Improvement of the Fuzzy Clustering Algorithm for Medical Image Segmentation and Denoising Using Fast Bilateral Filtering

Abstract

Medical image analysis often faces challenges due to noise, which can obscure crucial diagnostic information and hinder precise segmentation. Traditional denoising methods often fail to effectively suppress noise while preserving image details, resulting in blurred or overly smoothed outputs. To address this, we propose an improved fuzzy clustering algorithm that introduces an innovative integration of fast bilateral filtering and adaptive parameter tuning, offering superior noise reduction and enhanced medical image segmentation accuracy. Our method introduces a novel combination of fast bilateral filtering and an enhanced fuzzy C-means (FCM) algorithm, which effectively balances noise suppression and detail preservation, outperforming existing methods in both accuracy and efficiency. The fast bilateral filter efficiently preserves edge details while reducing spatial and local intensity variations, serving as a robust preprocessing step that mitigates noise-induced clustering errors. Additionally, we introduce an innovative strategy that calculates the absolute difference between the original and filtered images to enhance clustering accuracy in noisy environments. To improve convergence speed and computational efficiency, we refine the FCM objective function by incorporating a logarithmic summation of membership degrees from previous iterations, reducing iteration counts and accelerating convergence. Finally, we apply sharpening and median filtering techniques to refine segmentation outputs and enhance detail clarity. Experimental results on benchmark medical images demonstrate that our proposed method achieves superior noise suppression, improved segmentation accuracy, and faster convergence compared to conventional FCM and recent denoising techniques.