This repository contains code associated with out publication entitled "Unsupervised Abnormality Detection in Neonatal MRI Brain Scans Using Deep Learning" - Jad Dino Raad, Ratna Babu Chinnam, Suzan Arslanturk, Sidhartha Tan, Jeong-Won Jeong, and Swati Mody (currently under review at Scientific Reports).
In this research, we explore 3D Autoencoder (AE) and Variational Autoencoder (VAE) architectures for identifying anomalies in neonatal MRI brain scans. We train our architectures on normal T2-weighted brain volumes from the Developing Human Connectome Project, and test our approaches on abnormal data from the same dataset.