Classification of Liver Cancer Images Based on Deep Learning

With the rapid development of deep Learning, research into Deep Learning is being increasingly applied to the field of medical imaging. Liver cancer, which has one of the highest rates of morbidity and mortality in the world, is a great threat to people’s health. This study aims to apply Convolutional Neural Networks in the grade classification of liver cancer images. DCE- MRI and DWI, two modes of hepatocellular carcinoma images, are originally used separately to grade liver cancers. We combine these two image modes to improve the prediction accuracy. The study finds that the features of the two modes can be complementary, and can improve the grading classification of liver cancer. From comparing the two methods of traditional Machine Learning and Deep Learning, the study demonstrates that the grading accuracy by Machine Learning from the integration of features is 87.8, while the accuracy rate from Deep Learning reaches 90.5. The improvement in grading accuracy is due to Deep Learning can extract the appropriate features. In addition, the presence of micro vascular invasion is an important factor for the recurrence of liver cancer after surgery. The experiment also uses Deep Learning to predict micro vascular invasion. The accuracy of the ADC map prediction reached 69.2, it demonstrates that liver cancer images can also predict micro vascular invasion to a certain extent.