Integrating Pixel-Based Algorithms for Area Measurement in Brain Tumor Classification

Abstract

Diagnosing brain cancers in medicine necessitates an examination utilizing magnetic resonance imaging (MRI). picture processing techniques in the medical domain are integral to computed tomography detection in MRI due to their excellent picture fidelity and little radiation exposure. Nonetheless, there remain deficiencies in the interpretation, analysis, and imaging of brain tumors in detection. This study seeks to identify brain tumors to ascertain their size and extent by a pixel-based methodology. The dataset utilized originates from Cipto Mangunkusumo Hospital in Jakarta and comprises T1 contrast and BMP sequences. The research procedure will employ many methodologies, including active contours, Otsu's method, and a combination of techniques, with comparisons utilizing the MRI MicroDicom viewer. The image testing phase utilizing Matlab and Python with thirteen image datasets. The findings from this study, which involved segmentation and extraction techniques to quantify the area of brain tumors using a pixel-based approach, indicate that the combined method outperforms alternative methods by achieving superior accuracy of 99%. Other methods fail to attain this level of accuracy, and the combined method also demonstrates optimal error differentiation in template matching.