A Robust Framework for De-speckling of Optical Coherence Tomography Images

Abstract

Recently, Optical Coherence Tomography (OCT) is emerging as a important diagnostic tool in medical application. OCT is widely used in detection of vision-related diseases. The analysis of retinal OCT images are very difficult due to speckle noise. The characteristic nature of this noise is multiplicative. A Number of de-speckling methods was proposed in the last few decades. All the existing de-speckling methods reduce the speckle noise, but these methods are not able to preserve the structure of the OCT image during de-speckling process. This  paper propose a robust framework using wiener filter, soft thresholding  and a weighted guided filter along with wavelet decomposition for  the purpose of speckle noise reduction. The main contribution of this research is to remove speckle noise and preserve the structure of  OCT image during de-noising process. In the first step of proposed framework, the original image is filtered by the wiener filter. After that a logarithmic transformation is used for the conversion of multiplicative noise into additive noise. Discrete wavelet transform (DWT) decompose the image into its constituents.  Soft  thresholding  and weighted guided filter is used for high frequency sub-image and low frequency sub-image part  respectively. In the last inverse DWT and antilog transformation are applied to acquire the de-noised image. Two different experiments are performed one on real OCT Images and other on natural images, to demonstrate the usefulness of the proposed framework.