Machine Learning in Clinical Neuroimaging

Speaker: Kerstin Ritter, Charité Universitätsmedizin

Machine learning techniques have become increasingly popular for diagnosing and characterizing neurological and psychiatric diseases based on structural and functional MRI data. In this talk, I will review research from machine learning to deep learning in clinical neuroimaging including own work on multiple sclerosis, Alzheimer's disease, and mental health. I will focus on transparent convolutional neural network models, novel architectures incorporating spatial priors, and discuss several challenges with respect to sample size, confounding variables, and transparency.