A two-step filtering mechanism for speckle noise reduction in OCT images

Abstract

Optical coherence tomography (OCT) has been widely adopted in various areas for its noninvasive and high-resolution properties. Due to it low-coherence interferometry nature, however, OCT inevitably suffers from speckle noise, which hides structural information in OCT images and thus degrades the clinical diagnosis accuracy. So far various algorithms have been proposed for OCT speckle denoising, yet few studies have evaluated the influences of speckle noise distributions on the denoising effects. This paper studies the influences of speckle noise distributions in OCT despeckling process, and a twostep filtering mechanism, namely, Augmented Lagrange function minimization and Rayleigh alpha-trimmed filtering (AR) scheme, is proposed for OCT speckle noise reductions. The speckle noise distribution models are established and estimated first, and then two different filtering mechanisms are designed for those noise distributions, respectively. Simulations with both synthetic and OCT images are conducted to verify the effectiveness of the AR scheme. Results show that AR method can suppress OCT speckle noises effectively, and outperforms the best existing methods in different cases, yet with less time computations.