Atlas-based Investigation of Human Brain Tissue Microstructural Spatial Heterogeneity and Interplay between Transverse Relaxation Time and Radial Diffusivity

Abstract

Microstructural metrics obtained using magnetic resonance imaging (MRI) such as transverse relaxation time and radial diffusivity have been used as in vivo markers of human brain tissue integrity. Considering the sensitivity of these parameters to some common biophysical contributors and their structural and spatial heterogeneity, we hypothesized that strong inter and intra-regional association exist between these variables providing evidence to possible interplay between transverse relaxation time and radial diffusivity. To validate our hypothesis we obtained high resolution anatomical T1-weighted data and fused it with T2-relaxomotry and diffusion tensor imaging (DTI) data on a cohort of healthy adults. The anatomical data were parcellated using FreeSurfer and then coaligned and fused with the T2 and DTI maps. Our data reveal some association between transverse relaxation and radial diffusivity that may help towards the interpretation and modeling of the biophysical contributors to the measured MRI metrics.

Introduction

Magnetic resonance imaging (MRI) provides several parameters that can be used as in vivo markers of the structural alterations to brain tissue that accompany development, natural aging and neurological diseases such as epilepsy, multiple sclerosis and Alzheimer’s disease. Macroscopic volumetry derived by T1-weighted MRI can be used to investigate age related volume atrophy (Walhovd et al., 2005; Fjell et al., 2009; Ostby et al., 2009) and disease-driven volume atrophy (Ramasamy, et al., 2009; Hasan et al., 2009a) of various regions of the brain.

Quantitative MRI tissue markers such as T2 relaxation time derived using multiple spin-echo maps (Ono et al., 1993; Whittall et al., 1997; Baratti et al., 1999) and diffusion tensor imaging (DTI) metrics (Pierpaoli et al., 1996; Basser et al., 1997) such as fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (λ_||) and radial diffusivity (λ_⊥) are sensitive to tissue microstructural parameters such as intra-axonal and cellular integrity, water distribution and myelin content in the brain (Le Bihan et al., 2001; Beaulieu, 2002). Transverse relaxation time has been hypothesized to be a sensitive marker of myelination (Ono et al., 1993; Whittall et al., 1997), structural integrity (Georgiades et al., 2001; Bartzokis et al., 2004) and has also been shown to be affected by non-heme iron deposition (Haacke et al., 2005; Jara et al., 2006; Bartzokis et al., 2007; House et al., 2008; Mitsumori et al., 2009; Yao et al., 2009). DTI metrics such as FA MD, (Beaulieu, 2002) and subsequently generated fiber tracks have been λ_|| and λ_⊥ utilized in assessing the microstructural integrity of compact white matter (WM) fiber bundles (Stieltjes et al., 2001, Wakana et al., 2007; Lebel et al., 2008; Hasan et al., 2009b; Wahl et al., 2010). Understanding the regional distribution of these metrics and their association is essential as these parameters vary with age and disease. However, the specificity of these metrics using the contrast provided by each modality may not be optimal (Assaf et al., 2008; Bar-Shir et al, 2009; Barazany et al., 2009; Paus, et al., 2009) due to low spatial resolution and low signal-to-noise ratio (SNR).

A comprehensive empirical investigation of the interplay between MRI derived parameters using standardized volume-based methods in the healthy human brain gray matter (GM) and white matter (WM) has not been attempted to-date.

Since T2 relaxation time has been used as a marker of iron accumulation (Haacke et al., 2005; Bartzokis et al., 2007; House et al., 2008; Mitsumori et al., 2009), water content and myelination (Bartzokis et al., 2004; Dyakin et al., 2010) and given the hypothesis that radial diffusivity in white matter is dominantly affected by myelin integrity (Song et al., 2005; Drobyshevsky et al., 2005; Harsan et al., 2008; Ou et al., 2009) we attempted to investigate the interplay between T2 relaxation time and radial diffusivity in WM and GM, cortical and subcortical structures using standardized human brain atlas-based methods. First, we show the inter-regional or spatial heterogeneity relations for all segmented brain regions. Second, we investigate the relation using important benchmark structures such as the anterior and posterior corpus callosum (CC), putamen (PUT), hippocampus (HC), the cortical GM and WM of the motor and sensory lobes, respectively.

Materials and methods

Participants

This work has been approved by the local institutional review board of the University of Texas Health Science Center at Houston and is compliant with the Health Insurance Portability and Accountability Act (HIPAA). Also a written informed consent was obtained from each participant prior to data acquisition. All participants were identified as neurologically normal by review of medical history and were medically stable at the time of the assessments. All scans were read as “normal” by a board certified radiologist. The participants included 89 right-handed healthy adults with 43 males (age range = 18.7–57.6 years; mean age ± SD = 35.9 ± 10.7 years) and 46 females (age range = 19–56.9; mean age ± SD = 37.6 ± 10.0 years).

MRI Data Acquisition

We acquired whole-brain data using a Philips 3.0 T Intera system with a dual quasar gradient system with a maximum gradient amplitude of 80 mT/m, maximum slew rate of 200 mT/m per millisecond and an eight-channel SENSE compatible head coil (Philips Medical Systems, Best, Netherlands).

T1-weighted data

T1-weighted data acquired using 3D spoiled gradient-echo sequence (3D-SPGR) with a field-of-view (FOV) of 240 × 240 mm² and isotropic voxel size of 0.9375 mm.

T2-weighted data

T2-weighted data acquired using 2D dual spin-echo sequence with T_E1/T_E2/T_R=10/90/5000 ms, in the axial plane (3mm slice thickness, square FOV = 240×240 mm² at 44 axial sections).

Diffusion-weighted data

The diffusion-weighted data were acquired using a single-shot spin echo diffusion sensitized echo-planar imaging (EPI) sequence with the balanced Icosa21 encoding scheme which uses twenty-one diffusion gradient orientations (Hasan et al., 2001; Hasan and Narayana, 2003), a diffusion sensitization of b=1000 s.mm⁻², a repetition and echo times of T_R=6.1 s, T_E= 84 ms, respectively. EPI image distortion artifacts were reduced by using a SENSE acceleration factor or k-space under sampling of R of two. The slice thickness was 3 mm with 44 axial slices covering the whole-brain (foramen magnum to vertex), FOV =240 × 240 mm², and an image matrix of 256 × 256 that matched the 3D-SPGR and 2D conventional MRI dual spin-echo sequences described above. The number of non-diffusion weighted or b~0 magnitude image averages was 8; in addition, each encoding was repeated twice and magnitude-averaged to enhance the SNR (Hasan and Narayana, 2003; Hasan, 2006).

Processing

A pictorial flowchart of the computational pipeline for conventional and DTI data processing is illustrated in Fig. 1. The MRI data processing pipeline used in this work is described in more details elsewhere (Walimuni et al., 2011).

Figure 1.

A pictorial flow chart of the processing pipeline adopted in this work. In-house developed software used to process DWI and dual echo data. FreeSurfer and ANTs used for segmentation and registration, respectively.

Segmentation and Parcellation of T1-weighted Data

The T1-weighted brain data were automatically segmented into cortical and sub-cortical regions using FreeSurfer software library (Fischl et al., 2002). We used FreeSurfer in our work as it has been previously validated (Fjell et al., 2008; Jovicich et al., 2009) and has been widely used in normal brain development (Ostby et al., 2009) and aging research (Fjell et al., 2009) and in disease (Sailer et al., 2003; Salat et al., 2009; Ramasamy et al., 2009; Bigler et al., 2010). Using the Cortical and sub-cortical segmentations provided by FreeSurfer (Desikan, et al., 2006), an atlas consisting of 179 white matter (WM), gray matter (GM), and cerebrospinal fluid (CSF) regions was generated for each subject in the T1-weighted native space. A tabulation of the brain anatomical labels (non-CSF) provided by FreeSurfer and used in this work is provided in Table 1 and Supplementary Material.

Table 1.

Classification of brain cortical and lobar parcellations (see Desikan et al., 2006).

Frontal	Temporal	Parietal	Occipital	Cingulate
Middle Frontal Rostral middle frontal Caudal middle frontal	Entorhinal	Postcentral	Lingual	Rostral anterior division
Inferior Frontal Pars opercularis Pars triangularis Pars orbitalis	Parahippocampal	Supramarginal	Pericalcarine	Caudal anterior division
Orbitofrontal Medial orbitofrontal Lateral orbitofrontal	Temporal pole	Inferior parietal	Cuneus	Posterior cingulate
Preccentral	Fusiform	Superior parietal	Lateral occipital	Isthmus cingulate
Paracentral Lobule	Superior temporal	Precuneus		Rostral and caudal extents of Corpus Callosum
Superior frontal	Middle temporal
Frontal pole	Inferior temporal
	Transverse temporal
	Banks of superior temporal sulcus

Computation of Relaxation Time from Dual Spin Echo Data

The T₂ relaxation times were estimated from the early and late echoes (T_E1, T_E2) volumes according to standard spin-echo procedures assuming a single compartment model:

$$S_i=S_{0}exp\left(\frac{-T_{Ei}}{T_2}\right).$$ (1)

where S_i is the signal intensity of the ith echo; the T₂ relaxation time is obtained as (Hasan et al., 2008a; Aubert-Broche et al., 2009):

$$T_2=\frac{(T_{E2}-T_{E1})}{ln\left(\frac{S(T_{E1})}{S(T_{E2})}\right)}.$$ (2)

Diffusion-weighted Data and DTI Metrics

Diffusion-weighted volumes were intra-registered to the baseline “b0” image (the volume without diffusion sensitization) to correct for eddy-current image distortions using the Philips PRIDE workstation (Philips Medical Systems). Subsequently, all DWI volumes were masked using brain extraction tool (BET) in order to remove non brain regions (Smith, 2002). The DWI data were decoded and the tensors were diagonalized to obtain the three eigenvalues. The eigenvalues were subsequently used to compute FA, mean, axial and radial diffusivities (Hasan and Narayana 2006). A detailed description of the DTI data processing pipeline is provided elsewhere (Hasan, 2006; Hasan et al., 2011).

Fusion of T2 Relaxation Time and DTI Metrics in T1-weighted or Atlas space

The DTI-derived and T2 relaxation time volumes were registered to the T1-weighted data space using advanced normalization tools (ANTs) with symmetric normalization (Avants et al., 2008; Klein et al., 2009). The ANTs tool has also been independently tested and compared against 14 different non-linear registration programs (Klein et al., 2009). This registration tool has been reported to provide the highest Dice overlap according to our previous evaluations (Walimuni et al., 2011). Fractional anisotropy and T2-weighted volumes were chosen as objects for the registration and the corresponding transformations were applied to the radial diffusivity and T2 maps, respectively. A detailed description of the methods used in the registration is presented previously (Walimuni et al., 2011). Since the T1-weighted volume and the brain atlas are in the same space, the atlas was used as a look-up table to locate chosen brain structures from all the other image modalities registered to the T1-weighted space.

Statistical analysis

Regional group comparisons were conducted using analysis-of-variance (ANOVA) methods. All analyses of global and regional atlas-based T2 variation and λ_⊥ were conducted using a generalized linear model. The correlations between T2 and λ_⊥ of grouped and inter-regional structures were computed using the Pearson correlation coefficient (Zou et al., 2003). All statistical analyses were conducted MATLAB R12.1 Statistical Toolbox v 3.0 (The Mathworks, Natick, MA).

Results

The healthy men and women in our cohort did not differ in age (p = 0.43). Given previous reports of insignificant gender-based differences in brain volumetry (Fjell et al., 2009) and to simplify the analyses we averaged microstructural brain data obtained from age-matched men and women used in this study.

Normal Human Brain Microstructural Spatial Heterogeneity

Figure 2 illustrates the utility of the multimodal quantitative MRI methods adopted in this work and applied to the human brain using FreeSurfer anatomical atlas labels of deep and cortical gray and white matter (Desikan et al., 2006). The spatial heterogeneity of fractional anisotropy, mean diffusivity and axial diffusivity are shown Fig. 2A, 2B and 2C, respectively. The radial diffusivity, T2 relaxation time and their corresponding intra-regional Pearson correlation coefficient maps are shown in Fig. 3A, 2B and 3C, respectively. Note the expected trend that FA values in white matter are greater than the average values in gray matter (Pierpaoli et al., 1996). The cortical gray matter FA values are generally smaller than the values in deep gray matter (Fig. 2A). The average radial diffusivity values are smaller in white matter structures than in gray matter (Fig. 3A). The T2 values in cortical gray matter are larger than T2 values in deep gray matter. (Fig. 3B). The T2 relaxation time is smallest in iron-rich structures such as the globus pallidus and putamen. The T2 relaxation time is generally smaller in white matter than in gray matter (Fig. 3B). Note that the posterior corpus callosum has the largest FA, largest axial diffusivity and smallest radial diffusivity of any other structure in the human brain. The mean T2 relaxation time in the anterior CC is smaller than that in the posterior CC. The regional microstructural heterogeneity of the human brain white matter is clearly exemplified by the trends exhibited by the corpus callosum (see regional quantitative analyses below).

Figure 2.

Visual illustration of FreeSurfer brain atlas-based spatial heterogeneity of (A) fractional anisotropy, (B) mean diffusivity, and (C) axial diffusivity. The entire group average values along with the corresponding colorbar are shown in MNI space.

Figure 3.

Visual illustration of FreeSurfer brain atlas-based spatial heterogeneity of (A) radial diffusivity, (B) T2 relaxation time, and (C) the intra-regional Pearson correlation coefficient of T2 and radial diffusivity (p < 0.1).

Interregional Heterogeneity of Normal Human Brain Tissue

The scatter of mean T2 relaxation time and radial diffusivity values for all bilateral 82 segmented brain regions (Fig. 3C) is shown in Fig. 4 (see Table 1 and Supplementary Material). When all WM and all GM are grouped separately for the entire cohort, significant correlations between T2 and radial diffusivity were observed for GM (r = 0.543; p = 0.0003) and WM (r = 0.493; p = 0.0011).

Figure 4.

Atlas-based and volume averaged inter-regional scatter of radial diffusivity and corresponding T2 relaxation time for the 82 GM and WM structures (see Table 1 and Supplementary Material). Each point corresponds to group average over the entire cohort.

Intra-regional Dependence of T2 and Radial Diffusivity

We explored the intra-regional dependence of T2 and radial diffusivity using the entire cohort on selected brain structures (see Fig. 3C). Fig. 5 shows a scatter plot and corresponding linear regression analyses of T2 relaxation time vs. λ_⊥ for the anterior corpus callosum (CC) and posterior CC (Fig. 5A), the left hippocampus compared with left putamen (Fig. 45), left precentral GM (motor cortex) compared with left postcentral GM (sensory cortex), and left precentral WM (Fig. 5C) compared with left postcentral WM (Fig. 5D). Note that the average T2 values of anterior (genu) CC were smaller than the average values in the posterior (splenium) CC (p < 0.0001), the T2 values in the left hippocampus were much larger than the left putamen (p< 0.000001), the T2 values in motor cortex were not significantly different from those in the sensory cortex (p ~ 0.12), and the T2 mean values of left motor WM was larger than left sensory WM (p = 0.00001). The average radial diffusivity of the posterior CC was much larger than the anterior CC (p ~ 0.000000004), the average radial diffusivity of the hippocampus was much larger than the left putamen (p < 0000001).

Figure 5.

Intra-regional comparison of interplay between radial diffusivity and the T2 relaxation time for (A) anterior corpus callosum compared with posterior corpus callosum, (B) left hippocampus compared with left putamen, (C) left precentral GM compared with left postcentral GM, and (D) left precentral WM compared with left postcentral WM (see Figure 3C for the correlation coefficient map)

The scatter plots in Fig. 5 show also the Pearson correlation coefficient and its significance between T2 relaxation and for the anterior CC (r = 0.348; p < 0.001), posterior λ_⊥ CC (r = 0.25; p = 0.018) left hippocampus (r = 0.475; p = 0.0000026), left putamen (r=0.188; p = 0.077), left motor cortex (r = 0.279; p = 0.008), left sensory cortex (r = 0.249; p = 0.019), left motor WM (r = −0.116; p = 0.277) and left sensory WM (r = −0.143; p = 181). Note that the strongest correlation between T2 relaxation time and radial diffusivity is observed in the hippocampus (Fig 3C; Fig. 5B).

Discussion

To the best of our knowledge, this is the first human brain atlas-based report on the spatial heterogeneity and interplay between T2 relaxation and radial diffusivity on a large cohort of healthy adults. We presented qMRI maps of the subcortical, cortical gray matter, deep and lobar white matter. The average normative FA, axial, radial and mean diffusivity values in both gray matter and white matter are generally consistent with published reports on healthy adults (Pierpaoli et al., 1996).

In this work we investigated the interplay between T2 relaxation time and as these λ_⊥ indices are affected by the water spin environment (Beaulieu et al., 1998). The T2 relaxation time measured using spin-echo methods is speculated to be dependent on tissue microstructural composition such as water content, myelin (Whittall et al., 1997; Dyakin et al., 2010; Bartzokis et al., 2010) and iron deposition (Haacke et al., 2005; Hikita et al., 2005; Jara et al., 2006; Bartozkis et al., 2007), whereas radial diffusivity has been hypothesized to be a sensitive marker of myelination (Beaulieu et al., 2002; Drobyshevsky et al., 2005; Song et al., 2005).

In accordance with previous reports we found significant region-dependent variations in transverse T2 relaxation (Whittall et al., 1997; Georgiades et al., 2001; Bartzokis et al., 2004; Aubert-Broche et al., 2009; Hasan et al., 2010a) and DTI metrics (Pierpaoli et al., 1996; Lebel et al., 2008; Hasan et al., 2010b) in both GM and WM. Our qMRI findings (e.g. caudate vs. putamen; anterior vs. posterior; motor vs. sensory) may generally reflect region-dependent water and iron content (Hallgren and Sourander, 1958) in addition to myelination (Lebel et al., 2008; Hasan et al., 2010b).

In this work we sought to explore the interplay between radial diffusivity and T2 relaxation time as these metrics relate to physical variables. In two separate studies using a mouse model of dysmyelination the radial diffusivity (Song al., 2005) and T2 relaxation time (Dyakin et al., 2010) measurements of the corpus callosum were related to myelination. For correlation testing we assumed that the radial and axial diffusivities (Beaulieu et al., 2002) are independent, but some confounding effects due to fiber orientation have been reported recently (Wheeler-Kingshott and Cercignani, 2009). In this report we considered the radial diffusivity as a marker of inter-axonal and extracellular environment integrity (Song et al., 2005; Ou et al., 2009). The axial diffusivity may be affected by the intra-cellular or intra-axonal cytoskeletal microstructure (Kinoshita et al., 1999; Beaulieu, 2002) in addition to intra-voxel tortuosuity (Takahashi et al., 2000; Le Bihan et al., 2001).

For correlation testing we did not adjust for regional volumetry as tissue macrostructure and microstructural integrity are not expected to be generally unrelated (Fjell et al., 2008). Both tissue quantity (macrostructure) and quality (microstructure) depended on age and the age trends were used for quality assurance; the intra-regional qMRI correlations in this work did not alter when age was covaried. We also did not examine the interplay between T2 relaxation time and other metrics such as mean diffusivity and FA. Mean diffusivity is a superposition of radial and axial diffusivities (e.g. MD = (2*Radial Diffusivity + Axial Diffusivity)/3)) and FA is a function of the standard deviation of the tensor eigenvalues divided by MD that can be expressed as a ratio of axial and radial diffusivities (Hasan and Narayana 2003; Hasan and Narayana, 2006). A commensurate alteration in both axial and radial diffusivities would not be captured by FA (Mukherjee et al., 2002; Bar-Shir et al., 2009).

Consistent with previous reports using T2 relaxometry (Whittall et al., 1997) and DTI (Hasan et al., 2008b; Lebel et al., 2008; Hasan et al., 2009b; Lebel et al., 2010), our results on the anterior and posterior CC (Fig. 3A) show a significant heterogeneity between the anterior and posterior sectors of the corpus callosum. This observation may be attributed to axonal caliber, axonal orientation, axonal packing and myelin distribution. Our data show that in the anterior λ_⊥ CC is significantly larger than that in posterior CC predicting that the anterior CC is populated with lightly-myelinated fibers compared to the posterior CC. This finding is consistent with histopathological data on the human CC (Aboitiz et al., 1992) and in vivo measurements using myelin water volume mapping methods (Whittall et al., 1997; Laule et al., 2004).

Consistent with previous reports on the regional heterogeneity of the corpus callosum using T2 relaxation time measurements (Whittall et al., 1997; Bartzokis et al., 2004; Kim et al., 2007; Levesque et al., 2010), our data indicate that T2 relaxation time values of the anterior CC are significantly smaller than that of the posterior CC (Fig. 3A). In view of the DTI findings discussed above, this is an unexpected result as the posterior CC contains more myelin than the frontal CC (Aboitiz et al., 1992). This paradoxical result indicates that other biophysical factors in addition to myelin may contribute to the measured T2 relaxation time. Potential contributors to T2 relaxation time measurements in unmyelinated axons may include axonal packing and fiber orientation (Denk et al., 2011). Published normative magnetization transfer ratio (Vavasour et al., 1998) is also paradoxically greater in the less-meyelinated genu of CC compared to the splenium. These observations indicate that axonal membranes or adjacent macromolecular layer (Beaulieu et al., 1998; Sled et al., 2004) with short relaxation time may contribute to the reduced relaxation time in the more packed callosal genu via fast exchange mechanisms (Minty et al., 2009, Dula et al., 2010). A relatively larger iron content in the frontal CC than the posterior CC would offer another plausible explanation to the finding on the CC. To the best of our knowledge, there has been no comprehensive histopathological report (Yao et al., 2009) on the iron content of the different subdivisions of the human corpus callosum. Moreover, a recent study did not report strong association between iron content and relaxation rate in compact white matter (Li et al., 2009). Higher iron content in the genu of the CC compared to the splenium CC has been extrapolated by Bartzokis et al. (2007) using the postmortem data in Hallgren and Sourander (1958).

Our finding of strong correlation between hippocampal T2 and radial diffusivity may be related to higher water content and reduced iron in this region (see Haacke et al., 2005). A reduction in obstacles or barriers to random translational diffusion as a result reduced myelin or increased water content in the extracellular space would increase the diffusion rate and hence would allow water spins to average out the local magnetic field gradients and hence increase the relaxation time (Weisskoff et al., 2004). If water spins experience such scenario then a strong relation between T2 relaxation and diffusivity may be postulated on theoretical grounds (Baratti et al., 1999; Jara et al., 2006).

Despite the use of validated tissue segmentation (Fischl et al., 2010) and advanced multimodal MRI registration methods (Klein et al., 2009), our study has some limitations. Our normative data were collected to help interpret data collected from patients and our whole brain MRI data acquisition protocol was limited to ~ 30 minutes. Due to scan time considerations, the DTI data were acquired using echo-planar imaging with intrinsic voxel size of ~ 2mm × 2mm × 3mm, while the dual spin echo spatial resolution was ~ 1 mm × 1 mm × 3 mm. The DTI-derived and T2 relaxation time maps were registered to the T1-weighted map with spatial resolution of ~ 1mm × 1mm × 1mm. Therefore, partial volume averaging of CSF with brain parenchyma may have affected the estimated metrics in some regions. Our data quality assurance tests (Walimuni et al., 2011), visual inspection and quantitative analyses on the deep and cortical (e.g. volumetry vs. age) indicate minimal contamination with CSF. In this work we adopted a single compartment model for both DTI and relaxation time measurements in both gray and white matter. More sophisticated models for white matter have been proposed (Lancaster et al., 2003). We have ignored the possible contribution from short-lived compartments such as water trapped in myelin (Madler et al., 2008; Mackay et al., 2009) or the fast-exchanging water layer adjacent to cellular or axonal membranes (Sled et al., 2004). Optimized three-dimensional high resolution multi-echo relaxometry and myelin mapping methods may provide more specific metrics to model these compartments (Deoni et al., 2011). Due to the lack of histological data, we could not relate our DTI and relaxometry measurements in gray matter to neuronal size or to iron content directly. Our DTI and relaxation time measurements in white matter could not be related to myelin or axonal size distribution, axonal packing and axonal cytoskelatal microstructure (Pierpaoli et al., 1996; Paus et al., 2010). Recent attempts to relate diffusion metrics in white matter to axonal size are promising (Assaf et al., 2008; Barazany et al., 2009; Alexander et al., 2010).

Additional histopathological and in situ standardized quantitative MRI measurements of key human brain structures such as amgydala, hippocampus, motor and sensory cortices, anterior and posterior CC are needed to help isolate or model the in vivo contributors to MRI microstructural metrics (Beaulieu, 2002). In this work we used a large cohort of right-handed healthy controls with a wide age range, but we did not examine the interaction with age, gender and side. A future extension of this work will use a larger healthy cohort and will stratify the population by age and gender to examine more thoroughly the interplay between qMRI metrics and side. In conclusion, our results underscore the utility of using standardized multimodal and atlas-based approaches and the need for additional statistical models to understand the yet unresolved biophysical contributors to MRI signal.