Utilizing MRI Scans and Machine Learning to Enhance Depression Diagnosis

Depressive disorder rates have been increasing rapidly in recent history, and our paper offers the groundwork for a tool to combat this rise. Many therapists rely largely on qualitative analysis when diagnosing depressive disorders, however, our aim is to provide a more quantitative analysis in classifying depression and its severity. Whereas most modern techniques for diagnosing depression are dependent upon surveys and questionnaires as well as interviews, our solution utilizes neuroimaging techniques, particularly, MRI scans. This paper demonstrates the usage of multiple machine learning methods acting on a set of MRI scans allowing the models to diagnose depression and give an indicator of its severity. Our MRI scans are obtained through a publicly available dataset on OpenNeuro. If our research is implemented in therapist and psychologists' offices, we can potentially increase the accuracy of the diagnoses that are made by professionals thereby improving the treatment that patients receive.