Thomas Tourdias

Generating Synthetic MR Perfusion Maps from DWI and FLAIR in Acute Ischemic Stroke: Development and External Validation of a Deep Learning Model

This paper develops and validates a deep learning model to generate synthetic T-max perfusion maps from non-contrast MRI data in acute stroke patients.

DeepDisco: A Deep Learning Tool for Rapid Brain Connectivity Estimation

DeepDisco predicts brain connectivity maps from binary regions or lesions using deep learning, providing a scalable and accessible tool for large datasets.

MRI R2* captures inflammation in disconnected brain structures after stroke: a translational study

This study shows that MRI R2* changes reflect inflammation in disconnected brain regions after stroke, both in humans and mice, linking disconnection to iron accumulation and microglial activation.

A global effort to benchmark predictive models and reveal mechanistic diversity in long-term stroke outcomes

This study benchmarks predictive models for long-term stroke outcomes using a global consortium, revealing mechanistic diversity and key predictors across motor, cognitive, and emotional domains.

Deep Learning disconnectomes to accelerate and improve long-term predictions for post-stroke symptoms

This paper uses deep learning to generate disconnectomes from stroke lesion masks, achieving faster computation and better prediction of post-stroke cognitive and behavioral outcomes.

Lifespan Neurodegeneration Of The Human Brain In Multiple Sclerosis

This study models lifespan brain atrophy trajectories in MS patients and healthy controls, identifying early subcortical structures affected before clinical diagnosis.

Multimodal Hippocampal Subfield Grading For Alzheimer’s Disease Classification

This study compares whole hippocampus and subfield biomarkers for Alzheimer’s disease detection and prediction using multimodal MRI data, finding subiculum biomarkers improve prediction accuracy.

Objective Evaluation of Multiple Sclerosis Lesion Segmentation using a Data Management and Processing Infrastructure

This paper evaluates MS lesion segmentation algorithms using a new open-science infrastructure, finding that automatic methods lag behind human experts in detection and delineation.