Skip to main content
NCBI home page
eLife logoLink to eLife
. 2023 Apr 3;12:e83727. doi: 10.7554/eLife.83727

Spatiotemporal tissue maturation of thalamocortical pathways in the human fetal brain

Siân Wilson 1,2, Maximilian Pietsch 1, Lucilio Cordero-Grande 1,3,4, Daan Christiaens 1,5, Alena Uus 6, Vyacheslav R Karolis 1, Vanessa Kyriakopoulou 1, Kathleen Colford 1, Anthony N Price 1, Jana Hutter 1, Mary A Rutherford 1, Emer J Hughes 1, Serena J Counsell 1, Jacques-Donald Tournier 1, Joseph V Hajnal 1, A David Edwards 1,2, Jonathan O’Muircheartaigh 1,2,7,8,✉, Tomoki Arichi 1,2,9,10,✉
Editors: Finnegan J Calabro11, Jonathan Roiser12
PMCID: PMC10125021  PMID: 37010273

Abstract

The development of connectivity between the thalamus and maturing cortex is a fundamental process in the second half of human gestation, establishing the neural circuits that are the basis for several important brain functions. In this study, we acquired high-resolution in utero diffusion magnetic resonance imaging (MRI) from 140 fetuses as part of the Developing Human Connectome Project, to examine the emergence of thalamocortical white matter over the second to third trimester. We delineate developing thalamocortical pathways and parcellate the fetal thalamus according to its cortical connectivity using diffusion tractography. We then quantify microstructural tissue components along the tracts in fetal compartments that are critical substrates for white matter maturation, such as the subplate and intermediate zone. We identify patterns of change in the diffusion metrics that reflect critical neurobiological transitions occurring in the second to third trimester, such as the disassembly of radial glial scaffolding and the lamination of the cortical plate. These maturational trajectories of MR signal in transient fetal compartments provide a normative reference to complement histological knowledge, facilitating future studies to establish how developmental disruptions in these regions contribute to pathophysiology.

Research organism: Human

Introduction

Thalamocortical connections provide important inputs into the developing cortex during the second half of human gestation, where they play a key role in guiding cortical areal differentiation and establishing the circuitry responsible for sensory integration across the lifespan (Jones, 2007; Price et al., 2006; Schummers et al., 2005; Sharma et al., 2000; Sur and Rubenstein, 2005). Their importance is highlighted by previous work implicating disruptions to thalamocortical development during the perinatal period in the pathophysiology of neurodevelopmental disorders, such as schizophrenia (Klingner et al., 2014; Marenco et al., 2012), bipolar disorder (Anticevic et al., 2014), and autism (Nair et al., 2013). Altered thalamocortical connectivity has also been described in preterm infants, and was used to predict cognitive outcome (Ball et al., 2013; Ball et al., 2015; Toulmin et al., 2021), highlighting the specific vulnerability of these pathways during the second to third trimester. Although thalamocortical development has been studied in rodents and non-human primates (Brody et al., 1987; Kostović and Jovanov-Milosević, 2006; Molnár and Blakemore, 1995; Yakovlev et al., 1960) and post-mortem human tissue (Krsnik et al., 2017; Takahashi et al., 2012; Wilkinson et al., 2017), little is known about in vivo white matter maturation during fetal development.

White matter development in the late second and third trimesters of human gestation (between 21 and 37 weeks) is characterised by a sequence of precisely timed biological processes occurring in transient compartments of the fetal brain. These processes include the migration of neurons along the radial glial scaffold, accumulation of thalamocortical axons in the superficial subplate, innervation of the target cortical area, conversion of radial glial cells into astrocytes, and ensheathment of axonal fibres (Krsnik et al., 2017; Molliver et al., 1973; Kostović et al., 2002, Kostovic and Judas, 2006). The challenge for in vivo neuroimaging studies is to disentangle the effect of these different neurobiological processes on the diffusion magnetic resonance imaging (dMRI) signal, to improve mechanistic insight about the transformation of transient fetal compartments into segments of developing white matter (Kostovic 2012).

Recent advances in diffusion weighted imaging now allow in vivo characterisation and estimation of white matter development during the fetal period. Tractography has been used to estimate the fetal brain’s major white matter bundles and quantitatively characterise the evolution of the microstructure across the second half of gestation (Bui et al., 2006; Zanin et al., 2011; Jaimes et al., 2020; Jakab et al., 2015; Keunen et al., 2018; Khan et al., 2019; Machado-Rivas et al., 2021; Wilson et al., 2021). Advanced acquisition and analysis methods enable the relative contribution of constituent tissue and fluid compartments to the diffusion signal to be estimated (Jeurissen et al., 2014; Pietsch et al., 2019). Using this approach, previous work has identified non-linear trends in diffusion metrics over the second to third trimester (Wilson et al., 2021). Namely, we observed an initial decrease in tissue fraction within developing white matter between 22 and 29 weeks, which could be due to the radial glial scaffold disassembling (Rakic, 2003). Subsequently, we observed an increase from 30 to 36 weeks, potentially linked to more coherent fibre organisation, axonal outgrowth, and ensheathment (Wimberger et al., 1995; Back et al., 2002; Haynes et al., 2005), increasing the structural integrity of maturing white matter. Interpreting these trends is especially challenging in the rapidly developing fetal brain, because of the high sensitivity and low specificity of diffusion metrics to various co-occurring biological processes.

We hypothesise that the biological processes occurring in different fetal compartments leads to predictable changes in diffusion metrics along tracts, reflecting the appearance and resolution of these transient zones. When a mean value across the whole tract is calculated, sensitivity to the unique neurobiological properties of each transient compartment is lost. For example, in the early prenatal and mid prenatal period, the subplate is a highly water-rich compartment containing extracellular matrix, whereas the cortical plate and the deep grey matter are relatively cell dense (Kostović, 2020). We therefore predict that the tissue fraction would be higher in the deep grey matter and the cortical plate and lower in the subplate. We investigate this by characterising the entire trajectory of tissue composition changes between the thalamus and the cortex, to explore the role of transient fetal brain developmental structures on white matter maturational trajectories.

We acquired diffusion weighted imaging from 140 fetuses over a wide gestational age (GA) range (21–37 weeks) and use tractography to delineate five distinct thalamocortical pathways. To investigate whether the immature axonal bundles can be traced back to specific and distinct locations within thalamus, we parcellate the thalamus according to streamline connectivity (Behrens et al., 2003). We find consistent and distinct origins of different tracts, resembling the adult topology of thalamic nuclei (Toulmin et al., 2015; Behrens et al., 2003) as early as 23 weeks’ gestation. We then apply a multi-shell multi-tissue constrained spherical deconvolution (MSMT-CSD) diffusion model (Jeurissen et al., 2014) and derive tissue and fluid fraction values, charting tract-specific maturational profiles over the second to third trimester. We overlay the tracts on an atlas of transitioning fetal compartments and correlate changes in the dMRI signal across time with critical neurodevelopmental processes, such as the dissolution of the subplate and lamination of the cortical plate. We demonstrate that along-tract sampling of diffusion metrics can capture temporal and compartmental differences in the second to third trimester, reflecting the maturing neurobiology of the fetal brain described in histological studies. With these methods, we provide a detailed, accurate reference of the unique developing microstructure in each tract that improves mechanistic insight about fibre maturation, bridging the gap between MRI and histology.

Results

Estimating thalamocortical pathways using probabilistic streamline tractography

High-angular-resolution multi-shell diffusion weighted imaging (HARDI) was acquired from 140 fetuses between 21 and 37 gestational weeks (70 male, 70 female) as part of the Developing Human Connectome Project (dHCP). Data were corrected for fetal head motion and other imaging artefacts (Christiaens et al., 2021). Individual subject orientation density functions (ODFs) were then computed using cohort-specific fluid and ‘tissue’ response functions and compiled to generate weekly diffusion templates (see Materials and methods). The diffusion templates were then registered to a T2-weighted brain atlas (Gholipour et al., 2017) of tissue segmentations, used to generate anatomically constrained whole-brain connectomes for each gestational week (Smith et al., 2012; Tournier et al., 2019). To constrain our investigation, we selected thalamocortical pathways that are at a critical stage in their development and are vulnerable to external influences in the second to third trimester (Batalle et al., 2017; Nosarti et al., 2014; Raybaud et al., 2013), the anterior thalamic radiation (AT), thalamic-motor tract (TM), thalamic-sensory tract (TS), posterior parietal tract (PP), and optic radiation (OR). The connectomes were filtered down to the pathways of interest using inclusion regions defined by the T2 atlas, including the thalamus and specific cortical areas (Figure 1). These included the primary motor cortex, primary sensory cortex, posterior parietal cortex, dorso-lateral prefrontal cortex, and the primary visual cortex. With this method, we were able to delineate five major thalamocortical pathways in each gestational week. To keep regions of interest more consistent across the cohort, we grouped all cases into 2-weekly intervals, starting at 23 weeks (Figure 2), replicating methods used previously (Wilson et al., 2021).

Figure 1. Methods pipeline to estimate and quantify thalamocortical tracts development.

Figure 1.

Open in a new tab

(Top row) (a) Whole-brain connectomes generated for each gestational week template. (b) Atlas-defined masks of the thalamus and cortical areas were used to extract white matter pathways of interest from the connectomes. (c) These pathways were transformed to the native fetal diffusion space, (d) the values were sampled along the tract. (f) Whole-tract average diffusion metrics were calculated or (g) values sampled along the tract were aligned to an atlas of transient fetal compartments.

Figure 2. Tractography of thalamocortical pathways in different gestational week templates across the second to third trimester.

Figure 2.

Open in a new tab

Tracts project to five different cortical areas, the prefrontal cortex, primary motor cortex, primary sensory cortex, posterior parietal cortex, and primary visual cortex, coloured according to the anterior-posterior axis.

Structural connectivity parcellation of the fetal thalamus resembles adult topology of thalamic nuclei

Tract density imaging (Calamante et al., 2010) was used in each ODF template to explore whether the different cortical areas were connected to distinct, specific regions of the thalamus (Figure 3a). We found that for all ages, there was symmetrical topographical representation of the cortical regions of interest in the thalamus. Furthermore, they spatially corresponded to the adult organisation of thalamic nuclei, demonstrated by the schematic (Figure 3a) which is based on Morel’s thalamus and other connectivity-derived parcellations from adult imaging studies (Morel et al., 1997; Najdenovska et al., 2018; Niemann et al., 2000). The tract projecting to the prefrontal cortex was connected to the anterior thalamus and in the younger ages (23–29 weeks) also to the medial thalamus. In the older templates (31, 33, and 35 weeks), frontal connectivity was more localised to the anterior thalamus and less evident in the medial area. There were distinct but neighbouring areas in the ventral thalamus connecting to the sensory and motor cortical areas, the motor-connected thalamic region being more frontal. The connectivity of the posterior parietal area was in the posterior part of the thalamus, and the most posterior voxels in the thalamic mask projected to the primary visual cortex.

Figure 3. Tract density imaging parcellation of thalamus at different fetal ages.

Figure 3.

Open in a new tab

(a) A schematic of expected cortical connectivity arrangement across the thalamus, based on Morel’s parcellation of the adult thalamic nuclei. (b) Axial slices of thalamic parcellation, thresholded for the top 20% of voxels, colour-coded according to streamline connectivity of different tracts at 23 weeks, (c) 27 weeks, (d) 31 weeks, and (e) 35 weeks.

Whole-tract average diffusion metrics have a characteristic U-shaped trend across the second to third trimester

The thalamocortical pathways were transformed from the age-matched templates to the native subject space for 140 fetal subjects (Figure 4a). The MSMT-CSD-derived voxel-average tissue and fluid ODF values were sampled along the warped group-average streamline tracts. Tract-specific values were derived by averaging these for each tract in each subject, replicating the approach that has been used in previous fetal studies (Wilson et al., 2021). The values for each tract were plotted against the GA of the subject. The Akaike information criterion suggested second-order polynomial relationships for all tracts for both tissue and fluid fraction metrics, except the fluid fraction in the AT which is linear (Figure 4b). Diffusion tensor metrics also displayed similar age-related polynomial trends (Figure 4—figure supplement 1). We compared male vs. female (two-sample test) in each gestational week and found no significant differences (p>0.1).

Figure 4. Diffusion metric age trajectories for each tract.

(a) Distribution of age among the fetal cohort (n=140) in gestational weeks. (b) Whole-tract average tissue (top) and fluid fractions (bottom) for each subject in the left (orange) and right (blue) hemisphere, plotted against gestational age (GA) of the subject, best fit by second-order polynomials (AT = anterior thalamic radiation, OR = optic radiation, PP = posterior parietal tract, TS = thalamic-sensory tract, TM = thalamic-motor tract).

Figure 4.

Figure 4—figure supplement 1. Whole-tract average of diffusion tensor metrics in the fetal cohort (n=140), fractional anisotropy (FA), and mean diffusivity (MD), in thalamocortical tracts across gestational age (every other week shown).

Figure 4—figure supplement 1.

Open in a new tab

Along-tract sampling reveals evolving properties of fetal brain transient compartments

To explore the origins of these trends in diffusion metrics, the values of tissue and fluid fraction were sampled in subject space at 100 equidistant intervals between the thalamus and the cortex. Tissue and fluid fraction are scaled jointly per scan such that they are approximately reciprocal of one another across the brain using a cubic polynomial spatial model (Pietsch et al., 2019). In each subject, we sampled the tissue and fluid fraction values beneath the streamlines from the thalamus to the cortex, plotting the microstructural tissue composition against the distance from the thalamus (Figure 5). We found that trajectories changed gradually between gestational weeks, and therefore we grouped them to match previous histological studies that define this fetal period according to three developmental windows, early (21–25.5 weeks), mid (26–31.5 weeks), and late (32–36 weeks) prenatal period (Kostović, 2020; Appendix 1—figures 5 and 6). When comparing the microstructural profiles of all the tracts in the different periods, the motor, sensory, and parietal tracts shared similar trajectories, whilst those in the AT and OR tracts were more distinct (Figure 5a, b, and c and Figure 5—figure supplement 1a,b).

Figure 5. Microstructural composition of fetal compartments traversed by developing thalamic white matter.

Tracts were overlayed on the atlas of fetal compartments (examples highlight the difference between fetal brain structure in early prenatal [25 weeks] on far left, and late prenatal [35 weeks] on far right). Tissue fraction trends (top row) and fluid fraction trends (bottom row), normalised to 1, between the thalamus and cortex (thalamocortical tract axis) for the (a) thalamic-motor tract, (b) optic radiation, and (c) anterior thalamic radiation. Subjects were grouped by age, and average trajectories plotted for early prenatal (22–25.5 weeks), mid prenatal (26–31.5 weeks), late prenatal (32–36 weeks). Error bars represent the standard deviation among all subjects in each group. Atlas-derived tissue boundaries are marked on the trajectories to reveal the changing tissue properties of each layer between early, mid, and late prenatal development (cortical spinal fluid = CSF, cortical plate = CP, subplate = SP, intermediate zone = IZ, ventricular zone = VZ, deep grey matter = GM, immature white matter = WM).

Figure 5.

Figure 5—figure supplement 1. Microstructural composition of fetal compartments traversed by developing thalamic white matter.

Figure 5—figure supplement 1.

(a) Thalamic-sensory tract and (b) posterior parietal tract. Tracts were overlayed on the atlas of fetal compartments (examples highlight the difference between fetal brain structure in early prenatal [25 weeks] on far left, and late prenatal [35 weeks] on far right). Tissue fraction trends (top row) and fluid fraction trends (bottom row) normalised to 1, between the thalamus and cortex (thalamocortical tract axis). Subjects were grouped by age, and average trajectories plotted for early prenatal (22–25.5 weeks), mid prenatal (26–31.5 weeks), late prenatal (32–36 weeks). Error bars represent the standard deviation among all subjects in each group. Atlas-derived tissue boundaries are marked on the trajectories to reveal the changing tissue properties of each layer between early, mid, and late prenatal development (cortical spinal fluid = CSF, cortical plate = CP, subplate = SP, intermediate zone = IZ, ventricular zone = VZ, deep grey matter = GM, immature white matter = WM).
Open in a new tab

To improve our ability to corroborate changes in the dMRI signal with observations from histological studies, we mapped the maturational trajectories to an atlas of fetal brain compartments (Gholipour et al., 2017) and overlayed the boundaries of these compartments on the tissue and fluid fraction trajectories (Figure 5). By overlaying the tracts on the atlas, and corroborating each sampling segment with an atlas region, we could characterise the diffusion properties of the different fetal compartments independent of their change in size across gestation. Tissue fraction values in the deep grey matter and the cortical plate areas increased with GA in all tracts. This increase was most marked in the tracts terminating in superior areas of the brain (motor, sensory, and superior parietal cortex) (Figure 5a TM, Figure 5—figure supplement 1 (a) TS and (b) PP). The tissue fraction of the ventricular and intermediate zones decreased between the early and mid prenatal period, in all tracts. This decrease was very pronounced in the motor, sensory, and superior parietal tracts. The subplate tissue fraction changes were more tract specific. In the subplate of sensorimotor and parietal tracts, there was initially a very high fluid fraction and low tissue fraction, which transitions across the second to third trimester, increasing in tissue fraction from early to mid and then to late prenatal. Whereas in the AT, there was a decrease in subplate tissue fraction with GA (and a reciprocal increase in fluid fraction). In the OR, the subplate tissue fraction decreases between early and mid prenatal to then increase again in late prenatal. Highest tissue fractions were generally observed in the ventricular zone, with the lowest tissue fraction in the subplate area. To statistically test if there was a difference between the values in each compartment across GA, we correlated the tissue fraction values against age for each of the 100 sampling points. After correcting for multiple comparisons, points 1–16, 24–42, and 70–100 had significant linear correlations with age. Sampling points 43–69 had significant second-order polynomial relationships with age.

Discussion

In this work, we studied in utero development of five distinct thalamocortical pathways using state-of-the-art dMRI methods and bespoke pre-processing pipeline (Christiaens et al., 2019a; Cordero-Grande et al., 2019; Hutter et al., 2018a; Pietsch et al., 2019; Wilson et al., 2021) in 140 fetuses aged 21–37 weeks’ gestation. We show that these pathways connect to distinct thalamic nuclei, which could be clearly defined at group level even at 23 weeks. To disentangle the impact of different neurobiological processes on diffusion metrics, we characterised the tissue composition profile along each of the thalamocortical tracts as they traverse the different developmental tissue layers of the fetal brain. We used MSMT-CSD to model this fetal DWI dataset because it does not mandate a specific set of b-values. The technique exploits the unique b-value dependencies of different tissue types, and so depends inherently on the characteristics of the tissue in the fetal brain. The distinct properties of different fetal compartments after the application of MSMT-CSD are highlighted by the ODFs (Appendix 1—figure 4). Readers are directed to Tournier et al., 2019, for a more comprehensive review of the optimisation process involved to select appropriate b-shell values for the acquisition.

We found that the spatiotemporal changes in the diffusion signal reflected known developmental processes that take place between the early, mid, and late prenatal period. The early period is characterised by higher tissue fractions in the middle of the tract, where there is a radial scaffold for migrating neurons. As this scaffold dissipates in the mid prenatal period, this is accompanied by a reduction in the tissue fraction in the middle of the tract, and an increase in tissue fraction towards the termination of the tracts as the neurons of the cortical plate mature. Finally in the late prenatal period, we observe the highest tissue fraction values at the start and end of the axis, as the pre-myelination phase of white matter development commences. This study demonstrates how the dMRI signal can be modelled to create in vivo spatiotemporal trajectories which relate to underlying neurobiological properties and are consistent with described trends from post-mortem histology (Kostović, 2020).

Early embryonic patterning of gene expression and cell division in the thalamus provide a template for specialised nuclei to emerge over the course of development, such that specific cells eventually occupy distinct locations within the thalamus (Clascá et al., 2012; Nakagawa, 2019, Kostovic and Goldman-Rakic, 1983). Rodent studies labelling the embryonic thalamus demonstrated that there is a characteristic topography of thalamic projections which roughly exists at their time of arrival; anterior to posterior movement along the convexity of the cortex is represented in a medial-to-lateral axis within the thalamus, while ventral-dorsal movement across the cortex is represented in an anterior-to-posterior axis within the thalamus (Molnár et al., 1998; Molnár et al., 2012). Thalamocortical tracts emerge over the same timescale as the thalamus parcellates and matures into its specialised group of nuclei (Clascá et al., 2012). Although the topography of thalamic nuclei and their cortical connectivity is acquired embryonically, no in vivo parcellation of the thalamus in the fetal brain has been published. Using tract density imaging, we observed that the cortical areas were connected to specific thalamic regions, organised in an anterior-posterior axis. This anterior-posterior representation of cortical connectivity in the thalamus was consistent across the second to third trimester and is in accordance with the topology of thalamic nuclei described in rodent studies and histology (Molnár and Blakemore, 1995; Molnár et al., 1998).

In addition, our fetal structural connectivity parcellation resembles the functionally derived thalamic parcellation in neonates, supporting the view that there is a strong association between structure and function in thalamocortical circuitry that begins early in life (Johansen-Berg et al., 2005; Toulmin et al., 2015; Alcauter et al., 2014). However, it is worth noting that this thalamic parcellation is dependent on streamline count through a voxel, and in the fetal brain streamlines are prone to spurious detection. This limitation is particularly relevant in the youngest fetuses, where we observe an extremely dense connectome (due to a fixed number of streamlines in a smaller brain) but there are very few coherent axonal bundles, so tracts might be overrepresented in the thalamic parcellation. The topography of thalamic nuclei is also not static during the embryonic and fetal period (Le Gros Clark, 1936), as pulvinar size increases and the dorsal lateral geniculate nucleus shifts its position from dorsolateral to ventromedial (Rakic, 1977; Mojsilović and Zecević, 1991). However, the image resolution of this study and the timespan in development over which this data was collected limit us from visualising these differences across age.

Recent studies characterising developing white matter pathways using human fetal MRI identified second-order polynomial maturational trends in diffusion metrics unique to this developmental period (Wilson et al., 2021; Machado-Rivas et al., 2021). Here, we replicated these methods with a different group of tracts and found the same U-shaped trends in thalamocortical white matter development. The inflection point at around 29–30 weeks was hypothesised to be the result of the dissipating radial glial scaffold followed by the pre-myelination phase of white matter development (Wilson et al., 2021; Machado-Rivas et al., 2021). The sensitivity of HARDI to radially organised structure in the fetal brain has been described by previous studies (Miyazaki et al., 2016; Takahashi et al., 2012; Xu et al., 2014) combining it with post-mortem tissue analysis to show that radially coherent diffusion signal corresponded to radial glial fibres in the early prenatal period, transitioning to cortico-cortical fibres around 30 weeks, coinciding with the appearance of astrocytes (Takahashi et al., 2012; Xu et al., 2014). However, with whole-tract average values, it is not possible to establish the precise effect of different neurodevelopmental processes on diffusion metrics across gestation.

To address this ambiguity, we characterised the entire trajectory of tissue composition changes between the thalamus and the cortex. We found that age-related changes in the tissue and fluid fraction along the tracts concurred with histological observations (Kostović and Judas, 2010). During the early prenatal period (22–25.5 GW), neuronal precursors migrate along the radial glial scaffold from proliferative zones to their destination in the cortical plate and thalamocortical axons accumulate in the superficial subplate, entering a ‘waiting phase’, forming transient synaptic connections (Ghosh et al., 1990; Kostovic and Rakic, 1984; Kostovic and Rakic, 1990). In terms of the diffusion signal, this strongly aligned microstructure of the radial glia is represented in our results by a higher tissue fraction in the transient compartments containing the most migratory cells (such as the VZ, IZ) (Kostović and Judas, 2010). Conversely, we observe the lowest tissue fraction in the early prenatal SP, which predominantly contains hydrophilic extracellular matrix, as demonstrated by rodent and non-human primate studies (Allendoerfer and Shatz, 1994; Miller et al., 2014; Molnár and Hoerder-Suabedissen, 2016).

By the mid prenatal period (26–31.5 weeks), we observe increased tissue fraction in the cortical plate, coinciding with the innervation of the cortical plate by thalamocortical axons, increasing soma volume and dendritic branching of CP neurons and CP synaptogenesis (Huttenlocher, 1979; Mrzljak et al., 1992; Huttenlocher and Dabholkar, 1997). We also observe increased tissue fraction in the SP zone in the mid prenatal period, consistent with histological observations of increased coherence of axonal fibres between cortical areas (Takahashi et al., 2012; Xu et al., 2014). The tissue fraction in the VZ and IZ decreases compared to the early prenatal period, corresponding to the timeframe when the radial glial scaffold dissipates (Kinney et al., 1988; Back et al., 2001; Haynes et al., 2005).

From the mid to late prenatal period, there is a marked increase in tissue fraction in last third of the axis between thalamus and cortex. By this point in development, the radial glia have converted into oligodendrocyte precursor cells which ensheath the axonal fibres to commence pre-myelination, enhancing the structural integrity of the fibre pathways (Back et al., 2001; Back et al., 2002; Haynes et al., 2005; Kinney et al., 1988; Kinney et al., 1994). A previous study in perinatal rabbits has shown that this oligodendrocyte lineage progression correlates with diffusion metrics (Drobyshevsky et al., 2005), suggesting it is likely to contribute to the increased tissue fraction we observe in the late prenatal period. The tissue fraction increase in the CP area is consistent in time with the lamination of the CP, the elaboration of thalamocortical terminals in layer IV, and a rapid growth of basal dendrites of layer III and V pyramidal neurons (Kostović and Jovanov-Milosević, 2006; Krsnik et al., 2017; Molliver et al., 1973). These high tissue fraction values at the origin and termination of the tracts suggest co-maturation between ascending and descending pathways between the thalamus and cortex to eventually form continuous, structurally mature fibre bundles. This concept was proposed in the 1990s by Blakemore and Molnar, termed the ‘handshake hypothesis’. They suggested that thalamocortical pathways ascending through the internal capsule project to their cortical targets with assistance from reciprocal descending cortical pathways (Molnár and Blakemore, 1995). We hypothesise that continuing this analysis over subsequent weeks into the neonatal period would lead to an increasing tissue fraction in the middle of the axis, as fibre bundles become more uniformly structurally mature and the subplate completely resolves (Kinney et al., 1988; Haynes et al., 2005; Kostović and Jovanov-Milosević, 2006).

We observed that tracts terminating superiorly (motor, sensory, and parietal) shared very similar trajectories in the early, mid, and late periods. However, the OR and the AT had more distinct trajectories. The microstructural change along the AT suggests increasing tissue fraction between the deep grey matter, VZ and IZ. We hypothesise that the high tissue fraction in the IZ is due to densely packed ascending and descending bundles within the anterior limb of the internal capsule (Emos et al., 2022). The precise timeline for outgrowth and intermingling of the thalamocortical and corticothalamic projections remains ambiguous, and appears to be different between rodents and non-human primates (Alzu’bi et al., 2019).

On the other hand, the OR traverses the deep parietal lobe along the border of the lateral ventricle and has smoother transitions in tissue fraction between the fetal compartments. This is likely due to the tract area running more parallel to the tissue interfaces. Another explanation for the regional differences in microstructural properties is the variation in subplate remnants. In the late prenatal trajectories, all tracts except the OR have a large dip in tissue fraction along the tract. In the primary visual cortex, the subplate disappears during the final weeks of gestation, whereas in the somatosensory cortex there are still subplate neurons present in term-born neonates (Kostovic and Rakic, 1990) and the subplate of the pre-frontal associative cortex gradually disappears over the 6 postnatal months. Therefore, the peaks of fluid fraction in the frontal and sensory trajectories might reflect the lasting presence of subplate in these areas (Kostović and Jovanov-Milosević, 2006; Kostovic and Rakic, 1990).

There are several important considerations to our work which may limit the interpretation of our findings. MRI has an inherently low signal:noise ratio as signal attenuation is deliberately introduced and EPI acquisitions are highly susceptible to motion and distortion artefacts. Furthermore, the biophysical properties of the fetal environment introduce a series of additional inter-dependent challenges for acquiring high-quality dMRI data (Christiaens et al., 2019a; Christiaens et al., 2019b, Tournier et al., 2020). In utero imaging methods must adapt and account for unpredictable, sometimes abrupt motion of fetus, in addition to the motion introduced by maternal breathing. Achieving even coverage of the magnetic coil is challenging due to varying fetal and maternal positions inside the scanner, introducing unique bias fields for every subject. The dHCP acquisition protocol and pre-processing pipelines have incorporated dynamic distortion and motion correction algorithms that were designed to specifically tackle the unique artefacts-associated imaging fetuses (Deprez et al., 2020; Cordero-Grande et al., 2018; Cordero-Grande et al., 2019; Christiaens et al., 2019a; Christiaens et al., 2019b; Hutter et al., 2018b; Price et al., 2019). However, even after the extensive correction procedures, there is still some residual motion and distortion in the data. For a more extensive and comprehensive review of challenges associated specifically with in utero diffusion imaging and constructive suggestions to address them, readers are directed to Christiaens et al., 2019b.

The conclusions that can be drawn from analysing the diffusion MR signal are limited more generally by the level of noise and lack of directionality in the signal, and therefore we are not able to distinguish between ascending and descending thalamic projections as they develop. The fibre-tracking process itself is also very sensitive to the residual motion and distortion in fetal data, producing highly variable results in terms of streamline count per tract. Therefore, the metrics generally used in this field to investigate fibre bundle morphology, such as streamline count, a proxy for ‘innervation density’, are not an accurate, quantitative measure (Calamante, 2019) and cannot be reliably used to assess fibre morphology. Consequently, we have just used tractography to delineate regions of interest and used the signal contrast in the diffusion maps to quantify microstructure.

The methods described allow the direct study of the maturational effects of the subplate and intermediate zones, which are known to represent critical substrates for early synaptogenesis and the spatial guidance of thalamocortical axons (Ghosh et al., 1990). Damage to this essential structural framework for developing cortical circuitry has been implicated in the origins of numerous developmental disorders and is suspected to underly the altered structural and functional connectivity of the thalamus in preterm infants (Back and Volpe, 1997; Volpe, 2001; Volpe, 2009; Hüppi et al., 2001; Counsell et al., 2003; Hadders-Algra et al., 2017; Mathur and Inder, 2009; Toulmin et al., 2015; Ball et al., 2012; Ball et al., 2015). It is therefore critical to use clinically relevant tools, such as in utero MRI, to relate the microstructural properties of these transient fetal compartments to neurobiological processes. This improves mechanistic insight about both healthy white matter maturation and the developmental origins of white matter pathologies.

With this study we explore the development of thalamocortical white matter by quantifying microstructure in the different layers of the fetal brain. Using diffusion metrics, we characterise the emergence of structural connectivity from the thalamus to spatially and functionally distinct cortical brain regions. We observe correlations between the transitioning tissue components and key neurobiological processes in white matter development. By providing a detailed normative reference of MR signal change during the second to third trimester, this will help future studies to identify if the tissue properties of specific compartments are affected by preterm birth or other perinatal injury. To this effect, all fetal MRI data is made available to the research community.

Materials and methods

Sample

The study was approved by the UK Health Research Authority (Research Ethics Committee reference number: 14/LO/1169) and written parental consent was obtained in every case for imaging and open data release of the anonymised data. All data was acquired in St Thomas Hospital, London, United Kingdom. The sociodemographic characteristics of this sample are representative of the diversity in the London population (see Appendix 1—figure 1).

Acquisition, pre-processing, and quality control

GA was determined by sonography at 12 post-ovulatory weeks as part of routine clinical care. Three-hundred fetal MRI datasets were acquired with a Philips Achieva 3T system, with a 32-channel cardiac coil in maternal supine position. dMRI data was collected with a combined spin echo and field echo (SAFE) sequence (Hutter et al., 2018a, Cordero-Grande et al., 2018) at 2 mm isotropic resolution, using a multi-shell diffusion encoding that consists of 15 volumes at b=0 s/mm2, 46 volumes at b=400 s/mm2, and 80 volumes at b=1000 s/mm2 lasting 14 min (Christiaens et al., 2019a). For more insight on the choice of b-shell values for this acquisition, readers are directed to Appendix 1—figure 4 (Tournier et al., 2019). The protocol also included the collection of structural T2w, T1w, and fMRI data, for a total imaging time of approximately 45 min (Price et al., 2019).

dMRI data were processed using a bespoke pipeline (Christiaens et al., 2019a) that includes generalised singular value shrinkage image denoising and debiasing from complex data (Cordero-Grande et al., 2019), dynamic distortion correction of susceptibility-induced B0 field changes using the SAFE information (Ghiglia and Romero, 1994; Cordero-Grande et al., 2018; Hutter et al., 2018b) and slice-to-volume motion correction based on a multi-shell spherical harmonics and radial decomposition (SHARD) representation (Christiaens et al., 2021). Quality control (QC) was implemented using summary metrics based on the gradient of the motion parameters over time and the percentage of slice dropouts in the data (Christiaens et al., 2021). This was followed up with expert visual assessment, which considered any residual or uncorrected artefacts. Image sharpness, residual distortion, and motion artefacts were visually assessed and scored between 0 and 3, with 0 (=failure, e.g. because the subject moved out of the field of view) to 3 (=high quality), based on the mean b=0, b=400, and b=1000 images and the ODFs estimated with MSMT-CSD. See Appendix 1—figure 2 for examples of subjects that were excluded. Both DWI and T2 for each fetus were required to facilitate co-registration to template space via a structural intermediate. After co-registration, the QC scores were checked and validated again by different authors (MP, AU, SW), only subjects scoring 2 or 3 that were well aligned in T2 space were admitted to this study. Based on the above criteria, 140 of the 300 subjects that were pre-processed were classified as high-quality reconstructions for both DWI and T2 modalities. Post hoc analysis was also conducted to check if any motion-related parameters correlate with GA (Appendix 1—figure 3).

Diffusion modelling and template generation

All diffusion processing and tractography was done using MRtrix3 (Tournier et al., 2019). To deconvolve the tissue and fluid components of the diffusion data, white matter and cortical spinal fluid (CSF) response functions were estimated for each subject using T2-based tissue segmentations as inclusion areas. White matter response functions were extracted from areas of relatively mature white matter (corticospinal tract and corpus callosum) using the ‘tournier’ algorithm and CSF responses using the ‘dhollander’ algorithm in MRtrix3 (Jeurissen et al., 2014; Tournier et al., 2019; Tournier et al., 2013). The white matter response functions of the oldest 20 subjects were averaged to obtain a group-average response function of relatively mature white matter, whilst a group-average CSF response function was calculated from the whole cohort of subjects. dMRI signal of all subjects was subsequently deconvolved into tissue ODF and fluid components using MSMT-CSD and the group-average white matter and CSF response functions (Jeurissen et al., 2014), and resulting components were intensity normalised for each subject (Raffelt et al., 2011). Subject ODFs warped into weekly templates through a series of coarse pose normalisation and non-linear diffeomorphic image registration steps (Jenkinson et al., 2002; Raffelt et al., 2011; Pietsch, 2018). These transformations were composed to obtain pairs of inverse consistent diffeomorphic subject-to-template and template-to subject warps.

Connectome generation and tractography

The ODF templates were co-registered to the Boston T2-fetal atlas (Gholipour et al., 2017) using non-linear registration (Avants et al., 2008). The tissue segmentations of the cortex, white matter, and deep grey matter were used for anatomically constrained tractography to generate whole-brain structural connectomes of 100 M streamlines in each gestational week (Smith et al., 2012; Tournier et al., 2019). The connectomes were filtered down to 10 M streamlines using the SIFT algorithm (Smith et al., 2013; Tournier et al., 2019), so that the number of streamlines connecting the two regions are approximately proportional to the cross-sectional area of the fibres connecting them (Smith et al., 2013). In each weekly template, thalamocortical pathways of interest were defined in both hemispheres by filtering the connectome using seed regions derived from the Boston T2-fetal atlas (Gholipour et al., 2017), including the thalamus, primary motor cortex, primary sensory cortex, posterior parietal cortex, dorso-lateral prefrontal cortex, and primary visual cortex. We also used additional ROIs to exclude spurious streamlines that were projecting away from the expected path of the tract (e.g. to exclude callosal fibres from the TM).

Tract density parcellation of thalamus

Tckmap was used to identify which voxels in the thalamus mask were traversed by the streamlines of each tract (Calamante et al., 2010). The tract density maps were merged using FSL (Jenkinson et al., 2012) and a colour-coded parcellation volume was constructed reflecting the maximum density tract for each voxel. For visualisation, the tract density maps for each tract were thresholded at 80%, only to include voxels with the highest streamline connectivity.

Extracting tissue and fluid fraction values

To extract diffusion metrics for analysis, tracts were transformed from the templates to age-matched subject space to be overlaid onto the normalised fluid ODF and the normalised tissue ODF. The mean value within the segmented tracts was calculated to give the tissue and fluid fractions.

Microstructural profiling

In each template, thalamocortical tracts were filtered so all the streamlines for each tract were of the same length, to ensure even sampling intervals along them. All template tracts were then registered into a standard space and resampled to 100 points, before being transformed to individual subjects and overlaid on the normalised tissue and fluid fraction maps. The average value for each sampling point was calculated to create a microstructural profile along the path between the thalamus and the cortical plate. To provide a reference for microstructural differences between fetal brain compartments, tracts were overlaid on the atlas-derived tissue parcellations. The value of the tissue labels underlying the tract were used to establish which sampling points corresponded to each fetal compartment. These boundaries between compartments were then used to label the plots in Figure 5.

Acknowledgements

We thank the patients who agreed to participate in this work and the staff of St Thomas’ Hospital London. This work was supported by the European Research Council under the European Union Seventh Framework Programme (FP/2007–2013)/ERC Grant Agreement No. 319456. We acknowledge infrastructure support from the National Institute for Health Research (NIHR) Mental Health Biomedical Research Centre (BRC) at South London and Maudsley NHS Foundation Trust, King’s College London, and the NIHR-BRC at Guy’s and St Thomas’ NHS Foundation Trust. We also acknowledge grant support in part from the Wellcome Engineering and Physical Sciences Research Council (EPSRC) Centre for Medical Engineering at King’s College London (WT 203148/Z/16/Z) and the Medical Research Council (UK) (MR/K006355/1) and (MR/L011530/1). SW was supported by PhD funding from the UK Medical Research Council/Sackler Foundation (MR/P502108/1). JO is supported by a Sir Henry Dale Fellowship jointly funded by the Wellcome Trust and the Royal Society (206675/Z/17/Z). JO, MR, ADE, JO, and TA received support from the Medical Research Council Centre for Neurodevelopmental Disorders, King’s College London (MR/N026063/1). TA was supported by an MRC Clinician Scientist Fellowship (MR/P008712/1). VK and TA were supported by an MRC Transition Support Award (MR/V036874/1). Support for this work was also provided by the NIHR-BRC at Kings College London, Guy’s and St Thomas’ NHS Foundation Trust in partnership with King’s College London, and King’s College Hospital NHS Foundation Trust.

Appendix 1

Extended details about the methodological decision-making process

Sociodemographic information

One of the aims of the dHCP was to ensure that the cohort analysed were representative of the diverse sociodemographic spread within the London population. We compared our cohort with information derived from the Mayor of London database (https://data.london.gov.uk/dataset/indices-of-deprivation) to check if the distributions were similar (Appendix 1—figure 1). We find that socioeconomic status, according to the index of multiple deprivation, and the ethnicity is comparable between our cohort and London. This suggests that our sample is representative, and our results should be generalisable across the London population. To ensure that we were not observing the effect of a preterm pregnancy, we also provide information on the GA at birth of the fetal cohort and the baby birth weight.

Appendix 1—figure 1. Sociodemographic and neonatal follow-up information.

Appendix 1—figure 1.

Open in a new tab

Including index of multiple deprivation (IMD), London. Gestational age at birth, birth weight, and ethnicity.

Quality control

QC was implemented using summary metrics based on the gradient of the motion parameters over time and the percentage of slice dropouts in the data (Christiaens et al., 2021). This was followed up with expert visual assessment, which considered any residual or uncorrected artefacts. Image sharpness, residual distortion, and motion artefacts were visually assessed and scored between 0 and 3, based on the mean b=0, b=400, and b=1000 images. The scoring system was as follows: 0=failure, brain outside field of view, 1=banding and large distortion, 2=blurring but no major artefacts, 3=relatively high quality.

Appendix 1—figure 2. Examples of QC scoring scale for subjects, between 0 and 3, with 0 (=failure, e.g. subject moved out of the field of view) to 3 (=high quality).

Appendix 1—figure 2.

Open in a new tab

As an additional check to see if the extent of motion affected our results, we correlated the indices derived from the SHARD pipeline with GA. There are three output parameters per subject, which are signal-to-noise ratio (SNR), rotation, and translation. SNR = mean b=0 signal/mean noise level, measured in the reconstruction mask using MP-PCA denoising. Rotation = mean change in rotation of the subject pose between slices (degrees). Translation = mean change in translation of the subject pose between slices (mm).

Appendix 1—figure 3. Correlations between spherical harmonics and radial decomposition (SHARD)-derived indices describing subject motion and gestational age (signal-to-noise ratio [SNR], rotation and translation).

Appendix 1—figure 3.

Open in a new tab

Diffusion data modelling

Previous publications provide detail on the optimisation process used to define the b-shell values used (Tournier et al., 2019 and Christiaens et al., 2019b; Christiaens et al., 2019a). This work demonstrates that there is a marked loss of signal at higher b-shells (>1000 s/mm3) in the neonatal brain compared to adults, due to the tissue being inherently much less restricted. There is also increased noise at higher b-shells. Both issues are amplified in the fetal brain, there are long relaxation times in the water-rich, relatively unrestricted fetal brain tissues and the amount of noise is higher.

To demonstrate the suitability of MSMT-CSD for this dataset, an important consideration is that the MSMT-CSD approach doesn’t mandate a specific set of b-values which are required for the modelling process, only that there are multiple shells in addition to b=0. The technique depends on the signals related to different tissue types having sufficiently distinct behaviours across the b-values sampled, and so depends inherently on the characteristics of the tissue. To demonstrate the distinct properties of the different compartments in the fetal brain, we have now included a figure in the Appendix (Appendix 1—figure 4) showing the signal amplitude per tissue type as a function of b-value, and the ODFs after MSMT-CSD has been applied.

Appendix 1—figure 4. Signal amplitude decay in tissue vs. fluid response functions at different b-shells (left).

Appendix 1—figure 4.

Open in a new tab

Polar plots of the distinct orientation density functions (ODFs) between the corpus callosum (white matter, tissue compartment) and ventricle (cortical spinal fluid [CSF], fluid compartment) after multi-shell multi-tissue constrained spherical deconvolution (MSMT-CSD) is applied to the data (right).

Along-tract changes for specific gestational weeks

Prior to grouping the data as seen in Figure 5. The analysis was conducted for individual subjects, then averaged for each gestational week. We present these trajectories (Appendix 1—figures 5 and 6) to demonstrate how there are gradual changes in the diffusion signal, on a weekly basis, that justify our decision (for easier interpretation) to group the data into early, mid, and late prenatal groups.

Appendix 1—figure 5. Tissue fraction trajectories along thalamocortical tract axis in each gestational week, prior to grouping into early, mid, and late prenatal (every other week shown).

Appendix 1—figure 5.

Open in a new tab
Appendix 1—figure 6. Fluid fraction trajectories along thalamocortical tract axis in each gestational week, prior to grouping into early, mid, and late prenatal (every other week shown).

Appendix 1—figure 6.

Open in a new tab

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. For the purpose of Open Access, the authors have applied a CC BY public copyright license to any Author Accepted Manuscript version arising from this submission.

Contributor Information

Jonathan O’Muircheartaigh, Email: jonathanom@kcl.ac.uk.

Tomoki Arichi, Email: tomoki.arichi@kcl.ac.uk.

Finnegan J Calabro, University of Pittsburgh, United States.

Jonathan Roiser, University College London, United Kingdom.

Funding Information

This paper was supported by the following grants:

  • European Research Council Seventh Framework Programme: FP/2007/2013 to Maximilian Pietsch, Lucilio Cordero-Grande, Daan Christiaens, Vyacheslav R Karolis, Vanessa Kyriakopoulou, Anthony N Price, Jana Hutter, Emer J Hughes, Jacques-Donald Tournier, Joseph V Hajnal, A David Edwards.

  • Wellcome Trust Sir Henry Dale Fellowship: 206675/Z/17/Z to Jonathan O’Muircheartaigh.

  • Medical Research Council Centre for Neurodevelopmental Disorders MR/N0266063/1 to Siân Wilson, Jonathan O’Muircheartaigh, Mary A Rutherford, A David Edwards, Tomoki Arichi.

  • Medical Research Council Translation support fellowship: MR/V036874/1 to Vyacheslav R Karolis, Tomoki Arichi.

  • Wellcome / EPSRC Centre for Biomedical Engineering, Kings College London WT 203148/Z/16/Z to Anthony N Price, Jana Hutter, Jacques-Donald Tournier, Joseph V Hajnal.

  • Medical Research Council Clinician Scientist Fellowship MR/P008712/1 to Tomoki Arichi.

  • Medical Research Council MR/K006355/1 to Siân Wilson, Maximilian Pietsch, Lucilio Cordero-Grande, Daan Christiaens, Alena Uus, Vyacheslav R Karolis, Vanessa Kyriakopoulou, Kathleen Colford, Anthony N Price, Jana Hutter, Mary A Rutherford, Emer J Hughes, Serena J Counsell, Jacques-Donald Tournier, Joseph V Hajnal, A David Edwards, Jonathan O’Muircheartaigh, Tomoki Arichi.

  • Medical Research Council MR/L011530/1 to Serena J Counsell.

Additional information

Competing interests

No competing interests declared.

No competing interests declared.

Author contributions

Conceptualization, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft, Writing – review and editing.

Data curation, Software, Formal analysis, Methodology, Writing – review and editing.

Data curation, Software, Methodology, Writing – review and editing.

Data curation, Formal analysis, Methodology, Writing – review and editing.

Data curation, Methodology.

Formal analysis, Investigation, Methodology, Writing – review and editing.

Data curation.

Data curation.

Data curation.

Data curation, Writing – review and editing.

Data curation, Funding acquisition.

Data curation.

Data curation, Investigation, Writing – review and editing.

Data curation, Software, Formal analysis.

Resources, Data curation, Funding acquisition, Methodology, Project administration.

Supervision, Funding acquisition, Investigation, Methodology, Project administration, Writing – review and editing.

Conceptualization, Resources, Data curation, Supervision, Funding acquisition, Validation, Investigation, Methodology, Project administration, Writing – review and editing.

Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Methodology, Project administration, Writing – review and editing.

Ethics

Human subjects: The study was approved by the UK Health Research Authority (Research Ethics Committee reference number: 14/LO/1169) and written parental consent was obtained in every case for imaging and open data release of the anonymized data.

Additional files

MDAR checklist

Data availability

Developing Human Connectome project data is open-access and available for download following completion of a data-usage agreement via: http://www.developingconnectome.org/. Data is also available at: https://nda.nih.gov/edit_collection.html?id=3955.

The following dataset was generated:

Uus A, Kyriakopoulou V, Cordero Grande L, Pietsch M, Price A, Wilson S, Patkee P, Karolis V, Schuch A, Gartner A, Williams L, Hughes E, Arichi T, O'Muircheartaigh J, Hutter J, Robinson E, Tournier JD, Rueckert D, Counsell SJ, Rutherford MA, Deprez M, Hajnal JV, Edwards AD. 2023. Multi-channel spatio-temporal MRI atlas of normal fetal brain development (dHCP project) G-Node Open Data.

The following previously published dataset was used:

Gholipour A, Rollins CK, Velasco-Annis C, Ouaalam A, Akhondi-Asl A, Afacan O, Ortinau C, Clancy S, Limperopoulos C, Yang E, Estroff JA, Warfield SK. 2017. A normative spatiotemporal MRI atlas of the fetal brain for automatic segmentation and analysis of early brain growth. Harcard CRL/fetal brain atlas. fetal_brain_atlas/

References

  1. Alcauter S, Lin W, Smith JK, Short SJ, Goldman BD, Reznick JS, Gilmore JH, Gao W. Development of thalamocortical connectivity during infancy and its cognitive correlations. The Journal of Neuroscience. 2014;34:9067–9075. doi: 10.1523/JNEUROSCI.0796-14.2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Allendoerfer KL, Shatz CJ. The subplate, a transient neocortical structure: Its role in the development of connections between thalamus and cortex. Annual Review of Neuroscience. 1994;17:185–218. doi: 10.1146/annurev.ne.17.030194.001153. [DOI] [PubMed] [Google Scholar]
  3. Alzu’bi A, Homman-Ludiye J, Bourne JA, Clowry GJ. Corrigendum: Thalamocortical afferents innervate the cortical subplate much earlier in development in primate than in rodent. Cerebral Cortex. 2019;29:5316. doi: 10.1093/cercor/bhz056. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Anticevic A, Cole MW, Repovs G, Murray JD, Brumbaugh MS, Winkler AM, Savic A, Krystal JH, Pearlson GD, Glahn DC. Characterizing thalamo-cortical disturbances in schizophrenia and bipolar illness. Cerebral Cortex. 2014;24:3116–3130. doi: 10.1093/cercor/bht165. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. 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]
  6. Back SA, Volpe JJ. Cellular and molecular pathogenesis of periventricular white matter injury. Mental Retardation and Developmental Disabilities Research Reviews. 1997;3:96–107. doi: 10.1002/(SICI)1098-2779(1997)3:1<96::AID-MRDD12>3.0.CO;2-M. [DOI] [Google Scholar]
  7. Back SA, Luo NL, Borenstein NS, Levine JM, Volpe JJ, Kinney HC. Late oligodendrocyte progenitors coincide with the developmental window of vulnerability for human perinatal white matter injury. The Journal of Neuroscience. 2001;21:1302–1312. doi: 10.1523/JNEUROSCI.21-04-01302.2001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Back SA, Luo NL, Borenstein NS, Volpe JJ, Kinney HC. Arrested oligodendrocyte lineage progression during human cerebral white matter development: Dissociation between the timing of progenitor differentiation and myelinogenesis. Journal of Neuropathology and Experimental Neurology. 2002;61:197–211. doi: 10.1093/jnen/61.2.197. [DOI] [PubMed] [Google Scholar]
  9. Ball G, Boardman JP, Rueckert D, Aljabar P, Arichi T, Merchant N, Gousias IS, Edwards AD, Counsell SJ. The effect of preterm birth on thalamic and cortical development. Cerebral Cortex. 2012;22:1016–1024. doi: 10.1093/cercor/bhr176. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Ball G, Boardman JP, Aljabar P, Pandit A, Arichi T, Merchant N, Rueckert D, Edwards AD, Counsell SJ. The influence of preterm birth on the developing thalamocortical connectome. Cortex; a Journal Devoted to the Study of the Nervous System and Behavior. 2013;49:1711–1721. doi: 10.1016/j.cortex.2012.07.006. [DOI] [PubMed] [Google Scholar]
  11. Ball G, Pazderova L, Chew A, Tusor N, Merchant N, Arichi T, Allsop JM, Cowan FM, Edwards AD, Counsell SJ. Thalamocortical connectivity predicts cognition in children born preterm. Cerebral Cortex. 2015;25:4310–4318. doi: 10.1093/cercor/bhu331. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Batalle D, Hughes EJ, Zhang H, Tournier JD, Tusor N, Aljabar P, Wali L, Alexander DC, Hajnal JV, Nosarti C, Edwards AD, Counsell SJ. Early development of structural networks and the impact of prematurity on brain connectivity. NeuroImage. 2017;149:379–392. doi: 10.1016/j.neuroimage.2017.01.065. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Behrens TEJ, Johansen-Berg H, Woolrich MW, Smith SM, Wheeler-Kingshott CAM, Boulby PA, Barker GJ, Sillery EL, Sheehan K, Ciccarelli O, Thompson AJ, Brady JM, Matthews PM. Non-Invasive mapping of connections between human thalamus and cortex using diffusion imaging. Nature Neuroscience. 2003;6:750–757. doi: 10.1038/nn1075. [DOI] [PubMed] [Google Scholar]
  14. Brody BA, Kinney HC, Kloman AS, Gilles FH. Sequence of central nervous system myelination in human infancy. I. An autopsy study of myelination. Journal of Neuropathology and Experimental Neurology. 1987;46:283–301. doi: 10.1097/00005072-198705000-00005. [DOI] [PubMed] [Google Scholar]
  15. Bui T, Daire JL, Chalard F, Zaccaria I, Alberti C, Elmaleh M, Garel C, Luton D, Blanc N, Sebag G. Microstructural development of human brain assessed in utero by diffusion tensor imaging. Pediatric Radiology. 2006;36:1133–1140. doi: 10.1007/s00247-006-0266-3. [DOI] [PubMed] [Google Scholar]
  16. Calamante F, Tournier JD, Jackson GD, Connelly A. Track-density imaging (TDI): Super-resolution white matter imaging using whole-brain track-density mapping. NeuroImage. 2010;53:1233–1243. doi: 10.1016/j.neuroimage.2010.07.024. [DOI] [PubMed] [Google Scholar]
  17. Calamante F. The seven deadly SINS of measuring brain structural connectivity using diffusion MRI streamlines fibre-tracking. Diagnostics. 2019;9:115. doi: 10.3390/diagnostics9030115. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Christiaens D, Cordero-Grande L, Price AN, Hutter J, Hughes E, Counsell SJ, Tournier JD, Hajnal JV. Fetal diffusion MRI acquisition and analysis in the developing Human Connectome Project. ISMRM Annual Meeting & Exhibition.2019a. [Google Scholar]
  19. Christiaens D, Slator PJ, Cordero-Grande L, Price AN, Deprez M, Alexander DC, Rutherford M, Hajnal JV, Hutter J. In utero diffusion MRI: Challenges, advances, and applications. Topics in Magnetic Resonance Imaging. 2019b;28:255–264. doi: 10.1097/RMR.0000000000000211. [DOI] [PubMed] [Google Scholar]
  20. Christiaens D, Cordero-Grande L, Pietsch M, Hutter J, Price AN, Hughes EJ, Vecchiato K, Deprez M, Edwards AD, Hajnal JV, Tournier JD. Scattered slice SHARD reconstruction for motion correction in multi-shell diffusion MRI. NeuroImage. 2021;225:117437. doi: 10.1016/j.neuroimage.2020.117437. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Clascá F, Rubio-Garrido P, Jabaudon D. Unveiling the diversity of thalamocortical neuron subtypes. European Journal of Neuroscience. 2012;35:1524–1532. doi: 10.1111/j.1460-9568.2012.08033.x. [DOI] [PubMed] [Google Scholar]
  22. Cordero-Grande L, Price AN, Ferrazzi G, Hutter J, Christiaens D, Hughes E, Hajnal JV. Spin And Field Echo (SAFE) dynamic field correction in 3T fetal EPI. Joint Annual Meeting ISMRM-ESMRMB; Paris, France. 2018. [Google Scholar]
  23. Cordero-Grande L, Christiaens D, Hutter J, Price AN, Hajnal JV. Complex diffusion-weighted image estimation via matrix recovery under general noise models. NeuroImage. 2019;200:391–404. doi: 10.1016/j.neuroimage.2019.06.039. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Counsell SJ, Allsop JM, Harrison MC, Larkman DJ, Kennea NL, Kapellou O, Cowan FM, Hajnal JV, Edwards AD, Rutherford MA. Diffusion-Weighted imaging of the brain in preterm infants with focal and diffuse white matter abnormality. Pediatrics. 2003;112:1–7. doi: 10.1542/peds.112.1.1. [DOI] [PubMed] [Google Scholar]
  25. Deprez M, Price A, Christiaens D, Lockwood Estrin G, Cordero-Grande L, Hutter J, Daducci A, Tournier JD, Rutherford M, Counsell SJ, Cuadra MB, Hajnal JV. Higher order spherical harmonics reconstruction of fetal diffusion MRI with intensity correction. IEEE Transactions on Medical Imaging. 2020;39:1104–1113. doi: 10.1109/TMI.2019.2943565. [DOI] [PubMed] [Google Scholar]
  26. Drobyshevsky A, Song S-K, Gamkrelidze G, Wyrwicz AM, Derrick M, Meng F, Li L, Ji X, Trommer B, Beardsley DJ, Luo NL, Back SA, Tan S. Developmental changes in diffusion anisotropy coincide with immature oligodendrocyte progression and maturation of compound action potential. The Journal of Neuroscience. 2005;25:5988–5997. doi: 10.1523/JNEUROSCI.4983-04.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Emos MC, Khan Suheb MZ, Agarwal S. Neuroanatomy, Internal Capsule. StatPearls Publishing; 2022. [PubMed] [Google Scholar]
  28. Ghiglia DC, Romero LA. Robust two-dimensional weighted and unweighted phase unwrapping that uses fast transforms and iterative methods. Journal of the Optical Society of America A. 1994;11:107. doi: 10.1364/JOSAA.11.000107. [DOI] [Google Scholar]
  29. Gholipour A, Rollins CK, Velasco-Annis C, Ouaalam A, Akhondi-Asl A, Afacan O, Ortinau CM, Clancy S, Limperopoulos C, Yang E, Estroff JA, Warfield SK. A normative spatiotemporal MRI atlas of the fetal brain for automatic segmentation and analysis of early brain growth. Scientific Reports. 2017;7:476. doi: 10.1038/s41598-017-00525-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Ghosh A, Antonini A, McConnell SK, Shatz CJ. Requirement for subplate neurons in the formation of thalamocortical connections. Nature. 1990;347:179–181. doi: 10.1038/347179a0. [DOI] [PubMed] [Google Scholar]
  31. Hadders-Algra M, Boxum AG, Hielkema T, Hamer EG. Effect of early intervention in infants at very high risk of cerebral palsy: a systematic review. Developmental Medicine and Child Neurology. 2017;59:246–258. doi: 10.1111/dmcn.13331. [DOI] [PubMed] [Google Scholar]
  32. Haynes RL, Borenstein NS, Desilva TM, Folkerth RD, Liu LG, Volpe JJ, Kinney HC. Axonal development in the cerebral white matter of the human fetus and infant. The Journal of Comparative Neurology. 2005;484:156–167. doi: 10.1002/cne.20453. [DOI] [PubMed] [Google Scholar]
  33. Hüppi PS, Murphy B, Maier SE, Zientara GP, Inder TE, Barnes PD, Kikinis R, Jolesz FA, Volpe JJ. Microstructural brain development after perinatal cerebral white matter injury assessed by diffusion tensor magnetic resonance imaging. Pediatrics. 2001;107:455–460. doi: 10.1542/peds.107.3.455. [DOI] [PubMed] [Google Scholar]
  34. Huttenlocher PR. Synaptic density in human frontal cortex-developmental changes and effects of aging. Brain Research. 1979;163:195–205. doi: 10.1016/0006-8993(79)90349-4. [DOI] [PubMed] [Google Scholar]
  35. Huttenlocher PR, Dabholkar AS. Regional differences in synaptogenesis in human cerebral cortex. The Journal of Comparative Neurology. 1997;387:167–178. doi: 10.1002/(sici)1096-9861(19971020)387:2<167::aid-cne1>3.0.co;2-z. [DOI] [PubMed] [Google Scholar]
  36. Hutter J, Christiaens DJ, Schneider T, Cordero-Grande L, Slator PJ, Deprez M, Price AN, Tournier J-D, Rutherford M, Hajnal JV. Slice-level diffusion encoding for motion and distortion correction. Medical Image Analysis. 2018a;48:214–229. doi: 10.1016/j.media.2018.06.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Hutter J, Slator PJ, Christiaens D, Teixeira RPAG, Roberts T, Jackson L, Price AN, Malik S, Hajnal JV. Integrated and efficient diffusion-relaxometry using zebra. Scientific Reports. 2018b;8:15138. doi: 10.1038/s41598-018-33463-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Jaimes C, Machado-Rivas F, Afacan O, Khan S, Marami B, Ortinau CM, Rollins CK, Velasco-Annis C, Warfield SK, Gholipour A. In vivo characterization of emerging white matter microstructure in the fetal brain in the third trimester. Human Brain Mapping. 2020;41:3177–3185. doi: 10.1002/hbm.25006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Jakab A, Pogledic I, Schwartz E, Gruber G, Mitter C, Brugger PC, Langs G, Schöpf V, Kasprian G, Prayer D. Fetal cerebral magnetic resonance imaging beyond morphology. Seminars in Ultrasound, CT, and MR. 2015;36:465–475. doi: 10.1053/j.sult.2015.06.003. [DOI] [PubMed] [Google Scholar]
  40. 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]
  41. Jenkinson M, Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM. FSL. NeuroImage. 2012;62:782–790. doi: 10.1016/j.neuroimage.2011.09.015. [DOI] [PubMed] [Google Scholar]
  42. Jeurissen B, Tournier JD, Dhollander T, Connelly A, Sijbers J. Multi-tissue constrained spherical deconvolution for improved analysis of multi-shell diffusion MRI data. NeuroImage. 2014;103:411–426. doi: 10.1016/j.neuroimage.2014.07.061. [DOI] [PubMed] [Google Scholar]
  43. Johansen-Berg H, Behrens TEJ, Sillery E, Ciccarelli O, Thompson AJ, Smith SM, Matthews PM. Functional-anatomical validation and individual variation of diffusion tractography-based segmentation of the human thalamus. Cerebral Cortex. 2005;15:31–39. doi: 10.1093/cercor/bhh105. [DOI] [PubMed] [Google Scholar]
  44. Jones EG. The thalamus. 2nd ed. Cambridge University Press; 2007. [Google Scholar]
  45. Keunen K, van der Burgh HK, de Reus MA, Moeskops P, Schmidt R, Stolwijk LJ, de Lange SC, Išgum I, de Vries LS, Benders MJ, van den Heuvel MP. Early human brain development: Insights into macroscale connectome wiring. Pediatric Research. 2018;84:829–836. doi: 10.1038/s41390-018-0138-1. [DOI] [PubMed] [Google Scholar]
  46. Khan S, Vasung L, Marami B, Rollins CK, Afacan O, Ortinau CM, Yang E, Warfield SK, Gholipour A. Fetal brain growth portrayed by a spatiotemporal diffusion tensor MRI atlas computed from in utero images. NeuroImage. 2019;185:593–608. doi: 10.1016/j.neuroimage.2018.08.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Kinney HC, Brody BA, Kloman AS, Gilles FH. Sequence of central nervous system myelination in human infancy. II. Patterns of myelination in autopsied infants. Journal of Neuropathology and Experimental Neurology. 1988;47:217–234. doi: 10.1097/00005072-198805000-00003. [DOI] [PubMed] [Google Scholar]
  48. Kinney HC, Karthigasan J, Borenshteyn NI, Flax JD, Kirschner DA. Myelination in the developing human brain: Biochemical correlates. Neurochemical Research. 1994;19:983–996. doi: 10.1007/BF00968708. [DOI] [PubMed] [Google Scholar]
  49. Klingner CM, Langbein K, Dietzek M, Smesny S, Witte OW, Sauer H, Nenadic I. Thalamocortical connectivity during resting state in schizophrenia. European Archives of Psychiatry and Clinical Neuroscience. 2014;264:111–119. doi: 10.1007/s00406-013-0417-0. [DOI] [PubMed] [Google Scholar]
  50. Kostovic I, Goldman-Rakic PS. Transient cholinesterase staining in the Mediodorsal nucleus of the thalamus and its connections in the developing human and monkey brain. The Journal of Comparative Neurology. 1983;219:431–447. doi: 10.1002/cne.902190405. [DOI] [PubMed] [Google Scholar]
  51. Kostovic I, Rakic P. Development of prestriate visual projections in the monkey and human fetal cerebrum revealed by transient cholinesterase staining. The Journal of Neuroscience. 1984;4:25–42. doi: 10.1523/JNEUROSCI.04-01-00025.1984. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Kostovic I, Rakic P. Developmental history of the transient subplate zone in the visual and somatosensory cortex of the macaque monkey and human brain. The Journal of Comparative Neurology. 1990;297:441–470. doi: 10.1002/cne.902970309. [DOI] [PubMed] [Google Scholar]
  53. Kostović I, Judas M, Rados M, Hrabac P. Laminar organization of the human fetal cerebrum revealed by histochemical markers and magnetic resonance imaging. Cerebral Cortex. 2002;12:536–544. doi: 10.1093/cercor/12.5.536. [DOI] [PubMed] [Google Scholar]
  54. Kostović I., Jovanov-Milosević N. The development of cerebral connections during the first 20-45 weeks’ gestation. Seminars in Fetal & Neonatal Medicine. 2006;11:415–422. doi: 10.1016/j.siny.2006.07.001. [DOI] [PubMed] [Google Scholar]
  55. Kostovic I, Judas M. Prolonged coexistence of transient and permanent circuitry elements in the developing cerebral cortex of fetuses and preterm infants. Developmental Medicine and Child Neurology. 2006;48:388–393. doi: 10.1017/S0012162206000831. [DOI] [PubMed] [Google Scholar]
  56. Kostović I, Judas M. The development of the subplate and thalamocortical connections in the human foetal brain. Acta Paediatrica. 2010;99:1119–1127. doi: 10.1111/j.1651-2227.2010.01811.x. [DOI] [PubMed] [Google Scholar]
  57. Kostović I. The enigmatic fetal subplate compartment forms an early tangential cortical nexus and provides the framework for construction of cortical connectivity. Progress in Neurobiology. 2020;194:101883. doi: 10.1016/j.pneurobio.2020.101883. [DOI] [PubMed] [Google Scholar]
  58. Krsnik Ž, Majić V, Vasung L, Huang H, Kostović I. Growth of thalamocortical fibers to the somatosensory cortex in the human fetal brain. Frontiers in Neuroscience. 2017;11:233. doi: 10.3389/fnins.2017.00233. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Le Gros Clark WE. The topography and homologies of the hypothalamic nuclei in man. Journal of Anatomy. 1936;70:203–214. [PMC free article] [PubMed] [Google Scholar]
  60. Machado-Rivas F, Afacan O, Khan S, Marami B, Velasco-Annis C, Lidov H, Warfield SK, Gholipour A, Jaimes C. Spatiotemporal changes in diffusivity and anisotropy in fetal brain tractography. Human Brain Mapping. 2021;42:5771–5784. doi: 10.1002/hbm.25653. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Marenco S, Stein JL, Savostyanova AA, Sambataro F, Tan HY, Goldman AL, Verchinski BA, Barnett AS, Dickinson D, Apud JA, Callicott JH, Meyer-Lindenberg A, Weinberger DR. Investigation of anatomical thalamo-cortical connectivity and fMRI activation in schizophrenia. Neuropsychopharmacology. 2012;37:499–507. doi: 10.1038/npp.2011.215. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Mathur A, Inder T. Magnetic resonance imaging -- insights into brain injury and outcomes in premature infants. Journal of Communication Disorders. 2009;42:248–255. doi: 10.1016/j.jcomdis.2009.03.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Miller JA, Ding S-L, Sunkin SM, Smith KA, Ng L, Szafer A, Ebbert A, Riley ZL, Royall JJ, Aiona K, Arnold JM, Bennet C, Bertagnolli D, Brouner K, Butler S, Caldejon S, Carey A, Cuhaciyan C, Dalley RA, Dee N, Dolbeare TA, Facer BAC, Feng D, Fliss TP, Gee G, Goldy J, Gourley L, Gregor BW, Gu G, Howard RE, Jochim JM, Kuan CL, Lau C, Lee C-K, Lee F, Lemon TA, Lesnar P, McMurray B, Mastan N, Mosqueda N, Naluai-Cecchini T, Ngo N-K, Nyhus J, Oldre A, Olson E, Parente J, Parker PD, Parry SE, Stevens A, Pletikos M, Reding M, Roll K, Sandman D, Sarreal M, Shapouri S, Shapovalova NV, Shen EH, Sjoquist N, Slaughterbeck CR, Smith M, Sodt AJ, Williams D, Zöllei L, Fischl B, Gerstein MB, Geschwind DH, Glass IA, Hawrylycz MJ, Hevner RF, Huang H, Jones AR, Knowles JA, Levitt P, Phillips JW, Sestan N, Wohnoutka P, Dang C, Bernard A, Hohmann JG, Lein ES. Transcriptional landscape of the prenatal human brain. Nature. 2014;508:199–206. doi: 10.1038/nature13185. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Miyazaki Y, Song JW, Takahashi E. Asymmetry of radial and symmetry of tangential neuronal migration pathways in developing human fetal brains. Frontiers in Neuroanatomy. 2016;10:2. doi: 10.3389/fnana.2016.00002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Mojsilović J, Zecević N. Early development of the human thalamus: Golgi and nissl study. Early Human Development. 1991;27:119–144. doi: 10.1016/0378-3782(91)90033-y. [DOI] [PubMed] [Google Scholar]
  66. Molliver ME, Kostović I, van der Loos H. The development of synapses in cerebral cortex of the human fetus. Brain Research. 1973;50:403–407. doi: 10.1016/0006-8993(73)90741-5. [DOI] [PubMed] [Google Scholar]
  67. Molnár Z, Blakemore C. How do thalamic axons find their way to the cortex? Trends in Neurosciences. 1995;18:389–397. doi: 10.1016/0166-2236(95)93935-q. [DOI] [PubMed] [Google Scholar]
  68. Molnár Z, Adams R, Blakemore C. Mechanisms underlying the early establishment of thalamocortical connections in the rat. The Journal of Neuroscience. 1998;18:5723–5745. doi: 10.1523/JNEUROSCI.18-15-05723.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Molnár Z, Garel S, López-Bendito G, Maness P, Price DJ. Mechanisms controlling the guidance of thalamocortical axons through the embryonic forebrain. The European Journal of Neuroscience. 2012;35:1573–1585. doi: 10.1111/j.1460-9568.2012.08119.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Molnár Z, Hoerder-Suabedissen A. Regional scattering of primate subplate. PNAS. 2016;113:9676–9678. doi: 10.1073/pnas.1611194113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  71. Morel A, Magnin M, Jeanmonod D. Multiarchitectonic and stereotactic atlas of the human thalamus. The Journal of Comparative Neurology. 1997;387:588–630. doi: 10.1002/(sici)1096-9861(19971103)387:4<588::aid-cne8>3.0.co;2-z. [DOI] [PubMed] [Google Scholar]
  72. Mrzljak L, Uylings HB, Kostovic I, van Eden CG. Prenatal development of neurons in the human prefrontal cortex. II. A quantitative Golgi study. The Journal of Comparative Neurology. 1992;316:485–496. doi: 10.1002/cne.903160408. [DOI] [PubMed] [Google Scholar]
  73. Nair A, Treiber JM, Shukla DK, Shih P, Müller RA. Impaired thalamocortical connectivity in autism spectrum disorder: A study of functional and anatomical connectivity. Brain. 2013;136:1942–1955. doi: 10.1093/brain/awt079. [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Najdenovska E, Alemán-Gómez Y, Battistella G, Descoteaux M, Hagmann P, Jacquemont S, Maeder P, Thiran JP, Fornari E, Bach Cuadra M. In-vivo probabilistic atlas of human thalamic nuclei based on diffusion- weighted magnetic resonance imaging. Scientific Data. 2018;5:180270. doi: 10.1038/sdata.2018.270. [DOI] [PMC free article] [PubMed] [Google Scholar]
  75. Nakagawa Y. Development of the thalamus: From early patterning to regulation of cortical functions. Wiley Interdisciplinary Reviews. Developmental Biology. 2019;8:e345. doi: 10.1002/wdev.345. [DOI] [PubMed] [Google Scholar]
  76. Niemann K, Mennicken VR, Jeanmonod D, Morel A. The morel stereotactic atlas of the human thalamus: Atlas-to-MR registration of internally consistent canonical model. NeuroImage. 2000;12:601–616. doi: 10.1006/nimg.2000.0650. [DOI] [PubMed] [Google Scholar]
  77. Nosarti C, Nam KW, Walshe M, Murray RM, Cuddy M, Rifkin L, Allin MPG. Preterm birth and structural brain alterations in early adulthood. NeuroImage. Clinical. 2014;6:180–191. doi: 10.1016/j.nicl.2014.08.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  78. Pietsch M. Advanced Diffusion MRI Analysis Methods for Neonatal Imaging. King’s College; 2018. [Google Scholar]
  79. Pietsch M, Christiaens D, Hutter J, Cordero-Grande L, Price AN, Hughes E, Edwards AD, Hajnal JV, Counsell SJ, Tournier JD. A framework for multi-component analysis of diffusion MRI data over the neonatal period. NeuroImage. 2019;186:321–337. doi: 10.1016/j.neuroimage.2018.10.060. [DOI] [PMC free article] [PubMed] [Google Scholar]
  80. Price DJ, Kennedy H, Dehay C, Zhou L, Mercier M, Jossin Y, Goffinet AM, Tissir F, Blakey D, Molnár Z. The development of cortical connections. The European Journal of Neuroscience. 2006;23:910–920. doi: 10.1111/j.1460-9568.2006.04620.x. [DOI] [PubMed] [Google Scholar]
  81. Price AN, Cordero-Grande L, Hughes E, Hiscocks S, Green E, McCabe L, Hutter J, Ferrazzi G, Deprez M, Roberts T, Christiaens D, Duff E, Karolis V, Malik S, Rutherford M, Edwards AD, Hajnal JV. The developing Human Connectome Project (dHCP): fetal acquisition protocol. ISMRM Annual Meeting & Exhibition.2019. [Google Scholar]
  82. Raffelt D, Tournier JD, Fripp J, Crozier S, Connelly A, Salvado O. Symmetric diffeomorphic registration of fibre orientation distributions. NeuroImage. 2011;56:1171–1180. doi: 10.1016/j.neuroimage.2011.02.014. [DOI] [PubMed] [Google Scholar]
  83. Rakic P. Prenatal development of the visual system in rhesus monkey. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences. 1977;278:245–260. doi: 10.1098/rstb.1977.0040. [DOI] [PubMed] [Google Scholar]
  84. Rakic P. Elusive radial glial cells: historical and evolutionary perspective. Glia. 2003;43:19–32. doi: 10.1002/glia.10244. [DOI] [PubMed] [Google Scholar]
  85. Raybaud C, Ahmad T, Rastegar N, Shroff M, Al Nassar M. The premature brain: developmental and lesional anatomy. Neuroradiology. 2013;55 Suppl 2:23–40. doi: 10.1007/s00234-013-1231-0. [DOI] [PubMed] [Google Scholar]
  86. Schummers J, Sharma J, Sur M. Bottom-up and top-down dynamics in visual cortex. Progress in Brain Research. 2005;149:65–81. doi: 10.1016/S0079-6123(05)49006-8. [DOI] [PubMed] [Google Scholar]
  87. Sharma J, Angelucci A, Sur M. Induction of visual orientation modules in auditory cortex. Nature. 2000;404:841–847. doi: 10.1038/35009043. [DOI] [PubMed] [Google Scholar]
  88. Smith RE, Tournier JD, Calamante F, Connelly A. Anatomically-constrained tractography: improved diffusion MRI streamlines tractography through effective use of anatomical information. NeuroImage. 2012;62:1924–1938. doi: 10.1016/j.neuroimage.2012.06.005. [DOI] [PubMed] [Google Scholar]
  89. Smith RE, Tournier JD, Calamante F, Connelly A. SIFT: spherical-deconvolution informed filtering of tractograms. NeuroImage. 2013;67:298–312. doi: 10.1016/j.neuroimage.2012.11.049. [DOI] [PubMed] [Google Scholar]
  90. Sur M, Rubenstein JLR. Patterning and plasticity of the cerebral cortex. Science. 2005;310:805–810. doi: 10.1126/science.1112070. [DOI] [PubMed] [Google Scholar]
  91. Takahashi E, Folkerth RD, Galaburda AM, Grant PE. Emerging cerebral connectivity in the human fetal brain: an mr tractography study. Cerebral Cortex. 2012;22:455–464. doi: 10.1093/cercor/bhr126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  92. Toulmin H, Beckmann CF, O’Muircheartaigh J, Ball G, Nongena P, Makropoulos A, Ederies A, Counsell SJ, Kennea N, Arichi T, Tusor N, Rutherford MA, Azzopardi D, Gonzalez-Cinca N, Hajnal JV, Edwards AD. Specialization and integration of functional thalamocortical connectivity in the human infant. PNAS. 2015;112:6485–6490. doi: 10.1073/pnas.1422638112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  93. Toulmin H, O’Muircheartaigh J, Counsell SJ, Falconer S, Chew A, Beckmann CF, Edwards AD. Functional thalamocortical connectivity at term equivalent age and outcome at 2 years in infants born preterm. Cortex; a Journal Devoted to the Study of the Nervous System and Behavior. 2021;135:17–29. doi: 10.1016/j.cortex.2020.09.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  94. Tournier JD, Calamante F, Connelly A. Determination of the appropriate B value and number of gradient directions for high-angular-resolution diffusion-weighted imaging. NMR in Biomedicine. 2013;26:1775–1786. doi: 10.1002/nbm.3017. [DOI] [PubMed] [Google Scholar]
  95. Tournier JD, Smith R, Raffelt D, Tabbara R, Dhollander T, Pietsch M, Christiaens D, Jeurissen B, Yeh CH, Connelly A. MRtrix3: a fast, flexible and open software framework for medical image processing and visualisation. NeuroImage. 2019;202:116137. doi: 10.1016/j.neuroimage.2019.116137. [DOI] [PubMed] [Google Scholar]
  96. Tournier J-D, Christiaens D, Hutter J, Price AN, Cordero-Grande L, Hughes E, Bastiani M, Sotiropoulos SN, Smith SM, Rueckert D, Counsell SJ, Edwards AD, Hajnal JV. A data-driven approach to optimising the encoding for multi-shell diffusion MRI with application to neonatal imaging. NMR in Biomedicine. 2020;33:e4348. doi: 10.1002/nbm.4348. [DOI] [PMC free article] [PubMed] [Google Scholar]
  97. Volpe JJ. Neurobiology of periventricular leukomalacia in the premature infant. Pediatric Research. 2001;50:553–562. doi: 10.1203/00006450-200111000-00003. [DOI] [PubMed] [Google Scholar]
  98. Volpe JJ. The encephalopathy of prematurity -- brain injury and impaired brain development inextricably intertwined. Seminars in Pediatric Neurology. 2009;16:167–178. doi: 10.1016/j.spen.2009.09.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  99. Wilkinson M, Kane T, Wang R, Takahashi E. Migration pathways of thalamic neurons and development of thalamocortical connections in humans revealed by diffusion mr tractography. Cerebral Cortex. 2017;27:5683–5695. doi: 10.1093/cercor/bhw339. [DOI] [PMC free article] [PubMed] [Google Scholar]
  100. Wilson S, Pietsch M, Cordero-Grande L, Price AN, Hutter J, Xiao J, McCabe L, Rutherford MA, Hughes EJ, Counsell SJ, Tournier J-D, Arichi T, Hajnal JV, Edwards AD, Christiaens D, O’Muircheartaigh J. Development of human white matter pathways in utero over the second and third trimester. PNAS. 2021;118:e2023598118. doi: 10.1073/pnas.2023598118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  101. Wimberger DM, Roberts TP, Barkovich AJ, Prayer LM, Moseley ME, Kucharczyk J. Identification of “premyelination” by diffusion-weighted MRI. Journal of Computer Assisted Tomography. 1995;19:28–33. doi: 10.1097/00004728-199501000-00005. [DOI] [PubMed] [Google Scholar]
  102. Xu G, Takahashi E, Folkerth RD, Haynes RL, Volpe JJ, Grant PE, Kinney HC. Radial coherence of diffusion tractography in the cerebral white matter of the human fetus: neuroanatomic insights. Cerebral Cortex. 2014;24:579–592. doi: 10.1093/cercor/bhs330. [DOI] [PMC free article] [PubMed] [Google Scholar]
  103. Yakovlev PI, Locke S, Koskoff DY, Patton RA. Limbic nuclei of thalamus and connections of limbic cortex. Archives of Neurology. 1960;3:620–641. doi: 10.1001/archneur.1960.00450060008002. [DOI] [PubMed] [Google Scholar]
  104. Zanin E, Ranjeva J-P, Confort-Gouny S, Guye M, Denis D, Cozzone PJ, Girard N. White matter maturation of normal human fetal brain. An in vivo diffusion tensor tractography study. Brain and Behavior. 2011;1:95–108. doi: 10.1002/brb3.17. [DOI] [PMC free article] [PubMed] [Google Scholar]

Editor's evaluation

Finnegan J Calabro 1

This study presents important new findings regarding prenatal thalamocortical development. The authors present convincing evidence, while overcoming substantial methodological challenges, in charting prenatal brain development in vivo. This work will be of interest to pediatric and developmental neuroscientists and neuroradiologists.

Decision letter

Editor: Finnegan J Calabro1
Reviewed by: Finnegan J Calabro2, Cassandra Hendrix3, Zoltan Molnar4

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 "Spatiotemporal tissue maturation of thalamocortical pathways in the human fetal brain" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including involved in the review of your submission have agreed to reveal their identity: Finnegan as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Jonathan Roiser as the Senior Editor. The following individuals involved in the review of your submission have agreed to reveal their identity: Cassandra Hendrix (Reviewer #2); Zoltan Molnar (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:

The reviewers were in agreement that the paper represents an impressive accomplishment that will be impactful to the field, and which generates multiple novel insights. Several points were raised for aspects of the manuscript that could be strengthened.

1) Justify and clarify aspects of the methodological approach, including the population characteristics, several aspects of the statistical approaches employed, and potential limitations of the acquisition being applied to this population.

2) Demonstrate and discuss the robustness of the sensitivity of the results presented.

3) Consider further implications of these findings in the context of known anatomy of the thalamus and its development in prior literature.

For further details please see the "Public reviews" and "Recommendations for the authors" below.

Reviewer #1 (Recommendations for the authors):

This is an impressive study that overcomes a number of obstacles in performing fetal imaging analysis.

Regarding the application of MSMT-CSD modeling, the b-values used in the current study (400, 1000) seem comparatively low to previous studies employing MSMT modeling which I am aware of. Some clarification regarding the validity of these approaches given the sequences used (e.g., in consideration of Tournier, et al., 2013), e.g., including citing prior studies demonstrating validity for a similar range of b-values, would lend support to the application of these analyses to the current data set.

Assessment of the location-specific developmental effects is accomplished by normalizing each tract on a 0-100 scale between the thalamus and cortex. Does this normalization, in the context of an expanding brain, imply that a given point may be comparatively closer to the terminal zone at different ages? Could this account for (all or some of the) age-related changes in microstructure identified (i.e., if certain features are age-invariant with respect to absolute, rather than relative, position along the tract)? Demonstrating that group differences persist when selecting (e.g.) constant length segments at the ends and middle of each tract would help rule out brain size-induced topographic change as driving these effects (though other statistical approaches could likely also accomplish this).

I'm unsure what statistics support conclusions regarding position-specific differences in tissue fraction. E.g., the discussion summarizes that "the early period is characterized by higher tissue fractions in the middle of the [thalamocortical] tract". While this appears likely based on visual inspection of the tracts in Figure 5, it is not clear what specific statistical analyses support this inference. Further, given that differences were assessed both as a function of GA (though perhaps exclusively by group) and as a function of distance, some degree of multiple comparison correction is probably warranted, but not currently employed (or at least, not described).

Reviewer #2 (Recommendations for the authors):

First, I would like to commend the authors on a thoughtful, creative combination of histological maps and dMRI data. One of the major challenges of fetal neuroimaging work is understanding how co-occurring and dynamic neurobiological processes contribute to metrics derived from in vivo imaging. The present study presents an exciting model for contextualizing dMRI findings in relation to what we know about brain development from histology studies. The following recommendations are offered with the hope that they can strengthen an already exceptional paper:

1. Please provide a table with sociodemographic information about the sample. This information is necessary for determining the representativeness of the sample, inferring the generalizability of the present findings, and working towards appropriate diversification of developmental neuroscience as a field. Given that age-related change is the primary question of interest, it would also strengthen the study to show that fetuses of different ages do not systematically differ from one another in terms of other factors that may alter developmental trajectories.

2. The manuscript would benefit from more detail regarding motion correction. The authors describe using slice-to-volume motion correction (SHARD) and implementing quality control based on slice dropouts and gradient of motion parameters. However, additional information is required to enhance transparency and aid future replication efforts. What cut-offs were used to determine whether to include vs. exclude a subject? How many fetuses were excluded based on the selected quality control thresholds? Can the authors please provide more information about the visual assessment that was conducted? Were post-processing motion parameters correlated with tissue or fluid fraction values or with fetal age? It would also be helpful to include visual examples of residual or uncorrected artefacts that resulted in subject exclusion, even if only in the supplemental materials. Finally, Christiaens et al., 2021 is cited several times in the text (including in reference to preprocessing techniques used for correcting head motion) but is not included in the reference list.

3. What size were the along-tract analysis segments and how was this segment size selected? Showing the results replicate (or don't) with different segment sizes would show the robustness of results and could inform an ideal resolution at which to examine tissue maturation in the fetal brain.

4. Additional analyses that show the primary findings do not replicate with noise data but do replicate across different preprocessing choices would enhance rigor, especially since the field of fetal neuroimaging lacks a widely agreed-upon set of best practices. For instance, if segments from the along-tract analysis are scrambled, does the outward-to-inward maturational trajectory disappear? The disappearance of associations with scrambled data would enhance confidence that this finding is not simply a byproduct of the way the data were preprocessed. In addition, showing the results replicate following different preprocessing choices would aid in showing effect robustness.

5. Several studies have identified sex differences in fetal brain maturation. Given the even split in the present sample, examining whether tissue maturation trajectories differ by fetal sex would be informative. I wonder if the authors would consider adding this analysis.

6. The limitations of fetal dMRI should be clearly described in the discussion to aid reader interpretation.

7. It would improve manuscript readability to clarify why the authors believe it is important to report both tissue and fluid fractions. They appear to be reciprocals of one another so the added value of reporting both is unclear.

8. I would avoid subjective phrases like "thalamocortical connections represent the most important inputs into the developing cortex" and instead use more neutral language like "thalamocortical connections play a key role in guiding…"

Reviewer #3 (Recommendations for the authors):

Overall, this is a very valuable study with excellent datasets. The material provides a unique and detailed description of the changes in microstructure in the thalamus and cortex. I am very supportive of publication, but I feel that more could have been done to bridge the gap between MRI and histology and to link the results to the basic principles of thalamocortical development.

Criticisms:

1. Using tract density imaging, the authors observed that the cortical areas were connected to specific thalamic regions, organised in an anterior-posterior axis. This anterior-posterior representation of cortical connectivity in the thalamus was consistent with previous descriptions in rodent thalamocortical connectivity in rodent. However, the relationship between the cortical surface and the volume of the developing thalamus is not explained clearly. In their studies, Molnar and Blakemore emphasized that at the time of arrival, the thalamic projections are already arranged according to rough topography. Anterior to posterior movement along the convexity of the cortex represented a medial-to-lateral movement of labelling in the embryonic rodent thalamus. Ventral to dorsal movement in the cortex resulted in an anterior-to-posterior movement within the thalamus https://doi.org/10.1523/JNEUROSCI.18-15-05723.1998; https://doi.org/10.1111/j.1460-9568.2012.08119.x. It would be good to improve Figure 3 and represent the overall pattern. The very general pattern should be described. Could the thalamic volumes be color coded and the same fainter shade used on the convexity of the cortex of the 3D outline of the brain?

2. The study does not attempt to link the thalamic fibre topography to the histological structure of the thalamus. Thalamic nuclei undergo considerable changes at these early stages, e.g. pulvinar increases enormously and the dorsal lateral geniculate nucleus is changing position from dorsolateral to ventromedial. Such changes were already emphasized in the publications of Le Gros Clark and subsequently Rakic. The authors note that the topography of thalamic nuclei and their cortical connectivity is acquired embryonically, but they do not emphasize the considerable changes in this topography at early stages. The in vivo parcellation of the thalamus is only described as part of the fibre topography, no references are made to the changes in relative positions of the thalamic nuclei, and only the adult similarity is emphasized.

3. The density of the thalamocortical innervation is not equal across the developing or adult cortex. E.g. retrogradely tracing thalamic projections from various parts of the cortex using carbocyanine dyes, it was observed that the same sized carbocyanine dye can label 3-5 times more backlabelled thalamic neurons from the putative somatosensory cortex than from the visual cortex (see references in Z. Molnár, Development of Thalamocortical Connections (Springer, 1998)). The methodologies and the data would have enabled the authors to compare the parameters of the various thalamocortical pathways that they resolved. It would have been good to include some results and comments on these area-specific comparisons.

4. The authors found the spatiotemporal changes in the diffusion signal reflected known developmental processes that take place between the early, mid, and late prenatal period. "The early period is characterized by higher tissue fractions in the middle of the tract, where there is a radial scaffold for migrating neurons. As this scaffold dissipates in the mid-prenatal period, this is accompanied by a reduction in the tissue fraction in the middle of the tract, and an increase in towards the termination of the tracts as the neurons of the cortical plate mature." The fact that these changes can be now resolved with in vivo imaging is excellent, but the data itself is under-analysed.

The authors correctly describe the possible anatomical explanations for the changes of the signals: "In terms of the diffusion signal, this strongly aligned microstructure of the radial glia is represented in our results by a higher tissue fraction in the transient compartments containing the most migratory cells (such as the VZ, IZ). Conversely, we observe the lowest tissue fraction in the early prenatal SP, as this compartment predominantly contains hydrophilic extracellular matrix." These are key processes of cortical development and the authors have the material to draw conclusions on the various gradients on the cortical plate and subplate changes. It would have been good to get a feel for these dynamic changes and include some representations of the area-specificity of these changes. Surely the increase of the cortical plate has areal variation and also the increase and resolution of the subplate are also area specific. Measuring and representing these differences would increase the impact of the study. Moreover, it would also open avenues to study left-right and gender differences.

5. The authors have interesting discussions on the corticofugal and thalamocortical connectivity through the internal capsule. "The microstructural change along the anterior thalamic radiation suggests increasing tissue fraction between the deep grey matter, VZ, and IZ. We hypothesise that the high tissue fraction in the IZ is due to densely packed ascending and descending bundles within the anterior limb of the internal capsule." The exact timing of the outgrowth, intermingling, and arrival of the thalamocortical and corticothalamic projections are not known, but there are some suggestions that these are relatively earlier than in rodents (https://doi.org/10.1093/cercor/bhz056; https://doi.org/10.1016/B978-0-444-53860-4.00003-9).

6. A wide range of references from various species are used. This is excellent, but it can be also somewhat confusing. I recommend stating the species explicitly when non-human studies are referenced.

7. The formatting of the references will need attention: e.g. in line 69: (I. Kostović and Judaš 2015; Ivica Kostović and Judaš 2010); In line 257: Hutter, Christiaens, et al. 2018; In lines 275 and 277: Clascá, Rubio-Garrido, and Jabaudon 2012; In line 299: Miyazaki, Song, and Takahashi 2016; in line 361: (Ivica Kostovic and Rakic 1990b); in line 364: (Ivica Kostović and Judaš 2006; Ivica Kostovic and Rakic 1990a). In line 374: "Toulmin cerebral cortex" is mentioned, but no specific reference is cited.

Overall, I am very enthusiastic about this very significant and high-quality human developmental imaging data. I understand that the authors are making all this data available for future studies in the field. Nevertheless, I feel that they are in the best position to present some fundamental initial descriptions firsthand and present them in a format with suitable illustrations that will make it into textbooks on human development.

eLife. 2023 Apr 3;12:e83727. doi: 10.7554/eLife.83727.sa2

Author response

Reviewer #1 (Recommendations for the authors):

This is an impressive study that overcomes a number of obstacles in performing fetal imaging analysis.

Regarding the application of MSMT-CSD modeling, the b-values used in the current study (400, 1000) seem comparatively low to previous studies employing MSMT modeling which I am aware of. Some clarification regarding the validity of these approaches given the sequences used (e.g., in consideration of Tournier, et al., 2013), e.g., including citing prior studies demonstrating validity for a similar range of b-values, would lend support to the application of these analyses to the current data set.

We agree that providing insight about the choice of acquisition parameters would benefit readers, as these were specifically optimised for the unique biophysical properties of the fetal environment. We have now signposted the publications that provide detail on the optimisation process used to define the b-shell values used (Tournier et al., 2019 and Christiaens et al., 2019) more clearly in the text (line 451). In essence, these papers demonstrate that there is a marked loss of signal at higher b-shells (> 1000 s/mm3) in the neonatal brain compared to adults, due to the tissue being inherently much less restricted. There is also increased noise at higher b-shells. Both issues are amplified in the fetal brain, there are long relaxation times in the water-rich, relatively unrestricted fetal brain tissues and the amount of noise is higher.

In terms of the validity of MSMT-CSD for this dataset, an important consideration is that the MSMT-CSD approach doesn't mandate a specific set of b-values which are required for the modelling process, only that there are multiple shells in addition to b = 0. The technique depends on the signals related to different tissue types having sufficiently distinct behaviours across the b-values sampled, and so depends inherently on the characteristics of the tissue. To demonstrate the distinct properties of the different compartments in the fetal brain, we have now included a figure in the Appendix (Appendix 1-Figure 4) showing the signal amplitude per tissue type as a function of b-value, and the ODFs after MSMT-CSD has been applied. We have also provided further detail on the motivation for this choice in the discussion (line 265-271).

There is a general scarcity of literature in the fetal diffusion imaging field, but to support that the MSMT-CSD model is as valid as traditional approaches (DTI), in our previous work (Wilson et al., 2021), we demonstrated a high linear correlation between diffusion tensor metrics and the MSMT-CSD components when whole-tract averages are calculated.

Assessment of the location-specific developmental effects is accomplished by normalizing each tract on a 0-100 scale between the thalamus and cortex. Does this normalization, in the context of an expanding brain, imply that a given point may be comparatively closer to the terminal zone at different ages? Could this account for (all or some of the) age-related changes in microstructure identified (i.e., if certain features are age-invariant with respect to absolute, rather than relative, position along the tract)? Demonstrating that group differences persist when selecting (e.g.) constant length segments at the ends and middle of each tract would help rule out brain size-induced topographic change as driving these effects (though other statistical approaches could likely also accomplish this).

We agree that it is important to consider the effect of increasing fetal brain size when studying maturational changes across gestational age. We have operated under the assumption that the morphology of the tracts is consistent with age and thus that the tracts consistently project from the same origin to termination point in all fetuses (for example, from the thalamus to the primary motor cortex). Therefore, any change in the length of tract across gestational age will remain proportional to the change in overall brain size (or distance between the start and end of the tract) and thus each sampling point will represent the same relative distance from the origin and termination across different fetal ages.

We also used this normalisation approach together with an atlas, to cross-reference our trajectories of diffusion values. By overlaying the tracts on the atlas, and corroborating each sampling point with an atlas region, we characterised the diffusion properties of the different fetal compartments independent of their change in size and maturity across gestation. We have included more detail on this choice in the Results section to clarify it for readers (line 218-220).

I'm unsure what statistics support conclusions regarding position-specific differences in tissue fraction. E.g., the discussion summarizes that "the early period is characterized by higher tissue fractions in the middle of the [thalamocortical] tract". While this appears likely based on visual inspection of the tracts in Figure 5, it is not clear what specific statistical analyses support this inference. Further, given that differences were assessed both as a function of GA (though perhaps exclusively by group) and as a function of distance, some degree of multiple comparison correction is probably warranted, but not currently employed (or at least, not described).

Thank you for this suggestion, we agree that a more formal statistical comparison supports interpretation and strengthens the conclusion about microstructural differences between compartments. We have therefore correlated the tissue fraction at each sampling point with gestational age in the thalamic motor tract to demonstrate that there are significant associations with age. As there are 100 sampling points, we also used strict Bonferroni correction for multiple comparisons. This additional analysis has been mentioned in the Results section (line 233-237) and is shown in Author response image 1.

Author response image 1.

Author response image 1.

Open in a new tab

Reviewer #2 (Recommendations for the authors):

First, I would like to commend the authors on a thoughtful, creative combination of histological maps and dMRI data. One of the major challenges of fetal neuroimaging work is understanding how co-occurring and dynamic neurobiological processes contribute to metrics derived from in vivo imaging. The present study presents an exciting model for contextualizing dMRI findings in relation to what we know about brain development from histology studies. The following recommendations are offered with the hope that they can strengthen an already exceptional paper:

1. Please provide a table with sociodemographic information about the sample. This information is necessary for determining the representativeness of the sample, inferring the generalizability of the present findings, and working towards appropriate diversification of developmental neuroscience as a field. Given that age-related change is the primary question of interest, it would also strengthen the study to show that fetuses of different ages do not systematically differ from one another in terms of other factors that may alter developmental trajectories.

We thank the reviewer for their enthusiasm about our work, and for highlighting the absence of this important information. We have included a sentence in the methods section (line 440) about the recruitment process, where this data was acquired and sociodemographic information about the cohort in the Appendix 1-Figure 1 (including histograms to demonstrate that IMD and ethnicity distributions are well matched to the diverse population in London). We have also included Author response image 2 to demonstrate that there is no significant relationship between IMD and gestational age, suggesting that this is not influencing the microstructural trajectories we observe. When the dataset becomes publicly available later this year, a supporting publication will also be released containing all this information.

Author response image 2.

Author response image 2.

Open in a new tab

2. The manuscript would benefit from more detail regarding motion correction. The authors describe using slice-to-volume motion correction (SHARD) and implementing quality control based on slice dropouts and gradient of motion parameters. However, additional information is required to enhance transparency and aid future replication efforts. What cut-offs were used to determine whether to include vs. exclude a subject? How many fetuses were excluded based on the selected quality control thresholds? Can the authors please provide more information about the visual assessment that was conducted? Were post-processing motion parameters correlated with tissue or fluid fraction values or with fetal age? It would also be helpful to include visual examples of residual or uncorrected artefacts that resulted in subject exclusion, even if only in the supplemental materials. Finally, Christiaens et al., 2021 is cited several times in the text (including in reference to preprocessing techniques used for correcting head motion) but is not included in the reference list.

We appreciate that for reproducibility, more detail about precise exclusion criteria would benefit the reader. The methods section now contains additional information about this process (line 464-474), and the appendix includes examples of subjects that were excluded due to motion and residual artefacts (Appendix 1 – Figure 2).

We have also included correlations between motion-related parameters extracted from the SHARD pipeline, (translation, rotation and SNR) (Christiaens et al., 2021) and gestational age, demonstrating no relationship with age. This is included in Appendix 1-Figure 3. Thank you for pointing out the referencing error, we have now corrected it. We would also like to highlight that when this dataset becomes publicly available later this year, QC scores and motion parameters will also be released that will enhance reproducibility for future studies.

3. What size were the along-tract analysis segments and how was this segment size selected? Showing the results replicate (or don't) with different segment sizes would show the robustness of results and could inform an ideal resolution at which to examine tissue maturation in the fetal brain.

The normalisation approach we used was to standardise the sampling interval along all the tracts to 100 points between the origin and termination. This was to ensure consistency across the cohort, account for changing brain size and image resolution. We have assumed that the morphology of the tracts remains consistent across all ages, as we were able to delineate tracts even in the youngest fetuses which projected from the same origin to termination points (for example, from the thalamus to the primary motor cortex). As the start and end points of the tracts are the same regardless of gestational age, any maturational change in length is proportional to a corresponding change in brain size. The smooth transition between each segment, as demonstrated by the consistency of the curves and their gradual shift with age, suggests that this sampling interval is appropriate for the aims of this study. The error bars shown on Figure 5 that reflect the standard error of the mean also support this, as they demonstrate minimal variation between subjects and a lack of overlap between sections in each of the time windows.

4. Additional analyses that show the primary findings do not replicate with noise data but do replicate across different preprocessing choices would enhance rigor, especially since the field of fetal neuroimaging lacks a widely agreed-upon set of best practices. For instance, if segments from the along-tract analysis are scrambled, does the outward-to-inward maturational trajectory disappear? The disappearance of associations with scrambled data would enhance confidence that this finding is not simply a byproduct of the way the data were preprocessed. In addition, showing the results replicate following different preprocessing choices would aid in showing effect robustness.

We agree with the reviewer that an important caveat in the field of fetal imaging is the lack of a standardised set of procedures with which to conduct analysis of MR data. However, we would like to emphasise that the pre-processing pipeline for the dHCP data has been optimised through an extensive and rigorous procedure, such that the methods used for the reconstruction, motion correction and distortion correction produce the highest quality data (more detail on this procedure is available in the following publications cited in the article: Christiaens et al., 2019, 2021, Cordero-Grande 2019, Pietsch et al., 2019). Please see Appendix 1-Figure 3, where we use parameters derived from the SHARD reconstruction pipeline to demonstrate there are no correlations between motion and gestational age. This figure also highlights that for fetuses of the same age, there is a broad distribution of SNR/translation/slice weight values. We feel that this enhances confidence in our results, because the differences between SNR within a gestational week are not reflected in the tissue and fluid fraction values, which in contrast are consistent for subjects of the same age.

In reference to the suggestion to scramble the segments, we feel that shuffling the values presented in figure 5 would lead to disappearance of the trajectory because of the non-overlapping error bars between compartments at each age. To strengthen our analysis, we have taken on board the initial lack of statistical testing and computed correlations between each sampling point tissue fraction and gestational age and corrected for multiple comparisons (see above, response to reviewer one).

5. Several studies have identified sex differences in fetal brain maturation. Given the even split in the present sample, examining whether tissue maturation trajectories differ by fetal sex would be informative. I wonder if the authors would consider adding this analysis.

We agree this would be beneficial to include and have tested for statistical significance between male and female in the average values across the whole tract. We found there was no significant difference between the average diffusion values in any of the tracts. Please see a statement about this finding in the Results section (line 189).

6. The limitations of fetal dMRI should be clearly described in the discussion to aid reader interpretation.

We agree this is a useful suggestion to help to contextualise the study and the methods used. We have now included a specific limitations section in the discussion (line 388-425) and refer reviewers to a previous publication that comprehensively describes these challenges (Christiaens et al., 2019).

7. It would improve manuscript readability to clarify why the authors believe it is important to report both tissue and fluid fractions. They appear to be reciprocals of one another so the added value of reporting both is unclear.

Thank you for highlighting this point. When fitting this model, the two components are reciprocals of one another if the modelling method has fit the data well and the normalization procedure has worked, so we include this to aid reader interpretation of the results and confirm that data validity. To make our rationale clearer to readers, we have included a sentence in the Results section (line 203-205).

8. I would avoid subjective phrases like "thalamocortical connections represent the most important inputs into the developing cortex" and instead use more neutral language like "thalamocortical connections play a key role in guiding…"

We thank the reviewer for this suggestion and agree more neutral language is appropriate. This suggestion has now been included in the revised article (line 50).

Reviewer #3 (Recommendations for the authors):

Overall, this is a very valuable study with excellent datasets. The material provides a unique and detailed description of the changes in microstructure in the thalamus and cortex. I am very supportive of publication, but I feel that more could have been done to bridge the gap between MRI and histology and to link the results to the basic principles of thalamocortical development.

Criticisms:

1. Using tract density imaging, the authors observed that the cortical areas were connected to specific thalamic regions, organised in an anterior-posterior axis. This anterior-posterior representation of cortical connectivity in the thalamus was consistent with previous descriptions in rodent thalamocortical connectivity in rodent. However, the relationship between the cortical surface and the volume of the developing thalamus is not explained clearly. In their studies, Molnar and Blakemore emphasized that at the time of arrival, the thalamic projections are already arranged according to rough topography. Anterior to posterior movement along the convexity of the cortex represented a medial-to-lateral movement of labelling in the embryonic rodent thalamus. Ventral to dorsal movement in the cortex resulted in an anterior-to-posterior movement within the thalamus https://doi.org/10.1523/JNEUROSCI.18-15-05723.1998; https://doi.org/10.1111/j.1460-9568.2012.08119.x. It would be good to improve Figure 3 and represent the overall pattern. The very general pattern should be described. Could the thalamic volumes be color coded and the same fainter shade used on the convexity of the cortex of the 3D outline of the brain?

We are grateful that this missing but extremely important element of our discussion has been highlighted and have now modified the manuscript accordingly, please see discussion (line 286). We have also adjusted Figure 3 to highlight the anterior-posterior connectivity gradient across the cortex and shaded the different nuclei accordingly. In addition, we have changed the colours of the tracts in Figure 2 according to their anterior-posterior connectivity to clearly demonstrate this pattern.

2. The study does not attempt to link the thalamic fibre topography to the histological structure of the thalamus. Thalamic nuclei undergo considerable changes at these early stages, e.g. pulvinar increases enormously and the dorsal lateral geniculate nucleus is changing position from dorsolateral to ventromedial. Such changes were already emphasized in the publications of Le Gros Clark and subsequently Rakic. The authors note that the topography of thalamic nuclei and their cortical connectivity is acquired embryonically, but they do not emphasize the considerable changes in this topography at early stages. The in vivo parcellation of the thalamus is only described as part of the fibre topography, no references are made to the changes in relative positions of the thalamic nuclei, and only the adult similarity is emphasized.

Thank you for pointing out that this key feature of thalamic development was missing from our discussion. With the resolution (8mm3) and image contrast of our images, smaller changes in the position or size of individual nuclei are unfortunately difficult to evaluate though we agree that being able to resolve these changes concurrently would further reinforce our findings. We have now added more on the histological structure of the thalamus to our discussion (line 305-310).

3. The density of the thalamocortical innervation is not equal across the developing or adult cortex. E.g. retrogradely tracing thalamic projections from various parts of the cortex using carbocyanine dyes, it was observed that the same sized carbocyanine dye can label 3-5 times more backlabelled thalamic neurons from the putative somatosensory cortex than from the visual cortex (see references in Z. Molnár, Development of Thalamocortical Connections (Springer, 1998)). The methodologies and the data would have enabled the authors to compare the parameters of the various thalamocortical pathways that they resolved. It would have been good to include some results and comments on these area-specific comparisons.

Thank you for this suggestion. While this would be an interesting question to address, unfortunately we feel that tractography itself (particularly in fetal data) cannot be reliably used to assess fibre morphology. The diffusion MR signal has no directionality, and therefore we are unfortunately not able to distinguish between ascending and descending thalamic projections as they develop. We also find that metrics generally used in this field to investigate fibre bundle morphology, such as streamline count, a proxy for ‘innervation density’, are not an accurate, quantitative measure. These are outlined well in this review article: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6787694/. Consequently, we have avoided using the data in this way, and just used tractography to delineate regions of interest. We have now added a limitations section to the article and mention this constraint of diffusion methodology specifically (line 388).

4. The authors found the spatiotemporal changes in the diffusion signal reflected known developmental processes that take place between the early, mid, and late prenatal period. "The early period is characterized by higher tissue fractions in the middle of the tract, where there is a radial scaffold for migrating neurons. As this scaffold dissipates in the mid-prenatal period, this is accompanied by a reduction in the tissue fraction in the middle of the tract, and an increase in towards the termination of the tracts as the neurons of the cortical plate mature." The fact that these changes can be now resolved with in vivo imaging is excellent, but the data itself is under-analysed.

The authors correctly describe the possible anatomical explanations for the changes of the signals: "In terms of the diffusion signal, this strongly aligned microstructure of the radial glia is represented in our results by a higher tissue fraction in the transient compartments containing the most migratory cells (such as the VZ, IZ). Conversely, we observe the lowest tissue fraction in the early prenatal SP, as this compartment predominantly contains hydrophilic extracellular matrix." These are key processes of cortical development and the authors have the material to draw conclusions on the various gradients on the cortical plate and subplate changes. It would have been good to get a feel for these dynamic changes and include some representations of the area-specificity of these changes. Surely the increase of the cortical plate has areal variation and also the increase and resolution of the subplate are also area specific. Measuring and representing these differences would increase the impact of the study. Moreover, it would also open avenues to study left-right and gender differences.

We appreciate that this is a very rich dataset and there are a lot of questions that we have not addressed or analysed within this paper. Incorporating measures of areal variation in cortical plate development will require surface projection to calculate measures such as cortical thickness and sulcal depth reliably across subjects. Unfortunately, the required processing is non-trivial for fetal MRI data and requires careful optimisation, and thus we felt was beyond the scope of this work. Work is thus currently underway to explore regional differences across the brain for a separate publication that will use a different approach with broader coverage across the whole brain, allowing us to examine areal trends more comprehensively. We will also plan to eventually release surface projected data with the dHCP data release. We imagine this subsequent study will support the findings presented in this paper however, we did not want to overcomplicate the results as we aimed to investigate thalamocortical development more specifically.

5. The authors have interesting discussions on the corticofugal and thalamocortical connectivity through the internal capsule. "The microstructural change along the anterior thalamic radiation suggests increasing tissue fraction between the deep grey matter, VZ, and IZ. We hypothesise that the high tissue fraction in the IZ is due to densely packed ascending and descending bundles within the anterior limb of the internal capsule." The exact timing of the outgrowth, intermingling, and arrival of the thalamocortical and corticothalamic projections are not known, but there are some suggestions that these are relatively earlier than in rodents (https://doi.org/10.1093/cercor/bhz056; https://doi.org/10.1016/B978-0-444-53860-4.00003-9).

We thank the reviewer for this suggestion and are grateful for them highlighting this incredibly relevant study for our discussion. Please see discussion (line 373).

6. A wide range of references from various species are used. This is excellent, but it can be also somewhat confusing. I recommend stating the species explicitly when non-human studies are referenced.

We have prefaced references to non-human studies with a description of the specific species in the revised manuscript to make it clearer for readers.

7. The formatting of the references will need attention: e.g. in line 69: (I. Kostović and Judaš 2015; Ivica Kostović and Judaš 2010); In line 257: Hutter, Christiaens, et al. 2018; In lines 275 and 277: Clascá, Rubio-Garrido, and Jabaudon 2012; In line 299: Miyazaki, Song, and Takahashi 2016; in line 361: (Ivica Kostovic and Rakic 1990b); in line 364: (Ivica Kostović and Judaš 2006; Ivica Kostovic and Rakic 1990a). In line 374: "Toulmin cerebral cortex" is mentioned, but no specific reference is cited.

Thank you for pointing out these inconsistences, we have corrected them and corrected the bibliography in the revised manuscript.

Overall, I am very enthusiastic about this very significant and high-quality human developmental imaging data. I understand that the authors are making all this data available for future studies in the field. Nevertheless, I feel that they are in the best position to present some fundamental initial descriptions firsthand and present them in a format with suitable illustrations that will make it into textbooks on human development.

Associated Data

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

    Data Citations

    1. Uus A, Kyriakopoulou V, Cordero Grande L, Pietsch M, Price A, Wilson S, Patkee P, Karolis V, Schuch A, Gartner A, Williams L, Hughes E, Arichi T, O'Muircheartaigh J, Hutter J, Robinson E, Tournier JD, Rueckert D, Counsell SJ, Rutherford MA, Deprez M, Hajnal JV, Edwards AD. 2023. Multi-channel spatio-temporal MRI atlas of normal fetal brain development (dHCP project) G-Node Open Data. [DOI]
    2. Gholipour A, Rollins CK, Velasco-Annis C, Ouaalam A, Akhondi-Asl A, Afacan O, Ortinau C, Clancy S, Limperopoulos C, Yang E, Estroff JA, Warfield SK. 2017. A normative spatiotemporal MRI atlas of the fetal brain for automatic segmentation and analysis of early brain growth. Harcard CRL/fetal brain atlas. fetal_brain_atlas/ [DOI] [PMC free article] [PubMed]

    Supplementary Materials

    MDAR checklist
    elife-83727-mdarchecklist1.docx (99.8KB, docx)

    Data Availability Statement

    Developing Human Connectome project data is open-access and available for download following completion of a data-usage agreement via: http://www.developingconnectome.org/. Data is also available at: https://nda.nih.gov/edit_collection.html?id=3955.

    The following dataset was generated:

    Uus A, Kyriakopoulou V, Cordero Grande L, Pietsch M, Price A, Wilson S, Patkee P, Karolis V, Schuch A, Gartner A, Williams L, Hughes E, Arichi T, O'Muircheartaigh J, Hutter J, Robinson E, Tournier JD, Rueckert D, Counsell SJ, Rutherford MA, Deprez M, Hajnal JV, Edwards AD. 2023. Multi-channel spatio-temporal MRI atlas of normal fetal brain development (dHCP project) G-Node Open Data.

    The following previously published dataset was used:

    Gholipour A, Rollins CK, Velasco-Annis C, Ouaalam A, Akhondi-Asl A, Afacan O, Ortinau C, Clancy S, Limperopoulos C, Yang E, Estroff JA, Warfield SK. 2017. A normative spatiotemporal MRI atlas of the fetal brain for automatic segmentation and analysis of early brain growth. Harcard CRL/fetal brain atlas. fetal_brain_atlas/

    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES