A Learnable Despeckling Framework for Optical Coherence Tomography Images

TL;DR

This paper introduces LDF, a learnable framework that automatically selects the most effective despeckling filter for OCT images, enhancing image quality without hardware changes.

Contribution

The proposed LDF framework is a novel, expandable neural network system that learns to choose optimal despeckling filters for OCT images based on a learned figure of merit.

Findings

LDF effectively identifies the best despeckling filter for OCT images.

The framework improves image quality by selecting appropriate filters.

LDF demonstrates adaptability across various filter categories.

Abstract

Optical coherence tomography (OCT) is a prevalent, interferometric, high-resolution imaging method with broad biomedical applications. Nonetheless, OCT images suffer from an artifact, called speckle which degrades the image quality. Digital filters offer an opportunity for image improvement in clinical OCT devices where hardware modification to enhance images is expensive. To reduce speckle, a wide variety of digital filters have been proposed, selecting the most appropriate filter for each OCT image/image set is a challenging decision. To tackle this challenge, we propose an expandable learnable despeckling framework, we called LDF. LDF decides which speckle reduction algorithm is most effective on a given image by learning a figure of merit (FOM) as a single quantitative image assessment measure. The architecture of LDF includes two main parts: (i) an autoencoder neural network, (ii) filter classifier. The autoencoder learns the figure of merit based on the quality assessment measures obtained from the OCT image including.Subsequently, the filter classifier identifies the most efficient filter from the following categories: (a) sliding window filters including median, mean, and symmetric nearest neighborhood, (b) adaptive statistical based filters including Wiener, homomorphic Lee, and Kuwahara, and, (c) edge preserved patch or pixel correlation based filters including non-local mean, total variation, and block matching 3D filtering.