Enhanced Brain Tumor Detection from MRI Scans Using Frequency Domain Features and Hybrid Machine Learning Models

Abstract

This research proposes a machine learning-based approach to enhance the accuracy of brain tumor detection by incorporating advanced feature extraction techniques. Texture and shape information, which are critical for precise tumor characterization, were extracted from Magnetic resonance imaging (MRI) scans using Gabor and Radon features. The dataset used consists of 3,160 brain tumor images, categorized into three types of brain tumors and one category representing no tumor. Four classifiers were employed for classification: Linear Discriminant Analysis (LDA), k-Nearest Neighbors (KNN), Support Vector Machine (SVM), and AdaBoost. The results demonstrate that the recognition accuracies for Radon, Gabor, and combined features vary across classifiers. KNN achieved the highest accuracy of 95.50% with Radon features, SVM attained 96.65% with Gabor features, and SVM reported the best overall accuracy of 98.75% with combined features.