Glioma is the most common brain intracranial malignancy, which accounts for about 80% of malignant brain tumors in adults and its median survival rate is 12 months. In clinical, how to accurately predict the glioma overall survival (GOS) is a crucial work and it will be beneficial to monitor tumor progression, execute surgery as well as plan radiotherapy and follow-up studies. However, the glioma generally has highly heterogeneity degrees in the histological tumor sub-regions. we propose a comprehensive multi-modality MRI radiomics way of predicting the GOS. Different features are proposed committing to different image modalities. A feature selection strategy is applied for the optimal features and then random forest is contributed to the classification of short-survivors and long-survivors. With the performance evaluation criteria, our model showed promising classification ability for the brain tumor.
This abstract and the presentation materials are available to members only; a login is required.