Efficient Noise Removal From Preterm Baby Retinopathy Images Using Various Filtering Approaches

Abstract

For healthcare practitioners to perform reliable and precise assessments, assuring early detection and appropriate treatment of retinal illnesses, high-quality, noise-free images, is essential. An essential step in improving the reliability and precision of medical diagnoses is the reduction of noise from premature newborns' retinopathy photos. Effective noise removal can be accomplished by using various filters and methods. In recent days, research has been on the effectiveness of noise removal methods is available, specifically the homomorphic filter (HF), laplacian of gaussian (LOG) filter and adaptive filter (AF), with a focus on improving the clarity of retinopathy photos in preterm infants. The authors carefully compared the results with those of homomorphic, LOG, and adaptive filters via thorough testing and assessment criteria such as mean squared error (MSE), peak signal-to-noise ratio (PSNR), and Structural similarity index (SSIM). Applying the LOG filter produced better outcomes for each studied output parameter, producing MSE of 0.000119, PSNR of 42.34and SSIM of 0.998, respectively. The tool used for execution is Python.