Speckle noise reduction in spectral domain optical coherence tomography retinal images using fuzzification method

摘要

In the present work the focus is on speckle noise reduction in Spectral Domain Optical Coherence Tomography (SDOCT) of retinal images using fuzzification technique. OCT in frequency domain is emerging as attractive technique since it simultaneously measures multiple wavelength of reflected light across the spectrum by acquiring volumetric data for the diagnosis of retinal diseases. Even with sophisticated technology SDOCT images has speckle noise and it is about 1–2 μm which degrades the quality of image. In the existing literature, the speckles are denoised by using nonlinear direction preserving digital filters and averaging approaches. In all these methods, there are practical limitations such as loss of detailed resolution, reduction of frame rates and other shortcomings. In this paper, fuzzification method is proposed which uses the maximum fuzzy entropy principle to reduce speckle noise in retinal images. To validate the results SDOCT images were collected from 30 different subjects on 15 different pathological conditions. The results show that the proposed method works well and has enhanced image quality by improving Signal to Noise Ratio (SNR).