Convolutional Neural Network Visualization for fMRI Brain Disease Classification Tasks

Neurologists face an increasingly overwhelming amount of data that they must use to determine diagnoses for patients with potential brain diseases. Our project aims to supplement the upcoming technology of automated brain disease classification using deep convolutional neural networks to develop a suite of visualization tools for these high dimensional deep convolutional neural networks. For this project we developed high quality visualizations for convolutional neural networks that classify fMRI brain scans. Visualizations such as those developed in this project could be used by doctors to identify which parts of the brain are indicators for mental conditions or by data scientists to understand more about how their networks work