Histopathologic Cancer Detection with Hybrid Deep Learning model
Xinyi Zhang, Xin Xiao, Mingda Huo, Xiaolong Bai · 2023
Histopathological examination, as the "gold standard" recognized by the medical community, can easily reach the resolution of microns and is a qualitative examination. With the rapid development of computer hardware and software, deep learning has been triggered in the computer-aided diagnosis of medical images. To create an algorithm to identify metastatic cancer in small image patches taken from larger digital pathology scans, we proposed a hybrid deep learning model which combining Resnet and Densenet We combined Resnet and Densenet to extract image features. We introduced some related work about the histopathological examination. In the experiment, we compared our model with other models, the elevation metrics is accuracy and Auc-Roc score. From the definition, the higher Auc Roc Score and accuracy are, the better performance the model will gain. On the modified version of the PatchCamelyon (PCam) benchmark dataset, Our model achieved the highest AUC score (0.971) and highest accuracy (0.982) on the test set.