David Lee Thomas

The paper introduces a fast segmentation tool for brain MRI scans using a hybrid method, achieving comparable accuracy to a Bayesian approach but with significantly reduced runtime.

This study uses MRI to measure blood-cerebrospinal fluid barrier function in a mouse model of Alzheimer's disease, finding increased water delivery through the barrier in early stages.

The paper introduces NextBrain, a probabilistic histological atlas of the human brain, and presents a tool for automated MRI segmentation of anatomical regions.

This study compares multi-echo combination methods for MRI quantitative susceptibility mapping (QSM) and finds that nonlinear phase fitting before Laplacian-based methods provides the highest accuracy and lowest noise.

This paper introduces a tissue-weighted mean method to reduce bias in NODDI metrics by accounting for CSF partial volume, showing it improves accuracy in ROI analysis for neurodegenerative disease studies.

The paper describes an MRI dataset and processing pipeline for the EPAD study, identifying quality control features and image-derived phenotypes related to aging and dementia.

This study evaluates how flow compensation and QSM reconstruction methods affect venous susceptibility measurements, finding that QSM methods have a larger impact than acquisition sequences.

This paper introduces a computational pipeline to reconstruct 3D histology of the human brain using MRI and blockface photography to overcome tissue sectioning distortions.

ExploreASL provides a standardized image processing pipeline for multi-center ASL MRI studies, improving reproducibility and enabling group-level statistical analysis across diverse clinical populations.

The paper investigates the relationship between gene expression and brain atrophy in genetic frontotemporal dementia, finding correlations between atrophy patterns and specific gene sets related to vascular, glial, and neuronal functions.

This study shows that diffusion tractography-based parcellations of subcortical structures can be obtained without internal anisotropy, suggesting caution in interpreting such results.

This paper introduces SuStaIn, a machine learning method that identifies disease subtypes and their temporal progression from cross-sectional data, improving prediction of diagnostic conversion in neurodegenerative diseases.