DEA: Data-efficient augmentation for interpretable medical image segmentation

Xing Wu, Zhi Li, Chenjie Tao, Xianhua Han, Yen-Wei Chen, Junfeng Yao, Jian Zhang, Qun Sun, Weimin Li, Yue Liu, Yike Guo

Abstract: Highlights•We propose the method DEA for medical image segmentation to enhance data efficiency.•We propose to extract data-efficient features with Grad-CAM.•With 70% of training data, DEA enables different segmentation methods to achieve over 95% accuracy.