<![CDATA[Abstract - Accurate segmentation of brain tumor images plays a crucial role in supporting precise diagnosis and treatment planning. Brain tumor image segmentation requires an effective image enhancement process to distinguish healthy and abnormal tissues. While deep learning methods are advancing, Histogram Equalization (HE) and Contrast-Limited Adaptive Histogram Equalization (CLAHE) techniques remain essential for lightweight and efficient preprocessing. The dataset consists of 100 brain MRI images, including 50 normal and 50 tumor cases. The evaluation employs quantitative metrics including Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), Mean Square Error (MSE), and processing time. Experimental results indicate that HE achieves an accuracy of 85% (PSNR = 18.4 dB, SSIM = 0.71, MSE = 0.015, average processing time = 0.86 s), while CLAHE achieves 95% accuracy (PSNR = 24.9 dB, SSIM = 0.89, MSE = 0.009, average processing time = 1.12 s). These findings indicate that CLAHE provides clearer and more detailed image improvements compared to HE. Keywords: Brain Tumor, Preprocessing, Histogram Equalization (HE), Contrast-Limited Adaptive Histogram Equalization (CLAHE)]]>
Copyrights © 2025
