Local-global multiple perception based deep multi-modality learning for sub-type of esophageal cancer classification

Zhicheng Zhou, Long Yu, Shengwei Tian, Yan Xing, Meng Liu, Guangli Xiao, Junwen Wang, Fan Wang

Abstract: Highlights•Considering the feature information from both the global information and the local information.•Extremely efficient- need only 3% of the normal time to achieve 0.9732 ACC.•Introducing dynamic images (dynamic CT) in the esophageal cancer task.•The network structure has the function of structural reparameterization.•The feature enhancement module can adjust the feature map weights.