Optimizing MRI Image Analysis for Brain Tumor Detection: A GLCM-Enabled U-Net Approach

Abstract

This study presents a thorough methodology for precisely identifying and segmenting brain tumors from magnetic resonance imaging (MRI) data. The study starts with the collection of a heterogeneous database of MRI images, which includes different tumor sizes, locations, and types. Using sophisticated image processing methods like normalization and noise reduction improves the quality of MRI pictures. For effective data representation, the methodology combines a patch-based extraction technique with Gray-Level Co-occurrence Matrix (GLCM) feature extraction. With deep learning-based segmentation using a U-Net architecture, the system exhibits reliable and precise automated brain tumor detection. The efficacy of the suggested methodology is demonstrated by thorough performance evaluations that include quantitative metrics and qualitative assessments on training and testing datasets. This research advances medical imaging and computer-aided diagnosis, giving medical personnel an important tool for brain tumor early detection and treatment planning.