Brain Tumor Mri Segmentation Using A Hybrid Otsu–Fcm Model

Abstract

The process of segmenting MRI images of brain tumors is one of the essential tasks in image-based medical diagnosis, as it directly contributes to improving the accuracy of determining the tumor's location, shape, and size. However, most segmentation techniques still suffer from poor performance in low-contrast or noisy images due to the overlap of gray levels and the similarity of brain tissues. This study aims to develop an effective hybrid model that combines Otsu’s Method and the Fuzzy C-Means (FCM) algorithm, in order to enhance segmentation accuracy and reduce the impact of noise in medical images. The model was implemented using the MATLAB environment and applied to the BraTS 2023 database, which includes MRI images of 1271 patients. The evaluation methodology included the extraction of multiple quantitative indicators such as the Dice coefficient, Mean Squared Error (MSE), Peak Signal-to-Noise Ratio (PSNR), accuracy, recall, and F1 score. The results showed that the hybrid model outperformed the individual methods; the average Dice coefficient was 0.9577, compared to 0.9262 and 0.8955 for the FCM and Otsu methods, respectively. The model also achieved the lowest mean squared error (MSE = 94.58) and the highest PSNR of 29.17 dB, with a classification accuracy of 98%. These results confirm that the proposed model provides an effective balance between computational accuracy and structural flexibility, making it suitable for application in medical decision support systems, especially in analyzing medium or low-quality images without the need for complex deep models.