LAUSR

Simple Summary Despite the highly aggressive nature of glioblastoma multiforme (GBM), survival time is in practice highly variable, and some of the patients remain stable for several years after treatment. The aim of this study was to develop a machine learning method that could precisely predict survival time of GBM patients. To do so, we integrated multi-modal MRI with non-supervised and supervised machines. We first identified compartments of the tumor then extracted their features. Then relevant useful features were selected by Random Forest-Recursive Feature Elimination (RF-RFE) to feed into Gradient Boosting Machine Algorithm with the aim of classifying GBM patients. By selecting the most relevant features, multi-modality MRI with tumor segmentation provided valuable independent and complete features to feed a machine learning model. Additionally, advanced machine-learning methods such as RF-RFE and GBoost are powerful tools for data mining. Hand-crafted feature-based methods have shown promising results, but there is no systematic way to determine survival-related hand-crafted features and existing methods mostly rely on experience. Abstract Despite advances in tumor treatment, the inconsistent response is a major challenge among glioblastoma multiform (GBM) that lead to different survival time. Our aim was to integrate multimodal MRI with non-supervised and supervised machine learning methods to predict GBM patients’ survival time. To this end, we identified different compartments of the tumor and extracted their features. Next, we applied Random Forest-Recursive Feature Elimination (RF-RFE) to identify the most relevant features to feed into a GBoost machine. This study included 29 GBM patients with known survival time. RF-RFE GBoost model was evaluated to assess the survival prediction performance using optimal features. Furthermore, overall survival (OS) was analyzed using univariate and multivariate Cox regression analyses, to evaluate the effect of ROIs and their features on survival. The results showed that a RF-RFE Gboost machine was able to predict survival time with 75% accuracy. The results also revealed that the rCBV in the low perfusion area was significantly different between groups and had the greatest effect size in terms of the rate of change of the response variable (survival time). In conclusion, not only integration of multi-modality MRI but also feature selection method can enhance the classifier performance.