Breast cancer detection from biopsy images using nucleus guided transfer learning and belief based fusion

Kalpana George, Shameer Faziludeen, Praveen Sankaran, Joseph K. Paul

Abstract: Highlights•A nucleus guided transfer learning method is proposed for breast cancer detection.•Pretrained CNNs as feature extractors generate a system of low complexity.•Feature fusion approach obtain a discriminative feature representing an image.•A belief theory based classifier fusion strategy is proposed to enhance accuracy.•The proposed framework ‘NucTraL+BCF’ achieves an average accuracy of 96.91%.