Psychiatric disorders are amongst the most difficult to accurately diagnose and design a treatment plan for. Imaging the structural and functional properties of an individual’s brain is the key to solving this challenge. Current machine learning approaches fail to utilize the information in these brain scans. Broadly, we propose finding a manifold of structural and/or functional brain scans in an embedding space that clusters mental disease states into clinically recognized classes. Specially, we propose a flavour of the adversarial autoencoder to accomplish this task. The goal is wide and ambitious enough to accommodate several other alternatives to solve this challenge.