The MDD model finished with an accuracy of 88.75% with a P-value of 0.043, significant at an alpha level of 0.05. This significance was determined by comparing the model performance to past deep learning networks. Extraction of the final weighted layer allowed for the generation of heatmaps by utilizing the jet function. These mappings demonstrated support for an altered Corpus Callosum based on the weights determined by ImageNet. Alteration of the Corpus Callosum supports recent MDD findings (van Velzen et al., 2020). However, this model size is large and therefore exceeds the ram limit for computation on most computers. Instead, below, another MobileNet model is available for trial (Accuracy 82.19%). The analysis will not be as accurate as ResNet50 due to the model architecture of MobileNet. Certain input sizes may be rejected by the model. (1 voxel @ 244 x 244 is the only allowed input)