Adaptive Feature Selection for Brain Tumor Classification in MRI Images using Genetic Algorithm Polar Bear Optimization and SVM

Abstract

Brain tumor detection is a critical task in medical imaging, and the integration of advanced technologies has significantly improved the accuracy and efficiency of diagnosis. Accurate identification of tumor types from MRI scans is essential for effective treatment planning and patient management. The proposed method aims to optimize the feature selection process to enhance classification performance while minimizing computational complexity. Initially, a diverse set of features extracted from MRI images is subjected to the GA-PBO hybrid algorithm, which intelligently selects the most discriminative features relevant to brain tumor classification. The GA-PBO algorithm leverages the exploration-exploitation capabilities of genetic algorithms and the powerful search mechanisms inspired by polar bear behavior to efficiently identify the optimal feature subset. Subsequently, the selected features are fed into an SVM classifier to perform tumor classification based on their distinctive patterns and characteristics. SVM is well-suited for handling high-dimensional data and has been widely employed in medical image analysis tasks due to its robustness and effectiveness in classification. Experimental evaluations are conducted using publicly available MRI datasets containing various types of brain tumors. Evaluation metrics, such as accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC-ROC), are used to assess the efficacy of the proposed method. The performance of the proposed approach is compared against current feature selection methods and classification techniques. The results show that the proposed adaptive feature selection approach achieves superior classification performance compared to baseline methods, effectively distinguishing between different tumor types with high accuracy and reliability. Furthermore, the computational efficiency of the proposed method is demonstrated, making it suitable for real-time applications in clinical settings. Overall, the proposed GA-PBO-based adaptive feature selection approach, integrated with SVM classification, offers a promising solution for accurate and efficient brain tumor classification in MRI images, thereby facilitating timely diagnosis and treatment planning for patients with brain tumors.