Enhancing Brain Tumor Classification with VGG-19 in Deep Learning Paradigms

The primary and pivotal stage in patient care lies in accurately categorizing brain tumors. This critical process not only identifies potentially life-threatening abnormalities but also lays the groundwork for tailoring effective treatment plans essential for a patient's recovery journey. The proposed methodology entails a structured approach comprising segmentation, classification, feature extraction, and preprocessing. These sequential steps serve as the foundational framework for comprehensively analyzing the data sourced from the Figshare dataset. In the initial phase, photos undergo preprocessing utilizing the Gaussian filter method. Subsequently, the preprocessed images are subjected to segmentation employing the DU-Net method. Following segmentation, feature extraction is performed on the delineated segments. For this task, DesNet-121 is employed to extract feature data. Finally, leveraging the resultant features, data classification is executed. This systematic approach ensures a comprehensive analysis of the data while maintaining consistency and accuracy throughout the process. In the final stage, a VGG-19 deep learning model is employed to classify the MRI pictures into distinct groups. This proposed model is then simulated on a dataset, and its performance metrics, including accuracy, precision, recall, and F1-score, are thoroughly evaluated. The results indicate significant enhancements in brain tumor categorization and detection, affirming the efficacy and reliability of the suggested model for clinical applications. The testing outcomes underscore the capability of the recommended strategy to achieve exceptional accuracy, reaching an impressive 98.15%.