Noise adaptive wavelet thresholding for speckle noise removal in optical coherence tomography

Abstract

Optical coherence tomography (OCT) is based on coherence detection of interferometric signals and hence inevitably suffers from speckle noise. To remove speckle noise in OCT images, wavelet domain thresholding has demonstrated significant advantages in suppressing noise magnitude while preserving image sharpness. However, speckle noise in OCT images has different characteristics in different spatial scales, which has not been considered in previous applications of wavelet domain thresholding. In this study, we demonstrate a noise adaptive wavelet thresholding (NAWT) algorithm that exploits the difference of noise characteristics in different wavelet sub-bands. The algorithm is simple, fast, effective and is closely related to the physical origin of speckle noise in OCT image. Our results demonstrate that NAWT outperforms conventional wavelet thresholding.

Improved wavelet hierarchical threshold filter method for optical coherence tomography image de-noising

“…After wavelet decomposition, useful signals concentrate on larger amplitude wavelet coe±cients and noise signals concentrate on smaller amplitude wavelet coe±cients. 36 For OCT images, noise intensity is di®erent at di®erent spatial or frequency positions, so the selected wavelet threshold for each decomposition level should also be di®erent. In conventional wavelet threshold de-nosing method, an unchanged global threshold is assessed and then used for each decomposition level.…”

“…[33][34][35][36] The speckle noise in OCT images is multiplicative noise. 34 By taking the logarithmical transform of fðx; yÞ, the nonlinear multiplicative noise could be turned into additive noise.…”

“…But these methods would increase system complexity, meanwhile decrease imaging speed and spatial resolution. 12,16 Software approaches based on post-processing¯ltering techniques, including adaptive Wiener¯ltering, 6 curvelet domain¯ltering, 19,20 contourlet domain¯ltering, 21 Csiszars I-divergence regularization, 22 interval type II fuzzy system, 23,24 regularized image restoration based on speckle characteristics, 25 compressed sensing (CS) reconstruction, [26][27][28] sparse reconstruction, [29][30][31][32] and wavelet domain¯ltering [33][34][35][36][37] can reduce speckle noise by¯l-tering in di®erent transform domain. Wavelet domain¯ltering is widely accepted as a promising method in de-noising for OCT images.…”

“…Large amplitude wavelet coe±cients are regarded as the carrier of useful information, while small amplitude wavelet coe±cients are more likely to be the carrier of noise. 36 Conventional wavelet de-noising algorithm assumes that the level of noise intensity is the same at di®erent sub-bands, so this algorithm uses a global and constant threshold for di®erent wavelet sub-bands. In fact, the speckle noise pattern is uncorrelated and di®erent in OCT images 33 and speckle noise can also be the carrier of useful information.…”

Assessment and removal of additive noise in a complex optical coherence tomography signal based on Doppler variation analysis

“…Second, most of the denoising algorithms that have been developed are applied to the magnitude of OCT signal and assume the noise in magnitude OCT signal to be additive Gaussian [6–9]. However, OCT imaging suffers from both additive noise and multiplicative noise (speckle noise) [18–20], and the additive Gaussian noise model is not valid for magnitude OCT signal, particularly for low signal to noise ratio conditions. In our denoising method, we first generate a map of additive noise through Doppler variation analysis, and then perform local Wiener filtering [21, 22].…”