Classification of Low-grade and High-grade Glioma using Multiparametric Radiomics Model

Glioma is the most common primary central nervous system malignancy. Accurate grading of glioma before surgery is essential for the patient's next treatment planning and prognosis. In recent s ears, radiomics has been increasingly popular in the field of image diagnosis, providing new ideas and techniques for conventional image interpretation modes. By mining the deep morphological and physiological features of tumor tissue inside the image, by using information gain and LASSO to select features, the SVM classifier classifies the high and low levels of glioma according to the selected features, and finally obtains that the information gain mechanism can be better selected to have a significant degree of discrimination. Features and achieves an accuracy of 0.91 and an AUC value of 0.862.