OCT-GAN: Single Step Shadow and Noise Removal from Optical Coherence Tomography Images of the Human Optic Nerve Head

TL;DR

OCT-GAN is a fast, single-step deep learning method that effectively removes noise and shadows from OCT images, enhancing image quality for better diagnosis while reducing acquisition time and hardware needs.

Contribution

This paper introduces OCT-GAN, a novel single-step GAN-based algorithm that simultaneously removes noise and shadows from OCT images with high efficiency and improved image quality.

Findings

AGM increased by 57.2% over state-of-the-art

PSNR, CNR, SSIM improved significantly

Reduces image acquisition time and hardware costs

Abstract

Speckle noise and retinal shadows within OCT B-scans occlude important edges, fine textures and deep tissues, preventing accurate and robust diagnosis by algorithms and clinicians. We developed a single process that successfully removed both noise and retinal shadows from unseen single-frame B-scans within 10.4ms. Mean average gradient magnitude (AGM) for the proposed algorithm was 57.2% higher than current state-of-the-art, while mean peak signal to noise ratio (PSNR), contrast to noise ratio (CNR), and structural similarity index metric (SSIM) increased by 11.1%, 154% and 187% respectively compared to single-frame B-scans. Mean intralayer contrast (ILC) improvement for the retinal nerve fiber layer (RNFL), photoreceptor layer (PR) and retinal pigment epithelium (RPE) layers decreased from 0.362 \pm 0.133 to 0.142 \pm 0.102, 0.449 \pm 0.116 to 0.0904 \pm 0.0769, 0.381 \pm 0.100 to 0.0590 \pm 0.0451 respectively. The proposed algorithm reduces the necessity for long image acquisition times, minimizes expensive hardware requirements and reduces motion artifacts in OCT images.