Enhanced CT Cancer Image Segmentation Using 2D-STAMF And 2D ACBHI Algorithms with Heuristic Hybrid Fuzzy C-Means Clustering

Abstract

In the recent past, the advancement of medical imaging techniques has underscored the critical need for robust image preprocessing and segmentation algorithms to enhance diagnostic accuracy, particularly in CT cancer imaging. This study presents a comprehensive approach encompassing image restoration and enhancement, followed by precise segmentation using advanced clustering techniques. For image restoration, we introduce the 2D Spatial Temporal Adaptive Median Filter (2D-STAMF), which effectively reduces noise while preserving essential image details. This method is benchmarked against existing algorithms such as the 2D Adaptive Median Filter, 2D Gaussian Filter, and 2D Adaptive Spatial Filter, utilizing metrics including Peak Signal-to-Noise Ratio (PSNR), Mean Squared Error (MSE), and Entropy for comparative analysis. In the image enhancement phase, the proposed 2D Adaptive Contrast Brightness Histogram Improvement (2D ACBHI) algorithm is employed, enhancing image contrast and brightness more effectively than Contrast Limited Adaptive Histogram Equalization (CLAHE), 2D Adaptive Mean Adjustment, and Edge Preservation CLAHE, as evaluated by Structural Similarity Index (SSIM) and Absolute Mean Brightness Error. Subsequently, for CT cancer image segmentation, we develop the Heuristic Hybrid Fuzzy C-Means Clustering (HHFCM) combined with Adaptive Mean Thresholding (AMT), termed as HHFCM-AMT. This segmentation approach is compared against K-Means Clustering, Fuzzy C-Means Clustering, and Fast FCM, using parameters such as Gradient Clusters, K values, and Intensity Pixels. Experimental results demonstrate that the proposed methodologies significantly outperform existing techniques, achieving higher accuracy and reliability in CT cancer image segmentation, thereby validating the efficacy of the integrated preprocessing and segmentation framework.