Research on the Auxiliary Classification and Diagnosis of Lung Cancer Subtypes Based on Histopathological Images

Lung cancer (LC) is one of the most serious cancers threatening human health. Histopathological examination gold standard for qualitative and clinical staging lung tumors. However, process doctors to examine thousands histopathological images very cumbersome, especially with less experience. Therefore, objective pathological diagnosis results can effectively help choose appropriate treatment mode, thereby improving survival rate patients. For current problem incomplete experimental subjects in computer-aided subtypes, this study included relatively rare adenosquamous carcinoma (ASC) samples first time, proposed a method based on ASC, squamous cell (LUSC) small (SCLC). Firstly, multidimensional features 121 LC were extracted, then relevant (Relief) algorithm was used feature selection. The support vector machines (SVMs) classifier classify receiver operating characteristic (ROC) curve area under (AUC) make it more intuitive evaluate generalization ability classifier. Finally, through horizontal comparison variety mainstream classification models, experiments show that effect achieved by Relief-SVM model best. LUSC-ASC accuracy 73.91%, LUSC-SCLC 83.91% ASC-SCLC 73.67%. Our verify potential auxiliary constructed machine learning (ML) LC.