
Click anywhere on the brain and see that region's projections. Fibers are diffusion tractography (dMRI) from Human Connectome Project amyleesterling.github.io/hum…

🧠UPenn Neuroscience PhD Candidate 🎓Haverford '20

Click anywhere on the brain and see that region's projections. Fibers are diffusion tractography (dMRI) from Human Connectome Project amyleesterling.github.io/hum…

I am so grateful for all of my amazing collaborators who helped make this research possible, particularly the outstanding mentorship and support from my advisor Kate Davis & close collaborator Ted Satterthwaite (@sattertt)

Takeaway #3: Microstructural factors are distinctly organized, which influences their sensitivity to microstructural changes in different regions. In TLE, factor alterations differed within clinically relevant circuits --> Disease mapping benefits from using complementary metrics

Takeaway #2: dMRI microstructural investigations may benefit from using at least 1 dMRI statistic per microstructural factor, as overall, non-Gaussian, and anisotropic diffusivity carry complementary information

Takeaway #1: We provide a reference for dMRI statistic similarity, releasing mean dMRI statistic & factor maps in a standard template (MNI) space (osf.io/xsr7y) for further study. dMRI statistic loadings are shared so investigators can map their data on these factors

Hippocampal overall/non-Gaussian diffusivity alterations were more pronounced in TLE than measures of anisotropy. Did this extend to patient-specific images? Yes: RTOP/MD (overall) & ICVF (non-Gaussian) exhibited clearer hippocampal asymmetries than FA (anisotropic)

TLE abnormalities were most pronounced in temporolimbic regions, prompting closer examination of the mesial limbic circuit. Factor alterations were heterogeneous here, suggesting that multiple microstructural factors may be needed to comprehensively map focal circuit disruptions

We characterized how each factor captured microstructural alterations in unilateral temporal lobe epilepsy (TLE). We mapped microstructural abnormalities & their asymmetries (ipsilateral vs. contralateral regions to the seizure focus) within each factor & across all statistics

We examined gradients of factor similarity across the brain. Regions closer along gradients --> more similar factor expression. Principal gradients separated grey & white matter regions across factors, though each factor showed distinct local organization

Factor analysis of dMRI statistic correlations across brain regions and controls revealed 3 dominant axes of dMRI statistic similarities, corresponding to diffusion magnitude ("overall diffusivity"), complexity ("non-Gaussian diffusivity"), & directional coherence ("anisotropy")

Many dMRI statistics have been investigated, but their whole-brain relationships are unclear. This makes it challenging to determine which statistics capture overlapping vs. unique microstructural information

Diffusion MRI (dMRI) is sensitive to the brain's microstructure, but a myriad of dMRI measures exist. Can their similarities be captured concisely by a few factors? How are these factors organized in the brain? How are they altered in focal epilepsy? A 🧵 biorxiv.org/content/10.64898…

I am so grateful for all of my amazing collaborators who helped make this research possible, particularly the outstanding mentorship and support from my advisor Kate Davis & close collaborator Ted Satterthwaite (@sattertt)

Takeaway #3: Microstructural factors are distinctly organized, which influences their sensitivity to microstructural changes in different regions. In TLE, factor alterations differed within clinically relevant circuits --> Disease mapping benefits from using complementary metrics

Takeaway #2: dMRI microstructural investigations may benefit from using at least 1 dMRI statistic per microstructural factor, as overall, non-Gaussian, and anisotropic diffusivity carry complementary information

Takeaway #1: We provide a reference for dMRI statistic similarity, releasing mean dMRI statistic & factor maps in a standard template (MNI) space (osf.io/xsr7y) for further study. dMRI statistic loadings are shared so investigators can map their data on these factors