Abstract:
Brain tumor is a serious problem when it is not diagnosed. Different levels of tumors are identified this decade. The severity of tumor can be reduced if it is identified in its early stage. The most important challenge of identifying tumor is its shape and location in the brain tissue. This paper proposes different technique for extracting features for identifying the tumor by reducing the computation time. The identified features are classified using Random Forest classifier. Our proposed framework is experimented on a challenging BRATS 2015 dataset. The investigational results obtained by the proposed method shows better in terms of qualitative metrics such as Dice Score, Positive Predictive Value (PPV) and Sensitivity with a little reduction in computation time when compared to other recent methods.