This study proposes a novel cancer scoring card (CSC) method, which uses the weights of each image feature to build a scorecard, and further optimized using an intelligent genetic algorithm. To apply image analysis of H&E stained colorectal lymph node slides. The degree of metastasis is determined by the sum of the feature weights. The system can provide interpretable results such as image scores and features to assist the doctor in interpreting the lymph nodes for metastasis. The area under the curve (AUC) of support vector machine (SVM) is 0.9995, and the AUC of CSC is 0.9992.
https://ndltd.ncl.edu.tw/cgi-bin/gs32/gsweb.cgi/login?o=dnclcdr&s=id=%22107NCTU5112002%22.&searchmode=basic