Efficient Deep Learning Approach for Brain Tumor Detection and Segmentation using Random Forest in U-Net over SVM

Abstract

Brain tumor detection and segmentation in MRI images are critical for accurate diagnosis and treatment planning. Traditional methods rely heavily on manual feature extraction, which is time-consuming and prone to inconsistencies. This study presents an innovative deep learning-based framework that leverages RF, SVM, and U-Net architecture to automate tumor detection and segmentation. The proposed method enhances medical image processing through optimized preprocessing techniques, transfer learning, and feature extraction. The evaluation of the system shows a classification accuracy of 87.5% and a segmentation dice coefficient of 0.70, demonstrating superior performance compared to existing methods. The study highlights the potential of deep learning models in medical imaging, offering a reliable and scalable solution for brain tumor analysis.