A Hybrid Model for Brain Tumor Classification via Deep Feature Extraction and Optimized Classifier Weighting

Accurate and automated brain tumor classification is crucial for early treatment planning and effective diagnosis. In this study, we proposed a novel multiclass brain tumor classification approach that utilizes deep features extracted from the pre-trained VGG-16 model, to enhance the classification of brain tumor using Support Vector Machine (SVM) and K-Nearest Neighbours (KNN) with majority voting. VGG-16, effectively captured complex and discriminative features from Magnetic Resonance Images. These features are then utilized for classification using SVM and KNN, leveraging majority voting to combine the decisions of both classifiers to enhance the prediction accuracy. The proposed method is evaluated on the benchmark figshare brain tumor dataset, demonstrating superior performance compared to individual classifiers. Achieved 99.54 % accuracy with the proposed model surpassing the state of the art and existing models. The ensemble approach significantly improves robustness, achieving high classification accuracy and minimizing misclassification errors. The results indicate that the combination of deep feature extraction with ensemble learning can effectively address the challenges of multiclass brain tumor diagnosis, offering a reliable tool for medical image analysis.