がん細胞のタンパク質情報を考慮に入れたSVMによる病理診断システムの可能性

抄録

The cancer cells are known that they have a difference of their growth rate and effect of the medicine even if they are taken from the same site. We can check the progress in the morphological feature extraction using a microscope, but we can’t yet to classify the inherent differences. In this study, we proposed a system that allows for the difficult classification by visual analysis. The method is that first we created a database using a laser scanning cytometer (LSC) from protein information on the cancer cells, and learn the relation between the cancer cells and protein using a support vector machine (SVM). We performed the performance evaluation of recognition system by simulate the unknown cancer cells, and confirmed that there were two kinds of cancer cells that have high recognition rate and slightly inferior rate. Therefore, we report about the experimental verification to investigate that the problem of recognition accuracy.