Continual medical image denoising based on triplet neural networks collaboration

Xianhua Zeng, Yongli Guo, Laquan Li, Yuhang Liu

Abstract: Highlights•Proposed a Triplet Neural-networks Collaboration-continuity DeNosing (TNCDN) model.•Designed a differential loss and a continual loss to solve catastrophic for getting.•Designed a co-training mechanism to maintain continual learning ability.•Results show the model has the lowest forgetting degree compared with other methods.