Quantification of Brain Pathology in Magnetic Resonance Imaging Using Image Analysis for Stroke Research

Magnetic resonance images of the brain have been investigated extensively to determine whether precursors to stroke exist. Several studies show that hyperintense objects scattered throughout the white matter, known as white matter lesions (WML), are an independent risk factor for future stroke events.

To prove the relationship between WML and stroke, the volumes of WML were found by visual analysis and correlated to patient outcome. Unfortunately, obtaining volumetric measurements manually is subjective (i.e., observer-dependent), qualitative, error-prone and laborious. As a result, it is not possible to conduct large-scale research studies to understand disease on a deeper level.

Image analysis techniques combat these challenges by automatically measuring image objects in a quantitative, objective, efficient and reproducible manner. Disease can be accurately quantified for diagnostic purposes, and large patient cohorts can be analyzed efficiently for research.

In line with these goals, the focus of this research program is to design and develop image-processing methodologies that quantitatively analyze magnetic resonance images for stroke prevention by uncovering disease causes, models, trajectories and relationships. In the future, such models can be used to develop targeted, personalized therapies.