Enhancing Colonoscopy Image Quality Through Multi-Step Computational Pre-Processing Techniques

Colonoscopy is a crucial procedure for gastrointestinal diagnostics, providing direct visualization of the colon's internal structure. The quality of acquired colonoscopy images significantly impacts diagnostic accuracy and treatment planning. This study focuses on enhancing colonoscopy image quality through computational multi step image processing techniques aimed at enhancing colonoscopy image quality and interpretability. The methodology involves multi step strategy for evaluating various noise reduction filters including Gaussian, bilateral, and hybrid bilateral-Gaussian filters, along with Contrast Limited Adaptive Histogram Equalization (CLAHE) and Unsharp Masking techniques. Evaluation metrics such as Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM) are used to quantify the efficacy of these techniques. The study employs the CVC Clinic DB dataset for experimentation, ensuring clinical relevance and diversity in the images analyzed. Results from ablation studies and quantitative analyses highlight the effectiveness of specific preprocessing techniques in preserving image details, enhancing contrast, and sharpening edges. In first step, the hybrid bilateral-Gaussian filter achieved as a suitable noise reduction filter, followed by CLAHE and edge enhancement using Unsharp masking. The PSNR and SSIM values from the first to third step shows increase of 2.88% (from 37.86 dB to 38.95 dB) and 1.56% (from 0.96 to 0.975) respectively. The study's findings contribute to advancing gastrointestinal diagnostics, aiding in more accurate diagnoses, treatment planning, and patient outcomes