Evaluation of different denoising algorithms for OCT image denoising

Abstract

Optical coherence tomography (OCT) is a useful non-invasive optical tool for imaging various biological tissues. As OCT imaging is based on interferometry, speckle noises are inherent and can degrade the quality of OCT image. The objective of this study was to evaluate the effectiveness of conventional denoising algorithms for OCT image denoising and for improving image quality. OCT images of human skin were obtained from a swept source OCT of 1300 nm. Three image denoising algorithms, including median filtering, mean filtering and Gaussian bilateral filtering, were applied for denoising OCT images of different quality. Five quality evaluation criteria, including signal to noise ratio (SNR), equivalent number of looks (ENL), contrast-to-noise ratio (CNR), cross correlation (XCOR), and peak signal to noise ratio (PSNR) were used for comparing the effectiveness of each denoising process. In terms of improving local contrast, three denoising algorithms showed similar effect. In terms of the equivalent views, Gaussian bilateral filtering algorithm showed the most significant increase and therefore caused certain degrees of blurry. For signal to noise ratio, all three denoising algorithms showed improvement while Gaussian bilateral filtering algorithm had better protection effect of the effective information and edge of the original image. Gaussian bilateral filtering algorithm provides better denoising outcomes for OCT image processing.