depth

Figure 4. Illustration of quantitative R1 across cortical areas.

Figure 4—figure supplement 1. Illustration of quantitative R2* across cortical areas.

Figure 4—figure supplement 2. Mean quantitative R1 (MP2RAGE) and PD values across cortical areas.

Higher myelination of pale stripes

Figure 5. Comparison of quantitative R1 and R2* values between V2 stripe types.

Figure 5—figure supplement 1. Comparison of quantitative R1 and R2* values between V2 stripe types excluding foveal parts.

Figure 5—figure supplement 2. Comparison of quantitative R1 values (MP2RAGE) between V2 stripe types.

Figure 5—figure supplement 3. Comparison of quantitative R1 values (MP2RAGE) between V2 stripe types excluding foveal parts.

Discussion

Figure 6. Illustration of segmentation and registration quality.

Figure 6—video 1. Illustration of registration quality (subject 1).

Figure 6—video 2. Illustration of registration quality (subject 2).

Figure 6—video 3. Illustration of registration quality (subject 3).

Figure 6—video 4. Illustration of registration quality (subject 4).

Materials and methods

Participants

General procedures

Visual stimulation

Experiment 1: Color-selective thin stripes

Experiment 2: Disparity-selective thick stripes

Retinotopic mapping

Imaging

Data analysis

Reliability analysis of fMRI sessions

Quantitative comparison of qMRI parameters between stripe types

Acknowledgements

Appendix 1

Appendix 1—figure 1. Individual functional magnetic resonance imaging (fMRI) activation maps and quantitative magnetic resonance imaging (qMRI) maps (subject 1).

Appendix 1—figure 2. Individual functional magnetic resonance imaging (fMRI) activation maps and quantitative magnetic resonance imaging (qMRI) maps (subject 2).

Appendix 1—figure 3. Individual functional magnetic resonance imaging (fMRI) activation maps and quantitative magnetic resonance imaging (qMRI) maps (subject 3).

Appendix 1—figure 4. Individual functional magnetic resonance imaging (fMRI) activation maps and quantitative magnetic resonance imaging (qMRI) maps (subject 4).

Appendix 2

Appendix 2—figure 1. Cortical profiles of mean quantitative magnetic resonance imaging (qMRI) parameters within V2.

Appendix 3

Appendix 3—figure 1. Visualization of thin and thick stripe regions of interest (ROIs).

Appendix 4

Appendix 4—figure 1. Sizes of stripe regions of interest (ROIs) for each threshold level.

Appendix 5

Appendix 5—figure 1. Exemplary motion data from one multi-parameter mapping (MPM) measurement.

Appendix 6

Appendix 6—figure 1. Measured mean isoluminance points across participants and sessions at different eccentricities.

Appendix 7

Analysis of registration accuracy

Appendix 7—figure 1. Quantitative analysis of registration accuracy.

Funding Statement

Contributor Information

Funding Information

Additional information

Competing interests

Author contributions

Ethics

Additional files

Data availability

References

Editor's evaluation

Figure 4. Illustration of quantitative R₁ across cortical areas.

Figure 4—figure supplement 1. Illustration of quantitative R₂* across cortical areas.

Figure 4—figure supplement 2. Mean quantitative R₁ (MP2RAGE) and PD values across cortical areas.

Figure 5. Comparison of quantitative R1 and R₂* values between V2 stripe types.

Figure 5—figure supplement 1. Comparison of quantitative R₁ and R₂* values between V2 stripe types excluding foveal parts.

Figure 5—figure supplement 2. Comparison of quantitative R₁ values (MP2RAGE) between V2 stripe types.

Figure 5—figure supplement 3. Comparison of quantitative R₁ values (MP2RAGE) between V2 stripe types excluding foveal parts.