Machine Learning Techniques to Classify Brain Tumor

Automated identification of defects in medical imaging, such as brain tumors in Magnetic Resonance Imaging (MRI), is essential for diagnostic applications. This research paper focuses on diagnosing normal and abnormal brain tumors, including meningioma, glioma, and pituitary tumors. To address the challenge of large data volumes, automated methods using machine learning algorithms are developed. Gradient Boosting (GB), Adaptive Boosting (ADA), and Support Vector Machine (SVM) algorithms are employed to enhance accuracy and reduce diagnosis time. The study utilizes a publicly available dataset and applies feature extraction techniques, including Fast Fourier Transform (FFT), K-means clustering, and Tamura texture analysis. Additionally, dataset dimensionality reduction techniques are employed. The reported accuracy rates for normal and abnormal cases are 92.6% for GB, 86% for ADA, and 82% for SVM. For specific tumor types, the accuracy rates are 67.9
^%
for GB, 65.3% for SVM, and 59.6% for ADA. Based on superior performance, GB is recommended as the algorithm of choice. This research paper demonstrates the effectiveness of these techniques in automating brain tumor detection and improving diagnostic accuracy.”