
==== Front
9215515
20498
Neuroimage
Neuroimage
NeuroImage
1053-8119
1095-9572

39032791
10.1016/j.neuroimage.2024.120734
nihpa2016101
Article
High-resolution diffusion magnetic resonance imaging and spatial-transcriptomic in developing mouse brain
Han Xinyue ab
Maharjan Surendra a
Chen Jie a
Zhao Yi c
Qi Yi d
White Leonard E. e
Johnson G. Allan df
Wang Nian abg*
a Department of Radiology and Imaging Sciences, Indiana University, Indianapolis, IN, USA
b Advanced Imaging Research Center, University of Texas Southwestern Medical Center, Dallas, TX, USA
c Department of Biostatistics and Health Data Science, Indiana University, Indianapolis, IN, USA
d Center for In Vivo Microscopy, Department of Radiology, Duke University, Durham, NC, USA
e Department of Neurology, Duke University Medical Center, Durham, NC, USA
f Department of Biomedical Engineering, Duke University, Durham, NC, USA
g Stark Neurosciences Research Institute, Indiana University, Indianapolis, IN, USA
* Corresponding author. nianwang@iu.edu (N. Wang).
30 8 2024
15 8 2024
20 7 2024
06 9 2024
297 120734120734
https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Brain development is a highly complex process regulated by numerous genes at the molecular and cellular levels. Brain tissue exhibits serial microstructural changes during the development process. High-resolution diffusion magnetic resonance imaging (dMRI) affords a unique opportunity to probe these changes in the developing brain non-destructively. In this study, we acquired multi-shell dMRI datasets at 32 μm isotropic resolution to investigate the tissue microstructure alterations, which we believe to be the highest spatial resolution dMRI datasets obtained for postnatal mouse brains. We adapted the Allen Developing Mouse Brain Atlas (ADMBA) to integrate quantitative MRI metrics and spatial transcriptomics. Diffusion tensor imaging (DTI), diffusion kurtosis imaging (DKI), and neurite orientation dispersion and density imaging (NODDI) metrics were used to quantify brain development at different postnatal days. We demonstrated that the differential evolutions of fiber orientation distributions contribute to the distinct development patterns in white matter (WM) and gray matter (GM). Furthermore, the genes enriched in the nervous system that regulate brain structure and function were expressed in spatial correlation with age-matched dMRI. This study is the first one providing high-resolution dMRI, including DTI, DKI, and NODDI models, to trace mouse brain microstructural changes in WM and GM during postnatal development. This study also highlighted the genotype-phenotype correlation of spatial transcriptomics and dMRI, which may improve our understanding of brain microstructure changes at the molecular level.

Diffusion magnetic resonance imaging
DTI
DKI
NODDI
Spatial transcriptomics
Brain development
==== Body
pmc1. Introduction

Diffusion magnetic resonance imaging (dMRI) is a powerful imaging technique for characterizing detailed neuroanatomy changes during brain development. It is non-destructive and can quantitatively characterize brain microstructure (Lerch et al., 2017; Ge, 2006) using statistical (Basser et al., 1994; Jensen et al., 2005) and biophysical models (Zhang et al., 2012). dMRI has been applied in assessing various neurological diseases, including Alzheimer’s disease (Bozzali et al., 2002; Mielke et al., 2009; Stebbins and Murphy, 2009), Parkinson’s disease (Prodoehl et al., 2013; Yoshikawa et al., 2004; Vaillancourt et al., 2009), and amyotrophic lateral sclerosis (Toosy et al., 2003; Cosottini et al., 2005; Graham et al., 2004). In particular, dMRI has been widely used in evaluating white matter (WM) related disorders (Raja et al., 2019; Horsfield and Jones, 2002; Silk et al., 2009; Szeszko et al., 2005). In WM structures, changes in fiber orientation patterns, such as the progressive formation of axon fascicles in compact fiber bundles and the development of myelin sheathes, will result in corresponding changes in the behavior of water diffusion that can be detected by dMRI (Le Bihan, 2003).

Diffusion tensor imaging (DTI) uses the diffusion tensor model to extract quantitatively diffusion metrics, and these metrics are associated with the physiological and biological properties of brain tissue (Le Bihan, 2003; Andica et al., 2020; Alexander et al., 2007). DTI has been successfully applied in studying developing brains, including mouse (Mori et al., 2001), rat (Bockhorst et al., 2008), and human (Hüppi and Dubois, 2006; Mukherjee and McKinstry, 2006). Compared to DTI, neurite orientation dispersion and density imaging (NODDI) uses advanced modeling techniques to characterize tissue properties more specifically (Zhang et al., 2012). NODDI parameterizes the dMRI signals as biophysical parameters, including neurite density index (NDI) describing the density of neurites, orientation dispersion index (ODI) referring to the dispersion of neurites, and isotropic volume fraction (ISO) characterizing the volume fraction of isotropic diffusion (Zhang et al., 2012; Andica et al., 2020). NODDI has been used in developing human brain studies (Zhao et al., 2021; Genc et al., 2017; Mah et al., 2017). dMRI has been used to document the developmental processes of gray matter and white matter structures in the mouse brain. DTI can discretely delineate the microstructure of embryonic and early postnatal mouse brains with improved contrast from structural MRI (Mori et al., 2001; Kronman et al., 2023; Chuang et al., 2011). Fractional anisotropy (FA) is a standard DTI metric exhibiting correlations to mouse brain development. For instance, a sharp drop in FA was observed in the cerebral cortex in early postnatal development (Baloch et al., 2009), corresponding with dendrite formation and increased synaptic density (Bayer and Altman, 1991). The hippocampus also shows a decreasing FA in the first 15 postnatal days (Baloch et al., 2009). In contrast, WM exhibits elevated FA due to tightly fasciculated axonal fibers, increased myelination, and enhanced brain connectivity (Le Bihan, 2003), and has relatively steady FA after postnatal day 30 due to the maturity of brain development (McKinstry et al., 2002). DTI has also been used to build developing mouse brain atlases by incorporating spatiotemporal gene expression profiles in the embryonic brain (Wu et al., 2022) and the adult brain (Kronman et al., 2023).

High-resolution dMRI, which has reached below 50 μm isotopically, has been achieved by technical advances, including higher magnetic fields, stronger gradients, and higher angular resolution (Aggarwal et al., 2015; Wang et al., 2020). Compressed sensing (CS), a method for accelerating imaging acquisition (Lustig et al., 2007), can further improve the spatial resolution to 25 μm and reduce the acquisition time up to 8 times (Wang et al., 2020; Wang et al., 2018). The improved spatial resolution provides promising capability to investigate brain cytoarchitecture and validate MRI findings with conventional histology at similar spatial resolution. The Allen Mouse Brain Common Coordinate Framework (CCFv3) is a 3D spatial template constructed by serial two-photon tomography (STP) at high in-plane resolution (Wang et al., 2020). CCFv3 can be used to construct a 3D whole mouse brain ontology and to compare multimodal images (Wang et al., 2020; MacKenzie-Graham et al., 2004; MacKenzie-Graham et al., 2003). Recently, we have successfully integrated dMRI maps into CCFv3 to correlate MRI metrics with STP and Nissl staining from the Allen Mouse Brain Atlas (AMBA) (Wang et al., 2023).

Genes with spatial profiles of various expression levels are associated with neural phenotypic variations in brain structure and function (Arnatkevičūtė et al., 2019; Fornito et al., 2019; Lein et al., 2017; Lein et al., 2007). Several genome-wide studies have recently used spatial transcriptomics in the developing brain. La Manno et al. (2021) reported a comprehensive single-cell transcriptomic atlas of embryonic mouse brains until birth and identified key developmental genes. Rosenberg et al. (2018) profiled and spatially mapped >100,000 single-cell transcriptomes on the developing mouse brain and spinal cord. Genomics and transcriptomics have opened many new research possibilities when combined with imaging techniques. Integrating spatial transcriptomics from AMBA with quantitative MRI parameters has seen encouraging success in investigating brain structures and functions (Patel et al., 2020; Wen et al., 2018; Edwards et al., 2023; Vértes et al., 2016). For developing the brain specifically, Allen Developing Mouse Brain Atlas (ADMBA) generated spatial and temporal profiles of ~2000 genes in embryonic and postnatal mouse brains, which provided the genoarchitecture of brain development at the cellular level (Thompson et al., 2014).

In this study, we acquired high-resolution (32 μm isotropic) ex vivo mouse brain diffusion MRI images at different postnatal days. We characterized the microstructural changes through the whole brain using DTI, DKI, NODDI models. With reference to the genoarchitectural patterns available in the ADMBA, we examined the spatiotemporal gene expression profiles at P4 and P14. We correlated the gene expression data with the age-matched dMRI (by DTI, NODDI, and diffusion kurtosis imaging (DKI) (Jensen et al., 2005) models) at P4 and P14 using a partial least squares (PLS) regression model. We further identified the most over-expressed genes associated with dMRI parameters using gene ontology (GO) enrichment analysis. Compared to prior studies on developing mouse brains using dMRI, which were conducted at lower spatial resolutions and only employed basic diffusion tensor models (Mori et al., 2001; Kronman et al., 2023; Chuang et al., 2011; Zhang et al., 2003), this study utilizes the DTI and advanced models (DKI and NODDI) with improved spatial resolution. The existing investigations on gene-MRI associations focus on structural MRI (Diehn et al., 2008), quantitative MRI (qMRI) (Edwards et al., 2023), and functional MRI (fMRI) (Vértes et al., 2016) on mature brains rather than developing brains. This study contributes to our understanding of brain development by providing high-resolution brain microstructural visualizations from postnatal day 1. It provides the opportunity to investigate genetic contributions to neurodevelopmental disorders using non-destructive imaging techniques.

2. Methods and materials

2.1. Animal study

Twenty-four wild-type C57BL/6 J mice (Jackson Laboratory, Bar Harbor, ME) at postnatal days 1, 4, 7, 10, 14, 21, 60, and 360 (N = 3 at each timing point) were chosen for MR imaging. Animal experiments were carried out in compliance with the Duke University Institutional Animal Care and Use Committee. Animals were sacrificed and perfusion-fixed with a 1:10 mixture of ProHance-buffered (Bracco Diagnostics, Princeton, NJ) formalin. Specimens were immersed in buffered formalin for 24 h and then moved to a 1:200 solution of ProHance/saline to shorten the T1 (to about 100 ms) and reduce scan time. The choice for ex vivo instead of in vivo study is to get higher anatomic resolution and better tissue contrast for improved performance in detecting microstructural changes (To et al., 2023; Holmes et al., 2017; Ma et al., 2019) and to be consistent with spatial transcriptomics acquired ex vivo.

2.2. MRI acquisition

All the MRI experiments were performed on a 9.4T Oxford 8.9-cm vertical bore magnet (Oxford Instruments, Abingdon, United Kingdom) and the maximum gradient amplitude on each axis was 2000 mT/m. A modified three-dimensional (3D) diffusion-weighted spin-echo pulse sequence with k-space undersampling was used in this study. The readout dimension was fully sampled and phase encoding dimensions were under-sampled using a sparsifying approach, described in detail previously (Wang et al., 2018). An acceleration factor (AF) of 4.0 was used in this study, where 1.0 stands for the fully sampled data. For multi-shell dMRI acquisition, the diffusion gradient directions were determined using the method proposed by Koay et al. (2012) to ensure uniformity within each shell while maintaining uniformity across all shells.

The dMRI under sampled datasets were acquired with four shells (b value = 1000, 4000, 6000, 8000 s/mm (Ge, 2006)) at 32 μm isotropic resolution with one repetition. The total scan time was approximately 48 h with 64 diffusion-weighted images (DWIs) and 4 b0 images. The temperature (~ 20 °C) was monitored throughout all the scans and fluctuated less than 1 °C. The FOV is 18.88 mm × 11.52 mm × 11.52 mm, the matrix size was 590 × 360 × 360, the repetition time (TR) was 100 ms, the time to echo (TE) was 16 ms, and the bandwidth (BW) was 156 kHz. The gradient separation time was 7.4 ms, the diffusion gradient duration time was 4.8 ms, and the maximum gradient amplitude was about 140 mT/m.

2.3. MRI reconstruction and analysis

2.3.1. Compressed sensing

The under-sampled k-space data reconstruction has been described in previous studies (Lustig et al., 2007; Wang et al., 2018; Hollingsworth, 2015) by minimizing the following function: (1) f(x)=∥Fx−y∥22+λ1∥Ψx∥1+λ2TV(x)

Where x is the image and y is its corresponding k-space, F is the FFT, Ψ is the sparse transform, λ1 and λ2 are regularization parameters, and TV is the total variation. In this study, λ1 equals 0.006 for the sparse solution and λ2 equals 0.0012 for the data consistency with Daubechies wavelet transform performed (Lustig et al., 2007,61). An acceleration factor of 4.0 means the acquisition time is 1/4th of the time required for a fully sampled dataset. The total scan time of the multi-shell dMRI datasets was approximately 48 h, the scan time for the fully sampled dataset would be 8 days. The reconstruction was implemented on a Dell Cluster where an initial Fourier transform along the readout axis yielded multiple 2D images from each acquisition that could be spread across the cluster for parallel reconstruction.

2.3.2. DTI metrics extraction

The individual DWIs were registered to the baseline (b0) image using Advanced Normalization Tools (ANTs). Both DTI and Generalized Q-sampling imaging (GQI) reconstruction methods were performed for reconstruction using DSI studio (Yeh et al., 2010; Yeh et al., 2013). From data with b-values of 1000 to 4000 s/mm (Ge, 2006), the diffusion tensor was calculated. The scalar indices include fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD).

2.3.3. NODDI metrics extraction

The NODDI toolbox for Matlab was used to fit the diffusion MRI data (Zhang et al., 2012). The default value of isotropic diffusivity of 2.0 × 10−3 mm (Ge, 2006)/s (Holz et al., 2000) and intrinsic diffusivity of 0.9 × 10−3 mm (Ge, 2006)/s according to our previous findings (Wang et al., 2019) was used for the Diso and Din values, respectively. For the ex vivo NODDI model, a relatively small dot-compartment (the diffusion is restricted in all directions) was included for the fitting (Dhital et al., 2018). The parameters derived from the NODDI model were the orientation dispersion index (ODI), neurite density index (NDI), and isotropic volume fraction (Viso).

2.3.4. DKI metrics extraction

Diffusion Imaging in Python (Dipy) (Garyfallidis et al., 2014) was used to compute DKI metrics. 3D Gaussian smoothing (Gaussian kernel fwhm = 1.25) was used before fitting to DKI model (Jensen et al., 2005). The DKI parameters generated were axial kurtosis (AK), kurtosis fractional anisotropy (KFA), mean kurtosis (MK), and radial kurtosis (RK).

2.3.5. 3D histology and image registration

The ADMBA’s 2D in situ hybridization (ISH) images from Nissl staining, 3D Nissl staining images, and their labels were downloaded from the ADMBA website (https://developingmouse.brain-map.org/) and converted to NIFTI format (Wang et al., 2020). Each 2D label was registered to the common coordinates of a 3D reference model according to the ADMBA documentation (Allen Institute for Brain Science 2010). The diffusion MRI images were registered to the ADMBA using ANTs automated image registration, described in previous studies in detail (Avants et al., 2011; Calabrese et al., 2015). The labels were transformed back to MRI space and the quantitative MRI measurements from each ROI were then extracted. Each dMRI metric was averaged across all animals for further visualization and analysis.

2.4. Genetic data pre-processing

A total of 2002 genes were selected as genes of interest. The selection criteria are based on known roles in brain development and known robust neuroanatomical markers during brain development. In detail, there were five categories of genes (Liscovitch and Chechik, 2013) involving: (1) Transcription factors, including homeobox, basic helix-loop-helix, forkhead, nuclear receptor, high mobility group, and POU domain genes. (2) Neuropeptides, neurotransmitters, and their receptors. (3) Neuroanatomical marker genes. (4) Genes relevant to brain development including axon guidance, receptor tyrosine kinases, and their ligands. (5) Genes of general interest including common drug targets, ion channels, cell adhesion molecules, genes involved in neurotransmission, G-protein-coupled receptors, and genes implicated in neurodevelopmental diseases.

Selection of brain regions-of-interest (ROI) was based on the 11 anatomic regions according to ADMBA reference atlases (Allen Institute for Brain Science 2013), including retrosplenial cortex (RSP), telencephalic vesicle (Tel), hypothalamus (HYP), prosomere 3 (p3), prosomere 2 (p2), prosomere 1 (p1), midbrain (M), prepontine hindbrain (PPH), pontine hindbrain (PH), pontomedullary hindbrain (PMH), and medullary hindbrain (MH). 2002 gene expression levels for each of the 11 brain ROIs were imported from ADMBA. Particularly, the expression level for each brain ROI was computed as the sum of expressing pixel intensities in this ROI divided by the total number of pixels that intersect this ROI from ISH images (Allen Institute for Brain Science 2010). This study only included gene expression data from P4 and P14, corresponding to the availability of ADMBA and aged-matched dMRI images. The genetic data was then organized as two 2D matrices of size 11 × 2002 (11 ROIs at ages P4 or P14 × 2002 genes).

2.5. Regression model and statistical tests

We used partial least squares (PLS) regression models to examine the association between gene expression profiles and age-matched dMRI across ROIs at P4 and P14 individually, using in-house built MATLAB program. PLS regression model means to project the predictor variables and response variables to a new space and perform least square regression on the new space. It is a powerful model when there are more predictor variables than the sample size, or predictors are highly multicollinear. In this study, we have a greater number of predictor variables (i.e., genes of interest) than the sample size (i.e., regions of interest), and predictor variables are multicollinear (by the nature of genes).

In each of the two regression models, predictor variables were composed of the genetic data sized 11 × 2002 from Section 2.4, and response variables were composed of age-matched dMRI data sized 11 × 10, including 4 DTI metrics (i.e., AD, FA, MD, and RD), 4 DKI metrics (i.e., AK, KFA, MK, and RK), and 2 NODDI metrics (i.e., NDI and ODI), where the 11 rows in both variables represented the averaged data over the 3 animals of the 11 brain ROIs. During PLS regression, firstly, both the dMRI data and the genetic data were projected into PLS components that explain the maximum covariance between the two. Then, the first two PLS components were selected for balancing data size and preserving data variation. Finally, linear regression was performed on PLS components.

We conducted a residual analysis to validate the proposed PLS model, including normality and homoscedasticity checks for each raw residual from 10 response variables. For the normality check, we performed the Shapiro-Wilk test at a 5 % significance level and created normal probability plots comparing the residuals’ distribution to the normal distribution. For the homoscedasticity check, we performed the Breusch-Pagan test at a 5 % significance level. To test the performance of the PLS model, we used Pearson’s correlation test to examine the pairwise correlation between PLS components and dMRI parameters. A significance level of p-value < 0.05 was used.

2.6. Gene ontology (GO) enrichment analysis

We used GO enrichment analysis to identify the GO terms that were largely enriched in association with dMRI. The list of genes was from 2002 genes-of-interest and was ranked according to the averaged variable importance in projection (VIP) scores (Chong and Jun, 2005) derived from the two PLS models. We used GOrilla (https://cbl-gorilla.cs.technion.ac.il/, version August 2023) (Eden et al., 2009; Eden et al., 2007) as a GO enrichment analysis tool for calculating the GO terms, p-values, and enrichment. Three GO aspects were covered in the analysis: “biological process”, “cellular component”, and “molecular function”. We then filtered the resulting list of GO terms from Gorilla by thresholding p-values to be lower than 10−3.

The filtered list of GO terms was then visualized as bubble plots, with the axes indicating the semantic similarity between GO terms and the bubble sizes and color indicating the enrichment of GO terms. The visualization was done by the online tool Revigo (http://revigo.irb.hr/, version 1.8.1) (Supek et al., 2011). The GO terms that ranked the top under each GO aspect were selected for further interpretation.

The data that support the findings of this study and corresponding codes are available from the Github repository: https://github.com/selinahxy/developingmousebraindmri.

3. Results

3.1. Brain microstructure changes during development by dMRI

Fig. 1 shows the developing mouse brain’s brain volume and qualitative dMRI maps (b0, DWI, and diffusion-encoded-color (DEC)). Both body weight and brain volume increase dramatically during the development stage (Fig. 1A and Supplementary Fig. 1). While b0 images exhibit little image contrast between the corpus callosum (cc, red arrows) and the cerebral cortex before P14 (Fig. 1B), DWI and DEC images already developed high contrast in the cc even at postnatal day 1 (Fig. 1C–D). Corpus callosum starts to be visualized with lower intensity in the b0 image only after postnatal day 14, probably due to the rapid myelination process at this age. Similarly, the anterior commissure (ac) can be visualized earlier in DWI (Supplementary Fig. 2B) and DEC images (Supplementary Fig. 2C) than in b0 images.

Fig. 2 shows the quantitative dMRI maps of the developing mouse brain derived from both DTI and NODDI models. The tissue microstructure changes are evident throughout different brain areas. Different parameters exhibit different trends during brain development. MD and NDI images show opposite contrast, as do FA and ODI images, regardless of the developmental stage of the brain. Overall, there is an increasing trend for FA in WM bundles (red and yellow arrows) and a decreasing trend for the ODI as the brain develops. In contrast, the cerebral cortex (white arrows) shows a decreasing trend for FA and an increasing trend for the ODI. WM bundles have already developed high contrast in MD and NDI maps even at postnatal day 1. The pyramidal cell layer and granule cell layer of the hippocampus also showed high contrast in MD and NDI images as early as postnatal day 1. Moreover, the cerebellum (green arrows) shows high contrast at P14, probably due to myelination achieved at this age (Sanchez-Molina et al., 2020; Peris et al., 2023).

The 32 μm high-resolution dMRI can visualize brain microstructures, including the isocortex and hippocampus. In detail, the 32 μm dMRI showed layered-structure isocortex (Supplementary Fig. 3 row “32 μm”, white boxes) and hippocampus (Supplementary Fig. 3 row “32 μm”, yellow boxes). In contrast, lower resolution imaging (Supplementary Fig. 3 row “64 μm” and “128 μm”) failed to display these structures.

3.2. WM and GM show different development patterns

Fig. 3 illustrates the ODI maps at different regions of the developing mouse brain, including the anterior commissure, corpus callosum, isocortex, hippocampus, and cerebellum. Fig. 4 shows the dMRI parameters as a function of age. As shown in these two figures, different developmental patterns are demonstrated in WM bundles and the cerebral cortex. Overall, WM regions, including ac and cc, show decreasing ODI (Fig. 3A–B, Fig. 4D, and Supplementary Fig. 4), increasing FA (Fig. 4C), increasing MD (Supplementary Fig. 5B) and decreasing NDI (Supplementary Fig. 5C). In contrast, GM regions, including isocortex, have increasing ODI (Fig. 3C, Fig. 4H in red line, and Supplementary Fig. 4 in white arrows), decreasing FA (Fig. 4G in red line), increasing MD (Supplementary Fig. 5E in red line) and decreasing NDI (Supplementary Fig. 5F in red line). As for regions with mixture structures like hippocampus and cerebellum, for which the dMRI signals were averaged from WM and GM, the dMRI parameters do not exhibit a monotonous trend as discovered in WM bundles and the cerebral cortex. As shown in Fig. 3 and Fig. 4, the ODI of the hippocampus (Fig. 3D and Fig. 4H in blue line) demonstrates an initial increasing trend and then shows a decreasing trend, ODI of the cerebellum (Fig. 3E and Fig. 4H in green line) shows a decreasing trend and then shows an increasing trend. Similar variations in FA, MD, and NDI of these two regions (in blue and green lines, respectively) were demonstrated in Fig. 4G and Supplementary Fig. 5E–F, respectively. On the other hand, the volume of different brain areas gradually increases with age, especially the cc, ac, and isocortex, showing 4.5 %, 8.3 %, and 1.6 % volume increase, respectively, from P14 to P21. However, more significant changes are observed by dMRI metrics, including FA (24.1 % in cc, 26.6 % in ac, and 5.2 % in isocortex), ODI (− 32.2 % in cc, − 44.1 % in ac, and − 3.3 % in isocortex).

To further explore the different development patterns through different brain areas, the fiber orientation distributions (FODs) in the corpus callosum (Fig. 5B) and isocortex (Fig. 5C) were depicted during brain development (P1, P7, and P60). The center part of cc has predominantly medial-lateral (red color) fiber orientation in P1, P7, and P60 (Fig. 5A, yellow box). However, compared to P60, there are more crossing fibers at P1 (Fig. 5B). This suggests that the directions of fiber bundles in cc become more uniformly distributed, and therefore result in increasing FA (Fig. 4C in red line) and decreasing ODI (Fig. 3B and Fig. 4D in red line). In contrast, the FODs in the isocortex become more complex during brain development, i.e., more crossing fibers are exhibited at P60 than at P1 (Fig. 5C). This change leads to decreased FA (Fig. 4G in red line) and increased ODI (Fig. 3C and Fig. 4H in red line) in isocortex.

3.3. Associations between dMRI metrics and gene expression

To understand mouse brain development at the gene expression level, the dMRI metrics at P4 and P14 were registered to age-matched ADMBA to compare with the spatial transcriptomics. The labels from the ADBMA (Fig. 6A, C) were then transformed back to MRI space for ROI-based analysis (Fig 6B, D).

The PLS regression models at P4 and P14 have been demonstrated to be a good fit for characterizing the associations between dMRI metrics and gene expression. The residuals all have low absolute values (Supplementary Fig. 6A and Supplementary Fig. 7A), and were all homoscedastic and mostly normally distributed (Supplementary Fig. 6B, Supplementary Fig. 7B, and Supplementary Table 1). We choose the first two PLS components (i.e., PLS1 and PLS2) for further analysis. Because the first two PLS components account for most of the variance in response variables (90 % for PLS model at P4, and 85 % for PLS model at P14), and cross-validation analysis confirmed that after two PLS components mean squared error (MSE) became stable (Supplementary Fig. 6C and Supplementary Fig. 7C).

Fig. 7 demonstrated that the gene expression profiles were associated with dMRI parameters spatially. Specifically, PLS components (For P4, PLS1 explains 80.2 % covariance of response variables, PLS2 explains 9.6 % covariance of response variables; For P14, PLS1 explains 72.9 % covariance of response variables, PLS2 explains 11.9 % covariance of response variables) from genes-of-interest were associated with spatial alternations of DTI, DKI, and NODDI metrics. In detail, at both P4 and P14 and from midbrain (M, red arrows) to prepontine hindbrain (PPH, yellow arrows), it showed increased DTI metrics (Fig. 7A), as well as decreased DKI metrics (AK and MK) (Fig. 7B) and NODDI metrics (Fig. 7C). Meanwhile it had decreased PLS1 and PLS2 values (Fig. 7D).

PLS components were negatively correlated with all DTI metrics and positively correlated with DKI metrics (AK and MK) and all NODDI metrics, as shown in Fig. 7E. The Pearson’s correlation r-values of both PLS1 and PLS2 (not shown) were negative for all DTI metrics and were positive for DKI metric (AK and MK) and all NODDI metrics. It can then be concluded that the expression levels of the genes of interest have a negative correlation with DTI (AD, FA, MD, and RD), and a positive correlation with DKI (AK and MK) and NODDI (NDI and ODI).

3.4. Genes related to nervous system structure and function are associated with dMRI

To identify which gene categories are associated with dMRI parameters, we have conducted GO enrichment analysis on the 2002 genes of interest that have been ranked by their VIP scores from the PLS model. The results of GO enrichment analysis were visualized in Fig. 8. Among the three GO aspects—biological process (Fig. 8, left), cellular component (Fig. 8, center), and molecular function (Fig. 8, right), the biological process has the largest number of overly expressed GO terms. Under the biological process, the most enriched GO terms were associated with signal transport and neuron communication; under the cellular component, the most enriched GO term was centriole; under molecular function, the most enriched GO terms were related to the construction of neuronal elements. In summary, the GO terms that were overly expressed from genes of interest were closely related to nervous system structure and function.

4. Discussions

dMRI is a promising imaging technique for characterizing microstructure changes in the developing brain with three main advantages. Firstly, high-resolution dMRI boasts impressive spatial resolution for brain neuroanatomy. dMRI can achieve isotropic resolutions of 100 μm for in vivo imaging (Wu and Zhang, 2016) and resolution greater than 25 μm for ex vivo imaging (Wang et al., 2020; Johnson et al., 2023). However, tissue physiology changes in ex vivo imaging compared to in vivo imaging on living brains, which leads to overestimated FA and underestimated MD, NDI, and ODI (To et al., 2023). Secondly, compared to the conventional histological methods, dMRI offers non-destructive capabilities, enables whole-brain imaging while maintaining tissue integrity, and facilitates the construction of volumetric scans (Annese, 2012). Thirdly, dMRI can provide quantitative measurements by fitting with various biophysical models. For instance, DTI can provide valuable insights into the direction of water movement and fiber orientation, while biophysical models derived from NODDI can elucidate specific developmental changes and/or neuropathologies within brain microstructures (Zhang et al., 2012; Colgan et al., 2016). Combining the outputs of all assessable quantitative biophysical models makes it possible to gain profound insights into the morphology and pathophysiology of brain tissue.

By performing high resolution multi-shell dMRI with different modeling, we observed that WM and GM showed opposite changes in dMRI parameters, corresponding to the different development patterns. For WM, axonal fiber bundles increase remarkably in anisotropic water diffusion, as they go through myelination, increased axon diameter, increased compaction in parallel fascicles, and axon addition and pruning as development progresses (Neil et al., 1998; Baratti et al., 1999; Takahashi et al., 2000; Stiles and Jernigan, 2010; Dubois et al., 2014). For GM, the neuronal and glial cell morphologies have increasing dispersion in neuritic process orientations (dendritic and axonal arborizations for neurons and protoplasmic extensions for glial cells), and thus, they showed decreased water diffusion anisotropy (Kroenke et al., 2009; Kroenke, 2018). These differences in development patterns can be reflected by dMRI parameters, as shown in Fig. 3 and Fig. 4.

We also noticed opposite changes in FA versus ODI and MD versus NDI. FA describes the anisotropy of water diffusion direction. A high FA value means a restricted, directed diffusion, and a low FA means a more isotropic diffusion. ODI characterizes angular variation of neurites (Zhang et al., 2012). A high ODI means a multidirectional dendritic structure, and a low ODI means a highly compacted and parallel structure (e.g., WM bundles). Therefore, a structure with higher FA usually has a low ODI. Similarly, MD and NDI also showed opposite changes. NDI represents neurite density and MD is shown to be negatively correlated with axon density (Mottershead et al., 2003; Schmierer et al., 2007). The opposite correlations between FA and ODI, between MD and NDI are consistent during mouse brain development, which agrees with the previous findings (Grussu et al., 2017; Fukutomi et al., 2018).

Brain development is a complex process, involving thousands of genes to coordinate molecular and cellular levels of brain organization and function. For instance, Hox genes exhibit functionality in generating selective muscle innervation subtypes and shaping synaptic specificity during brain development (McGinnis and Krumlauf, 1992; Philippidou and Dasen, 2013). In contrast, microcephaly genes assume a pivotal role in regulating brain size during development (Gilbert et al., 2005). In a broader context, the spatial transcriptomics of the mouse brain can reveal mechanisms related to genetic specification that underlie aspects of brain development and disorders at the molecular level (Wang and Wang, 2019; Piwecka et al., 2023; Lein et al., 2017). In this study, we have focused our investigation on a set of 2002 genes of interest, demonstrating a discernible impact on brain development (Allen Institute for Brain Science 2013). PLS components from the genes of interest exhibit spatial correlations with DTI, DKI, and NODDI metrics. Notably, results from GO enrichment analysis highlighted the prominent enrichment of GO terms associated with signaling pathways, neuroanatomical markers, and neurotransmission, all of which are of profound significance in the context of brain development.

On the other hand, dynamic changes in brain microstructure throughout development can be elucidated via dMRI. dMRI characterizes alterations in water diffusion patterns due to microstructural changes in brain tissue including axons, dendrites, myelin, and extracellular volumes (Lebel and Deoni, 2018; Alexander et al., 2019; DiPiero et al., 2023). Therefore, gene spatial transcriptomics and dMRI are valuable techniques for characterizing brain development. Within the framework of this study, we have undertaken a comprehensive analysis to explore the interplay between these two modalities during development, revealing a noteworthy spatial correlation between gene expression levels and dMRI data.

There are some limitations in this study. First, there are significant variations in dMRI parameters in the brain ROIs that cover both WM and GM structures. We are unable to separate WM and GM structures within the individual ROI, primarily because the available gene expression data are collected with coarse resolution (11 ROIs for the whole brain). In future studies, it is warranted to utilize single-cell gene expression data to distinguish specific subregions within their larger composite region. Second, we only used multiple rather than single response variables in the PLS regression model. In theory, the multiple response variables model offers a computational advantage due to its reduced computational complexity. In contrast, the single response variable PLS model excels in predictive accuracy, as each response variable possesses its own set of unique PLS loadings. In the future, we will also perform the single response variable model for better prediction. Thirdly, the residuals of the PLS regression model showed some degree of non-normality due to the limited sample size. In the future we will improve the normality by including more brain samples and refining the regression model. Fourthly, the animal study only enrolled three mice per age group due to the long image acquisition time (48 h per animal). Given the current conclusions, larger-scale studies, that enroll more mice and age groups, will be considered. Fifthly, the current analysis considered spatially correlated ROIs as samples in the PLS regression. Thus, the interpretation focuses on the point estimate of the association between gene expression profiles and dMRI metrics, rather than the inference. A future direction is to properly accountfor the dependence in a larger scale study. Lastly, for NODDI model, we kept the intrinsic diffusivity parameter Din constant throughout the developmental process for consistent comparison. Intrinsic diffusivity changes in neonatal brains as their axons and neurites may change rapidly (Guerrero et al., 2019). In the future, we will include quantitative validations by conventional histology for NODDI metrics.

5. Conclusions

In this study, we demonstrated tissue microstructure changes during mouse brain development using high-resolution dMRI. Both FA-ODI and MD-NDI showed opposite changes during development, and alternation in fiber orientation distribution could differentiate WM from GM as they have distinct development patterns. Gene expression displayed unique spatial patterns corresponding with dMRI alternations during brain development. Genes related to nerve system structure and functions were demonstrated to correlate with different quantitative parameters derived from dMRI.

Supplementary Material

1

Acknowledgment

The MRI images were acquired in the Center for In Vivo Microscopy at Duke University with the support of NIH P41 EB015897 and NIH S10OD010683.

Funding

This work was supported by the NIH R01 NS125020, NIH S10OD010683, Ralph W. and Grace M. Showalter Research Award, Indiana Center for Diabetes and Metabolic Diseases Pilot and Feasibility Grant, and IUSM Roberts Drug Discovery Fund & TREAT-AD Center Grant.

Data availability

The data that support the findings of this study and corresponding codes are available from the Github repository: https://github.com/selinahxy/developingmousebraindmri.

Abbreviations:

dMRI diffusion magnetic resonance imaging

DTI diffusion tensor imaging

NODDI neurite orientation dispersion and density imaging

DKI diffusion kurtosis imaging

WM white matter

GM gray matter

FA fractional anisotropy

MD mean diffusivity

AD axial diffusivity

RD radial diffusivity

NDI neurite density index

ODI orientation dispersion index

ISO isotropic volume fraction

AK axial kurtosis

KFA kurtosis fractional anisotropy

MK mean kurtosis

RK radial kurtosis

CS compressed sensing

CCFv3 Allen Mouse Brain Common Coordinate Framework

STP serial two-photon tomography

AMBA Allen Mouse Brain Atlas

ADMBA Allen Developing Mouse Brain Atlas

PLS partial least squares

GO gene ontology

qMRI quantitative magnetic resonance imaging

fMRI functional magnetic resonance imaging

3D three dimension

AF acceleration factor

DWI diffusion-weighted image

TR repetition time

TE time to echo

BW bandwidth

b0 baseline

ANTs Advanced Normalization Tools

GQI generalized q-sampling imaging

ISH in situ hybridization

ROI region of interest

VIP variable importance in projection

DEC diffusion-encoded-color

cc corpus callosum

ac anterior commissure

FOD fiber orientation distribution

RSP retrosplenial cortex

Tel telencephalic vesicle

HYP hypothalamus

p3 prosomere 3

p2 prosomere 2

p1 prosomere 1

M midbrain

PPH prepontine hindbrain

PH pontine hindbrain

PMH pontomedullary hindbrain

MH medullary hindbrain

Fig. 1. Brain volume and dMRI qualitative maps of mouse brains at different postnatal days. Brain volume (A) and dMRI qualitative maps including b0 (B), diffusion-weighted image (DWI, C), and diffusion-encoded-color (DEC, D) image of mouse brains at different postnatal days. The brain volume increases with age (A). Compared to DWI (C) and DEC (D) images, the corpus callosum (red arrows) only shows dark in the b0 (B) image after P14.

Fig. 2. dMRI quantitative maps of mouse brains at different postnatal days. The dMRI quantitative maps included FA (A), MD (B), NDI (C), and ODI (D) of mouse brains at different postnatal days. White arrows = isocortex, red arrows = corpus callosum, yellow arrows = hippocampus commissure, green arrows = cerebellum.(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 3. The ODI maps of the developing mouse brain for different structures. The ODI maps of the developing mouse brain for different structures, including (A) anterior commissure, (B) corpus callosum, (C) isocortex, (D) hippocampus, and (E) cerebellum. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 4. The volume, FA, and ODI changes during brain development. The volume, FA, and ODI changes in anterior commissure (A-D, blue), corpus callosum (A-D, red), isocortex (E-H, red), hippocampus (E-H, blue), and cerebellum (E-H, green) during brain development. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 5. The fiber orientation distributions during the mouse brain development. Example structures include corpus callosum (yellow box in A and B) and isocortex (green box in A and C). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 6. The registered ADMBA images (A, C) and dMRI images (B, D) at P4 and P14.

Fig. 7. dMRI parameters, partial least squares components, and their correlation plots at P4 and P14. (A-D) dMRI parameter changes, including DTI metrics(A), DKI metrics (B), and NODDI metrics (C), as well as PLS components changes (D) at P4 and P14. Red arrows = midbrain ROI, yellow arrows = prepontine hindbrain ROI. (E) Red dots are ROIs from P4, and blue dots are ROIs from P14. Pearson’s correlation r-values are denoted at each correlation plot. All correlations are significant (p < 0.05). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 8. Gene ontology enrichment analysis results. Gene ontology enrichment analysis results. 2002 genes-of-interest were ranked by variance importance in projection scores and input into GO enrichment analysis. GO aspects include biological process (left), cellular component (center), and molecular function (right). Semantically similar GO terms remain close together in the plot. Markers are scaled according to the enrichment and colored according to the log10 of the enrichment of each term. The GO term names of the most enriched genes under each GO aspect were numbered and denoted below.

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.neuroimage.2024.120734.

Declaration of competing interest

The authors declare there is no conflict of interests.

Compliance with ethical standards

Ethical approval

All animal studies have been approved by the appropriate ethics committee: Duke University Institutional Animal Care and Use Committee. Approval code A226-17-09.

Informed consent

No human subject was used in this study.

Declarations of generative AI and AI-assisted technologies in the writing process

During the preparation of this work, the authors used ChatGPT 3.5 and Grammarly in order to improve readability and grammar. After using these tools, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

CRediT authorship contribution statement

Xinyue Han: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Resources, Project administration, Methodology, Investigation, Formal analysis, Data curation. Surendra Maharjan: Visualization, Validation. Jie Chen: Visualization, Validation. Yi Zhao: Writing – review & editing, Formal analysis, Conceptualization. Yi Qi: Data curation. Leonard E. White: Writing – review & editing, Supervision, Conceptualization. G. Allan Johnson: Writing – review & editing, Conceptualization. Nian Wang: Writing – review & editing, Visualization, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.
==== Refs
References

Aggarwal M , Gobius I , Richards LJ , Mori S , 2015. Diffusion MR microscopy of cortical development in the mouse embryo. Cereb. Cortex 25 (7 ), 1970–1980.24518754
Alexander AL , Lee JE , Lazar M , Field AS , 2007. Diffusion tensor imaging of the brain. Neurotherapeutics 4 (3 ), 316–329.17599699
Alexander DC , Dyrby TB , Nilsson M , Zhang H , 2019. Imaging brain microstructure with diffusion MRI: practicality and applications. NMR Biomed. 32 (4 ), e3841.29193413
Allen Institute for Brain Science, Allen Institute for Cell Science. Informatics Data Processing For the Allen Developing Mouse Brain Atlas. Published online March 2010. https://help.brain-map.org/download/attachments/4325389/DevMouse_InformaticsDataProcessing.pdf.
Allen Institute for Brain Science, Allen Institute for Cell Science. Overview: Allen Developing Mouse Brain Atlas. Published online June 2013. https://help.brain-map.org/download/attachments/4325389/DevMouse_Overview.pdf?version=1&modificationDate=1370999986503&api=v2.
Allen Institute for Brain Science, Allen Institute for Cell Science. References Atlases for the Allen Developing Mouse Brain Atlas. Published online June 2013. https://community.brain-map.org/uploads/short-url/bkmaWYpLwmHXuhGDJt22Czt6FLH.pdf.
Andica C , Kamagata K , Hatano T , , 2020. MR biomarkers of degenerative brain disorders derived from diffusion imaging. J. Magn. Reson. Imaging 52 (6 ),1620–1636.31837086
Annese J , 2012. The importance of combining MRI and large-scale digital histology in neuroimaging studies of brain connectivity and disease. Front. Neuroinform 6 , 13.22536182
Arnatkevičūtė A , Fulcher BD , Fornito A , 2019. A practical guide to linking brainwide gene expression and neuroimaging data. Neuroimage 189 , 353–367.30648605
Avants BB , Tustison NJ , Song G , Cook PA , Klein A , Gee JC , 2011. A reproducible evaluation of ANTs similarity metric performance in brain image registration. Neuroimage 54 (3 ), 2033–2044.20851191
Baloch S , Verma R , Huang H , , 2009. Quantification of brain maturation and growth patterns in C57BL/6J mice via computational neuroanatomy of diffusion tensor images. Cereb. Cortex 19 (3 ), 675–687.18653668
Baratti C , Barnett AS , Pierpaoli C , 1999. Comparative MR imaging study of brain maturation in kittens with T1, T2, and the trace of the diffusion tensor. Radiology 210 (1 ), 133–142.9885598
BART Toolbox for Computational Magnetic Resonance Imaging. 10.5281/zenodo.592960.
Basser PJ , Mattiello J , LeBihan D , 1994. Estimation of the effective self-diffusion tensor from the NMR spin echo. J. Magn. Reson. B 103 (3 ), 247–254.8019776
Bayer SA , Altman J Neocortical development. No Title. Published online 1991.
Bockhorst K , Narayana P , Liu R , , 2008. Early postnatal development of rat brain: in vivo diffusion tensor imaging. J. Neurosci. Res 86 (7 ), 1520–1528.18189320
Bozzali M , Falini A , Franceschi M , , 2002. White matter damage in Alzheimer’s disease assessed in vivo using diffusion tensor magnetic resonance imaging. J. Neurol. Neurosurg. Psychiatry 72 (6 ), 742–746.12023417
Calabrese E , Badea A , Cofer G , Qi Y , Johnson GA , 2015. A diffusion MRI tractography connectome of the mouse brain and comparison with neuronal tracer data. Cereb. Cortex 25 (11 ), 4628–4637.26048951
Chong IG , Jun CH , 2005. Performance of some variable selection methods when multicollinearity is present. Chemom. Intell. Lab. Syst 78 (1–2 ), 103–112.
Chuang N , Mori S , Yamamoto A , , 2011. An MRI-based atlas and database of the developing mouse brain. Neuroimage 54 (1 ), 80–89.20656042
Colgan N , Siow B , O’Callaghan JM , , 2016. Application of neurite orientation dispersion and density imaging (NODDI) to a tau pathology model of Alzheimer’s disease. Neuroimage 125 , 739–744.26505297
Cosottini M , Giannelli M , Siciliano G , , 2005. Diffusion-tensor MR imaging of corticospinal tract in amyotrophic lateral sclerosis and progressive muscular atrophy. Radiology. 237 (1 ), 258–264.16183935
Dhital B , Kellner E , Kiselev VG , Reisert M , 2018. The absence of restricted water pool in brain white matter. Neuroimage 182 , 398–406.29129672
Diehn M , Nardini C , Wang DS , , 2008. Identification of noninvasive imaging surrogates for brain tumor gene-expression modules. Proc. Natl. Acad. Sci 105 (13 ), 5213–5218.18362333
DiPiero M , Rodrigues PG , Gromala A , Dean III DC , 2023. Applications of advanced diffusion MRI in early brain development: a comprehensive review. Brain Struct. Funct 228 (2 ), 367–392.36585970
Dubois J , Dehaene-Lambertz G , Kulikova S , Poupon C , Hüppi PS , Hertz-Pannier L , 2014. The early development of brain white matter: a review of imaging studies in fetuses, newborns and infants. Neuroscience 276 , 48–71.24378955
Eden E , Lipson D , Yogev S , Yakhini Z , 2007. Discovering motifs in ranked lists of DNA sequences. PLoS Comput. Biol 3 (3 ), e39.17381235
Eden E , Navon R , Steinfeld I , Lipson D , Yakhini Z , 2009. GOrilla: a tool for discovery and visualization of enriched GO terms in ranked gene lists. BMC Bioinform. 10 (1 ), 1–7.
Edwards LJ , McColgan P , Helbling S , , 2023. Quantitative MRI maps of human neocortex explored using cell type-specific gene expression analysis. Cereb. Cortex 33 (9 ), 5704–5716.36520483
Fornito A , Arnatkevičūtė A , Fulcher BD , 2019. Bridging the gap between connectome and transcriptome. Trends. Cogn. Sci 23 (1 ), 34–50.30455082
Fukutomi H , Glasser MF , Zhang H , , 2018. Neurite imaging reveals microstructural variations in human cerebral cortical gray matter. Neuroimage 182 , 488–499.29448073
Garyfallidis E , Brett M , Amirbekian B , , 2014. Dipy, a library for the analysis of diffusion MRI data. Front. Neuroinform 8 , 8.24600385
Ge Y , 2006. Multiple sclerosis: the role of MR imaging. Am. J. Neuroradiol 27 (6 ),1165–1176.16775258
Genc S , Malpas CB , Holland SK , Beare R , Silk TJ , 2017. Neurite density index is sensitive to age related differences in the developing brain. Neuroimage 148 ,373–380.28087489
Gilbert SL , Dobyns WB , Lahn BT , 2005. Genetic links between brain development and brain evolution. Nat. Rev. Genet 6 (7 ), 581–590.15951746
Graham J , Papadakis N , Evans J , , 2004. Diffusion tensor imaging for the assessment of upper motor neuron integrity in ALS. Neurology 63 (11 ), 2111–2119.15596758
Grussu F , Schneider T , Tur C , , 2017. Neurite dispersion: a new marker of multiple sclerosis spinal cord pathology? Ann. Clin. Transl. Neurol 4 (9 ), 663–679.28904988
Guerrero JM , Adluru N , Bendlin BB , , 2019. Optimizing the intrinsic parallel diffusivity in NODDI: an extensive empirical evaluation. PLoS ONE 14 (9 ), e0217118.
Hollingsworth KG , 2015. Reducing acquisition time in clinical MRI by data undersampling and compressed sensing reconstruction. Phys. Med. Biol 60 (21 ), R297.26448064
Holmes HE , Powell NM , Ma D , , 2017. Comparison of in vivo and ex vivo MRI for the detection of structural abnormalities in a mouse model of tauopathy. Front. Neuroinform 11 , 20.28408879
Holz M , Heil SR , Sacco A , 2000. Temperature-dependent self-diffusion coefficients of water and six selected molecular liquids for calibration in accurate 1H NMR PFG measurements. Phys. Chem. Chem. Phys 2 (20 ), 4740–4742.
Horsfield MA , Jones DK , 2002. Applications of diffusion-weighted and diffusion tensor MRI to white matter diseases–a review. NMR Biomed. Int. J. Devoted. Dev.Appl. Magn. Reson. Vivo 15 (7–8 ), 570–577.
Hüppi PS , Dubois J , 2006. Diffusion Tensor Imaging of Brain Development. Elsevier, pp. 489–497. Vol 11 .
Jensen JH , Helpern JA , Ramani A , Lu H , Kaczynski K , 2005. Diffusional kurtosis imaging: the quantification of non-Gaussian water diffusion by means of magnetic resonance imaging. Magn. Reson. Med 53 (6 ), 1432–1440.15906300
Johnson GA , Tian Y , Ashbrook DG , , 2023. Merged magnetic resonance and light sheet microscopy of the whole mouse brain. Proc. Natl. Acad. Sci 120 (17 ), e2218617120.
Koay CG , Özarslan E , Johnson KM , Meyerand ME , 2012. Sparse and optimal acquisition design for diffusion MRI and beyond. Med. Phys 39 (5 ), 2499–2511.22559620
Kroenke CD , 2018. Using diffusion anisotropy to study cerebral cortical gray matter development. J. Magn. Reson 292 , 106–116.29705039
Kroenke CD , Taber EN , Leigland LA , Knutsen AK , Bayly PV , 2009. Regional patterns of cerebral cortical differentiation determined by diffusion tensor MRI.Cereb. Cortex. 19 (12 ), 2916–2929.
Kronman FA , Liwang JK , Betty R , Developmental Mouse Brain Common Coordinate Framework. bioRxiv. Published online 2023:2023–09.
La Manno G , Siletti K , Furlan A , , 2021. Molecular architecture of the developing mouse brain. Nature 596 (7870 ), 92–96.34321664
Lebel C , Deoni S , 2018. The development of brain white matter microstructure.Neuroimage 182 , 207–218.29305910
Le Bihan D , 2003. Looking into the functional architecture of the brain with diffusion MRI. Nat. Rev. Neurosci 4 (6 ), 469–480.12778119
Lein E , Borm LE , Linnarsson S , 2017. The promise of spatial transcriptomics for neuroscience in the era of molecular cell typing. Science 358 (6359 ), 64–69.28983044
Lein ES , Belgard TG , Hawrylycz M , Molnaŕ Z , 2017. Transcriptomic perspectives on neocortical structure, development, evolution, and disease. Annu Rev. Neurosci 40 , 629–652.28661727
Lein ES , Hawrylycz MJ , Ao N , , 2007. Genome-wide atlas of gene expression in the adult mouse brain. Nature 445 (7124 ), 168–176.17151600
Lerch JP , Van Der Kouwe AJ , Raznahan A , , 2017. Studying neuroanatomy using MRI. Nat. Neurosci 20 (3 ), 314–326.28230838
Liscovitch N , Chechik G , 2013. Specialization of gene expression during mouse brain development. Sporns O. PLoS Comput. Biol 9 (9 ), e1003185 10.1371/journal.pcbi.1003185.
Lustig M , Donoho D , Pauly JM , 2007. Sparse MRI: The application of compressed sensing for rapid MR imaging. Magn. Reson. Med 58 (6 ), 1182–1195.17969013
Ma D , Holmes HE , Cardoso MJ , , 2019. Study the longitudinal in vivo and cross-sectional ex vivo brain volume difference for disease progression and treatment effect on mouse model of tauopathy using automated MRI structural parcellation. Front. Neurosci 13 , 408510.
MacKenzie-Graham A , Jones ES , Shattuck DW , Dinov ID , Bota M , Toga AW , 2003. The informatics of a C57BL/6J mouse brain atlas. Neuroinformatics 1 ,397–410.15043223
MacKenzie-Graham A , Lee E , Dinov ID , , 2004. A multimodal, multidimensional atlas of the C57BL/6J mouse brain. J. Anat 204 (2 ), 93–102.15032916
Mah A , Geeraert B , Lebel C , 2017. Detailing neuroanatomical development in late childhood and early adolescence using NODDI. PLoS ONE 12 (8 ), e0182340.
McGinnis W , Krumlauf R , 1992. Homeobox genes and axial patterning. Cell 68 (2 ),283–302.1346368
McKinstry RC , Mathur A , Miller JH , , 2002. Radial organization of developing preterm human cerebral cortex revealed by non-invasive water diffusion anisotropy MRI. Cereb. Cortex 12 (12 ), 1237–1243.12427675
Mielke MM , Kozauer N , Chan K , , 2009. Regionally-specific diffusion tensor imaging in mild cognitive impairment and Alzheimer’s disease. Neuroimage 46 (1 ),47–55.19457371
Mori S , Itoh R , Zhang J , , 2001. Diffusion tensor imaging of the developing mouse brain. Magn. Reson. Med 46 (1 ), 18–23.11443706
Mottershead J , Schmierer K , Clemence M , , 2003. High field MRI correlates of myelin content and axonal density in multiple sclerosis: a post-mortem study of the spinal cord. J. Neurol 250 , 1293–1301.14648144
Mukherjee P , McKinstry RC , 2006. Diffusion tensor imaging and tractography of human brain development. Neuroimaging Clin. 16 (1 ), 19–43.
Neil JJ , Shiran SI , McKinstry RC , , 1998. Normal brain in human newborns: apparent diffusion coefficient and diffusion anisotropy measured by using diffusion tensor MR imaging. Radiology. 209 (1 ), 57–66.9769812
Patel Y , Shin J , Drakesmith M , Evans J , Pausova Z , Paus T , 2020. Virtual histology of multi-modal magnetic resonance imaging of cerebral cortex in young men. Neuroimage 218 , 116968.32450248
Peris M , Benseny-Cases N , Manich G , , 2023. Roadmap for postnatal brain maturation: changes in gray and white matter composition during development measured by fourier transformed infrared microspectroscopy. ACS Chem. Neurosci 14 (17 ), 3088–3102.37540627
Philippidou P , Dasen JS , 2013. Hox genes: choreographers in neural development, architects of circuit organization. Neuron 80 (1 ), 12–34.24094100
Piwecka M , Rajewsky N , Rybak-Wolf A , 2023. Single-cell and spatial transcriptomics: deciphering brain complexity in health and disease. Nat. Rev. Neurol 1–17.Published online.36418496
Prodoehl J , Li H , Planetta PJ , , 2013. Diffusion tensor imaging of Parkinson’s disease, atypical parkinsonism, and essential tremor. Mov. Disord 28 (13 ),1816–1822.23674400
Rosenberg AB , Roco CM , Muscat RA , , 2018. Single-cell profiling of the developing mouse brain and spinal cord with split-pool barcoding. Science 360 (6385 ), 176–182.29545511
Sanchez-Molina P , Kreuzer M , Benseny-Cases N , , 2020. From mouse to human: Comparative analysis between grey and white matter by synchrotron-Fourier transformed infrared microspectroscopy. Biomolecules 10 (8 ), 1099.32722088
Schmierer K , Wheeler-Kingshott CA , Boulby PA , , 2007. Diffusion tensor imaging of post mortem multiple sclerosis brain. Neuroimage 35 (2 ), 467–477.17258908
Silk TJ , Vance A , Rinehart N , Bradshaw JL , Cunnington R , 2009. White-matter abnormalities in attention deficit hyperactivity disorder: a diffusion tensor imaging study. Hum. Brain Mapp. 30 (9 ), 2757–2765.19107752
Stebbins G , Murphy C , 2009. Diffusion tensor imaging in Alzheimer’s disease and mild cognitive impairment. Behav. Neurol 21 (1–2 ), 39–49.19847044
Stiles J , Jernigan TL , 2010. The basics of brain development. Neuropsychol. Rev 20 (4 ), 327–348.21042938
Supek F , Bošnjak M , Škunca N , Šmuc T , 2011. REVIGO summarizes and visualizes long lists of gene ontology terms. PLoS ONE 6 (7 ), e21800.21789182
Szeszko PR , Ardekani BA , Ashtari M , , 2005. White matter abnormalities in obsessive-compulsive disorder: a diffusion tensor imaging study. Arch. Gen. Psychiatry 62 (7 ), 782–790.15997020
Takahashi M , Ono J , Harada K , Maeda M , Hackney DB , 2000. Diffusional anisotropy in cranial nerves with maturation: quantitative evaluation with diffusion MR imaging in rats. Radiology 216 (3 ), 881–885.10966726
Thompson CL , Ng L , Menon V , , 2014. A high-resolution spatiotemporal atlas of gene expression of the developing mouse brain. Neuron 83 (2 ), 309–323.24952961
To XV , Kurniawan ND , Cumming P , Nasrallah FA , 2023. A cross-comparative analysis of in vivo versus ex vivo MRI indices in a mouse model of concussion. Brain Res. 1820 , 148562.37673379
Toosy A , Werring D , Orrell R , , 2003. Diffusion tensor imaging detects corticospinal tract involvement at multiple levels in amyotrophic lateral sclerosis. J. Neurol. Neurosurg. Psychiatry 74 (9 ), 1250–1257.12933929
Vaillancourt D , Spraker M , Prodoehl J , , 2009. High-resolution diffusion tensor imaging in the substantia nigra of de novo Parkinson disease. Neurology 72 (16 ),1378–1384.19129507
Vértes PE , Rittman T , Whitaker KJ , , 2016. Gene transcription profiles associated with inter-modular hubs and connection distance in human functional magnetic resonance imaging networks. Philos. Trans. R. Soc. B 371 (1705 ), 20150362.
Wang N , Anderson RJ , Badea A , , 2018. Whole mouse brain structural connectomics using magnetic resonance histology. Brain Struct. Funct 223 ,4323–4335.30225830
Wang N , Maharjan S , Tsai AP , , 2023. Integrating multimodality magnetic resonance imaging to the Allen Mouse Brain Common Coordinate Framework. NMR Biomed. 36 (5 ), e4887. 10.1002/nbm.4887.36454009
Wang N , White LE , Qi Y , Cofer G , Johnson GA , 2020. Cytoarchitecture of the mouse brain by high resolution diffusion magnetic resonance imaging. Neuroimage 216 , 116876.32344062
Wang N , Zhang J , Cofer G , , 2019. Neurite orientation dispersion and density imaging of mouse brain microstructure. Brain Struct. Funct 224 , 1797–1813.31006072
Wang Q , Ding SL , Li Y , , 2020. The Allen mouse brain common coordinate framework: a 3D reference atlas. Cell 181 (4 ), 936–953.32386544
Wang W , Wang GZ , 2019. Understanding molecular mechanisms of the brain through transcriptomics. Front. Physiol 10 , 214.30930788
Wen J , Goyal MS , Astafiev SV , Raichle ME , Yablonskiy DA , 2018. Genetically defined cellular correlates of the baseline brain MRI signal. Proc. Natl. Acad. Sci 115 (41 ), E9727–E9736.30254176
Wu D , Richards LJ , Zhao Z , , 2022. A diffusion MRI-based spatiotemporal continuum of the embryonic mouse brain for probing gene–neuroanatomy connections. Proc. Natl. Acad. Sci 119 (7 ), e2111869119.
Wu D , Zhang J , 2016. In vivo mapping of macroscopic neuronal projections in the mouse hippocampus using high-resolution diffusion MRI. Neuroimage 125 , 84–93.26499812
Yeh FC , Verstynen TD , Wang Y , Fernández-Miranda JC , Tseng WYI , 2013. Deterministic diffusion fiber tracking improved by quantitative anisotropy. PLoS ONE 8 (11 ), e80713.24348913
Yeh FC , Wedeen VJ , Tseng WYI , 2010. Generalized ${q} $-sampling imaging. IEEE Trans. Med. ImAging 29 (9 ), 1626–1635.20304721
Yoshikawa K , Nakata Y , Yamada K , Nakagawa M , 2004. Early pathological changes in the parkinsonian brain demonstrated by diffusion tensor MRI. J. Neurol.Neurosurg. Psychiatry 75 (3 ), 481–484.14966170
Zhang H , Schneider T , Wheeler-Kingshott CA , Alexander DC , 2012. NODDI: practical in vivo neurite orientation dispersion and density imaging of the human brain. Neuroimage 61 (4 ), 1000–1016.22484410
Zhang J , Richards LJ , Yarowsky P , Huang H , Van Zijl PC , Mori S , 2003. Three-dimensional anatomical characterization of the developing mouse brain by diffusion tensor microimaging. Neuroimage 20 (3 ), 1639–1648.14642474
Zhao X , Shi J , Dai F , , 2021. Brain development from newborn to adolescence:Evaluation by neurite orientation dispersion and density imaging. Front. Hum. Neurosci 15 , 616132.33790750
