Reduction of Speckle Noise in SDOCT Retinal Images by Fuzzification and Anisotropic Diffusion Filtering

Abstract

Spectral Domain Optical Coherence Tomography (SDOCT) is a well accepted clinical standard for diagnosing and monitoring the pathological changes in optic disc and retinal layers. In recent times, even though spectral domain detection technique has proved higher potential in terms of extremely high sensitivity and image acquisition speed, the major shortcomings is existence of speckle and its effect on interpretation and diagnosis. In this paper, image enhancement is performed using fuzzification method followed by anisotropic diffusion filtering for effective reduction of speckle noise. Fuzzification technique uses the maximum fuzzy entropy principle. For the removal of speckles of size 2-3µm, anisotropic diffusion filtering is performed. On an average Structural Similarity Index Measure (SSIM) is calculated to be 0.9997 which proves that the image quality has been significantly improved. Moreover edges are perfectly preserved in this filtering technique.