Federal Learning and Deep Learning Framework for MRI and Facial Expression Single Bases Multi Modal of Mental Illness

Mental Health disorders are becoming highly common, specifically among students and young adults who face daily pressure with respect to academics, career and personal life affecting personal balances both emotionally and physically, change in behavior and very less social interaction. Medical images such as MRI scan and easily seen facial expression compel us to have insights of mental condition of a person. Although deep learning techniques is considered effective in image recognition but simultaneously this old centralized method raise serious issues regarding security, data privacy and valid use of medical data. The objective of this study is to overcome these challenges through conceptual federated multimodal framework that aligns face expression with MRI image for better detection of mental situation. Through this approach we solve the data security and above stated technical issues as data remains within the institution while only trained model share updates during training. The need of the study is not to derive result from single resource or an experiment but to explain collaborative working of federated learning and multimodal deep learning, have a view on their advantages and disadvantages and check for future challenges.