Evolutionary Model for Brain Cancer-Grading and Classification

Abstract

Brain cancer is a bad disease and affects millions of people in worldwide. Approximately 70% of patients diagnosed with this disease do not survive. The Machine learning is a promising and recent development in this area. However, very limited research is performed in this direction. Therefore, in this research, we propose an evolutionary lightweight model aimed at detecting brain cancer and classification, starting from the analysis of magnetic resonance images. The proposed model named lightweight ensemble combines (weighted average and lightweight combines multiple XGBoost decision trees) is the modified version of the recent Multimodal Lightweight XGBoost. Herein, we provide prediction explain ability by considering the preprocessing of Magnetic Resonance Imaging (MRI) data and the feature extraction (Intensity, texture, and shape). The process in the evolutionary model involves a various step - first, prepare the data, extract important features, and finally, merge together using a special kind of classification called ensemble classification. We evaluate our proposed model using BraTS 2020 dataset. The dataset consists of 285 MRI scans of patients diagnosed with gliomas. The simulation results showed that our proposed model achieved 93.0% accuracy, 0.94 precision, 0.93 recall, 0.94 F1 score, and an area under Receiver Operating Characteristic Curve (AUC-ROC) value of 0.984. The efficient results demonstrate the effectiveness of our proposed model for brain tumor grading and classification using four grades. The efficient results show the potential of our proposed approach as a valuable tool for early diagnosis and effective treatment planning of brain tumors. Finally, the proposed model holds promise for aiding in early cancer diagnosis and treatment.