Enhancing Brain Tumor Diagnosis: Comparative Analysis of Random Forest and Support Vector Machine for Brain Tumor Mri Classification

Segmentation and classification of brain tumors from magnetic resonance imaging (MRI) are significant issues in medical imaging, since accurate delineation of tumor volumes has immediate implications on diagnosis, treatment planning, and patient outcome. Manual delineation by a radiologist, although traditional, is operator dependent and time-consuming. In this work, an automated pipeline including image Preprocessing, segmentation and machine learning-based classification is proposed. Preprocessing consists of Gaussian filtering, skull stripping, histogram equalization, and intensity normalization followed by adaptive thresholding-based segmentation, edge detection, morphological processing, and clustering. Classification of the segmented regions is accomplished using Random Forest (RF) and Support Vector Machine (SVM) models. The experiments were conducted on an available dataset comprising 5,712 training and 1,311 test MRI scans across four classes: glioma, meningioma, pituitary adenoma, and no tumor. Overall accuracy of 0.92 was achieved by Random Forest classifier having F1-score = 0.926}, and SVM yielded 0.87 having F1-score =0.879. Both values indicate that RF outperformed SVM regarding overall generalizability and hardness, particularly in detecting glioma and pituitary adenoma. Compared to the available literature, the proposed system demonstrates comparable performance and maintains computational efficiency and interpretability.