FL-SiCNN: An improved brain tumor diagnosis using siamese convolutional neural network in a peer-to-peer federated learning approach

Artificial Intelligence has been an essential component for successful data-driven medical applications. Considering today's conditions, Deep Learning holds the leading role in advancing the field of Artificial Intelligence and ensuring positive results for even the most complicated medical problems. More specifically, deep learning has been effectively used, especially in medical image-based analysis and diagnosis problems. In this context, cancer diagnosis has value in research studies and it still has space for alternative solution ways. On the other hand, the use of private patient data, keeping the data from cyber threats, and building a collaborative way for improving the learning from medical image data have been open questions in recent research efforts. The objective of this study is to provide a Deep Learning-based approach for dealing with the related open questions and advancing the way of cancer diagnosis via Artificial Intelligence. The study targeted brain tumor diagnosis and designed a Siamese Convolutional Neural Network (SiCNN) to advance the diagnosis mechanisms. At this point, the whole Deep Learning approach has been supported with a peer-to-peer (P2P) Federated Learning environment where local systems are collaboratively working to provide a good performing, privacy-preserving solution methodology for classifying brain tumors from MRI images. After developing the SiCNN model and the Federated Learning architecture, the whole system called as FL-SiCNN was examined through evaluation works. The obtained results showed that the SiCNN was effective enough in brain tumor diagnosis while ensuring data privacy and safety along the Deep Learning flow.