Wavelet-Based MRI Brain Image Analysis for Tumor Detection and Classification Using SVM and Random Forest

Abstract

Brain tumor detection and classification in MRI image data is a significant and challenging task in medical image analysis. This paper presents an efficient method that integrates Support Vector Machine (SVM) and Random Forest algorithms, developed with a Graphical User Interface (GUI) in MATLAB. The interface allows flexible combinations of segmentation, filtering, and other techniques to achieve optimal results. The proposed approach starts with preprocessing steps, including Gaussian filtering and morphological operations, followed by feature extraction using Discrete Wavelet Transform (DWT) and Continuous Wavelet Transform (CWT). Principal Component Analysis (PCA) is applied to decrease the feature set for more effective classification. The extracted first and second-order features are used to train the kernel SVM. Classification is then performed using both SVM and Random Forest to improve accuracy. Watershed segmentation is applied for precise tumor localization. The hybrid model achieved a classification accuracy of approximately 93% using only SVM, and an improved accuracy of 96% when combining SVM with Random Forest, outperforming traditional approaches in both accuracy and computational efficiency. Benchmark evaluation plays a crucial role in enhancing both accuracy and reliability. To ensure precise tumor localization, watershed segmentation was applied. These findings indicate that the proposed method offers a reliable, automated solution for brain tumor detection, demonstrating significant potential for integration into clinical diagnostic workflows.