DeshadowGAN: A Deep Learning Approach to Remove Shadows from Optical Coherence Tomography Images

Abstract

Purpose: To remove retinal shadows from optical coherence tomography (OCT) images of the optic nerve head (ONH). Methods: 2328 OCT images acquired through the center of the ONH using a Spectralis OCT machine for both eyes of 13 subjects were used to train a generative adversarial network (GAN) using a custom loss function.Image quality was assessed qualitatively (for artifacts) and quantitatively using the intralayer contrast a measure of shadow visibility ranging from 0 (shadow-free) to 1 (strong shadow) and compared to compensated images. is was computed in the Retinal Nerve Fiber Layer (RNFL), the Inner Plexiform Layer (IPL), the Photoreceptor layer (PR) and the Retinal Pigment Epithelium (RPE) layers. Results: Output images had improved intralayer contrast in all ONH tissue layers. On average the intralayer contrast decreased by 33.7 ± 6.81%, 28.8 ± 10.4%, 35.9 ± 13.0%, and 43.0 ± 19.5% for the RNFL, IPL, PR, and RPE layers respectively, indicating successful shadow removal across all depths. is compared to 70.3 ± 22.7%, 33.9 ± 11.5%, 47.0 ± 11.2%, 26.7 ± 19.0% for compensation. Output images were also free from artifacts commonly observed with compensation. Conclusions: DeshadowGAN signi cantly corrected blood vessel shadows in OCT images of the ONH. Our algorithm may be considered as a pre-processing step to improve the performance of a wide range of algorithms including those currently being used for OCT image segmentation, denoising, and classi cation. Translational Relevance: DeshadowGAN could be integrated to existing OCT devices to improve the diagnosis and prognosis of ocular pathologies.

Diagnosis of Choroidal Disease With Deep Learning-Based Image Enhancement and Volumetric Quantification of Optical Coherence Tomography

“…22 , 23 Notably, our preprocessing method works on a single OCT volume obtained without any special data acquisition protocols, and thus offers a significant practical advantage in that it can be applied widely and retrospectively to existing databases to gain new insights into the vasculature of the choroid in health and disease. Recently, other techniques addressing speckle noise 24 – 27 and shadow artifact 28 have been proposed and can also be used as alternatives to our methods in the processing pipeline to achieve the quantitative analysis.…”

Deepfakes in Ophthalmology

“…We found 20 published implementations of GANs specific to ophthalmology. Of these, 11 manuscripts synthesized fundus images, 9,19,20,23,32,37,[40][41][42][43][44] 6 manuscripts synthesized OCT images, [27][28][29][45][46][47] 2 manuscripts synthesized fluorescein angiography images, 48,49 and 1 manuscript synthesized infrared images 21 (Table 2). The majority of GANs were proof-ofconcept studies demonstrating feasibility of generating realistic-appearing synthetic images, specific implementations of GANs were published in 9 for diagnosis of ophthalmic diseases, including diabetic retinopathy (DR), 9,20,32,40 glaucoma, 28,45 age-related macular degeneration (AMD), 19,46 and meibomian gland dysfunction.…”

“…Second, GANs are often used to improve signal quality or fill in missing information in an image; however, the resulting "improved" image might obscure the presence of real pathology that would have been visible without the artifact or on a better-quality scan/image. 47 Additionally, they can produce so-called image hallucinations, i.e. the addition of image features not actually present, which may or may not be useful.…”

Applying a Pix2Pix Generative Adversarial Network to a Fourier-Domain Optical Coherence Tomography System for Artifact Elimination

“…The quantitative measurements of the DL method showed an improvement in the mean signal-to-noise ratio, mean contrast-to-noise ratio, and mean structural similarity index. In addition, the DL method was used to extract capillarylevel angiograms from a single OCT volume [14] and detect retinal nerve fiber layer segmentation errors on SD-OCT [15]. Further DL role in medical imaging was showed in [16] for automated age-related macular degeneration (AMD) detection by utilizing support vector machine (SVM), AlexNet, GoogLeNet, and Inception-ResNet for AMD detection while a block-matching and 3-Dimension filter (BM3DF), a hybrid median filter (HMF), and an adaptive wiener filter (AWF) were used to denoise the OCT images.…”