Gastric Cancer Diagnosis with Mask R-CNN

It is a reliable way to judge gastric cancer by pathological section. Using deep learning method to detect medical images, as an auxiliary diagnosis method, it can improve the speed and accuracy of doctors to diagnose gastric cancer, and reduce misdiagnosis and missed diagnosis. Mask R-CNN is the latest method in the related field at the beginning of the research. It is mainly used to segment the objects in daily life and achieve good results. The medical image is very different from the scene in life, and the detection effect is also weakened. We use the Mask R-CNN method to detect the pathological sections of gastric cancer, and segment the cancer nest, and then optimize it by adjusting parameters. The method finally allows it to obtain a test result with an AP value of 61.2 when detecting medical images.