CCE-Net: A rib fracture diagnosis network based on contralateral, contextual, and edge enhanced modules

Rib fracture is a common disease that requires prompt treatment. This study focuses on developing rib diagnosis deep learning method using contralateral, contextual, and edge enhanced modules evaluating its detection performance. A novel was designed, named CCE-Net. To evaluate the performance of this method, 1639 digital radiography (DR) images were enrolled. Fracture features extracted for three modules: modules. These can be used to identify in DR images, imitating experience broad-certificated radiologists. The contralateral module assists diagnosing fractures by comparing difference between detected target region region. contextual helps aid extracting features. improves accuracy enhancing information bone. head two-stage network uses multi-path fusion mechanism as main architecture integrate utilize above qualitative results show with ground truth evaluation standard, CCE-Net achieve better visual effect than other methods. quantitative best various indicators include AP50 0.911, AP75 0.794, AP25 0.913, Recall 0.934. Experimental acquire excellent ability diagnosis. We invasion approach will applied clinical study.