Reproducibility Study of Prediction of Brain Tumors Response to Bevacizumab Treatment

Abstract

Glioblastoma Multiform (GBM) is the most common cause of cancer death in both men and women. Bevacizumab is a recent therapy for stopping the tumor growth. The purpose of this paper is to present our reproducibility study of predicting response of the brain tumors to Bevacizumab treatment. This method allows physicians to select most effective treatment plans. We take two image series of patients before and after the treatment. After constructing Eigen images, we extract their statistical histogram features and then use regression analysis to develop a predictive model. Predictive models of response are developed with large regression coefficients (maximum R2=0.8). This method is dependent on the operator. To decrease the operator’s role, this method is repeated four times for each patient. Then, the average of the achieved results is used for regression analysis. As a result, the regression coefficient increases (maximum R=0.86). The result of this approach is compared to that of a previous work at the University of Tehran showing excellent reproducibility of the proposed method. 