Mental health awareness is a growing movement, but prevention and diagnoses of mental health disease is difficult. Unlike many other parts of the body, the human brain is not understood to the extent that direct relationships between external stimuli and temporary moods, long-term depression, anxiety, etc. are fully understood. Deep learning algorithms provide a way to extract high-level information from images, and thus present a promising method to aid in diagnosis of mental health disorders. In this project, we applied and evaluated several neural network architectures to the binary classification problem of depression diagnosis in the fMRI dataset openneuro-ds000171 and present several promising results.