Physics (Internal)

Deep learning for Alzheimer's disease assessment

This event is part of the Preliminary Oral Exam.

Despite the recent prosperity of machine learning research in medicine, its clinical applications remain challenging. Current clinical diagnosis of Alzheimer’s disease (AD) is complicated and requires expert doctors to carefully review patient’s medical history, various neuropsychological and functional tests, and brain imaging scans. Efficient, accurate and interpretable AI-aided diagnostic method for AD is thus in urgent demand. We created an interpretable deep learning framework to generate AD risk maps on brain MRI scans and validated our approach with independent datasets. Ongoing work is focused on incorporating this framework for the diagnosis of the pre-dementia stage, AD dementia, and other forms of dementia. The broad diagnostic spectrum makes the current project a potential tool to fit in the clinical environment.