Cross population study of white matter connectivity using dictionary learning and sparse coding

Understanding the structure and connections in the brain is an outstanding research problem. Many diseases impact the structure and hence connections between them, analyzing which may help in detection and diagnosis of these diseases. In this project, we try to learn an atlas of major connecting fibers (tube like connections) for a group of people, using dictionary learning based framework. This atlas is then used to observe the similarities and differences in fibers across the group, and to segment fibers from new person. We will use twin dataset of the Human Connectome Project to study heritability of white matter clusters. The method will also be used to understand the differences in structural connectivity between healthy subjects, mild cognitive impairment (MCI) subjects, and Alzheimer’s disease (AD) subjects.

Faculty Supervisor:

Christian Desrosiers

Student:

Kuldeep Kumar

Partner:

Discipline:

Engineering - biomedical

Sector:

University:

École de technologie supérieure

Program:

Globalink Research Award