Federated Learning to Safeguard Patients Data: A Medical Image Retrieval Case

Healthcare data are distributed and confidential, making it difficult to use centralized automatic diagnostic techniques. For example, different hospitals hold the electronic health records (EHRs) of patient populations; however, transferring this between is due sensitive nature information. This presents a significant obstacle development efficient generalizable analytical methods that require large amount diverse Big Data. Federated learning allows multiple institutions work together develop machine algorithm without sharing their data. We conducted systematic study analyze current state FL in healthcare industry explore both limitations technology its potential. Organizations share parameters models with each other. them reap benefits model developed richer set while protecting confidentiality Standard for large-scale learning, optimization, privacy-friendly analytics need be fundamentally rethought address new problems posed by training on networks may contain amounts In article, we discuss particular qualities difficulties federated provide comprehensive overview approaches, outline several directions future relevant variety research communities. These issues important many