Automated Artifacts and Noise Removal from Optical Coherence Tomography Images Using Deep Learning Technique

Abstract

Optical Coherence Tomography (OCT) is a popular non-invasive clinical tool for the diagnosis of ocular diseases that provides micron-scale images of ocular pathology in vivo and in real-time. The cross-sectional OCT B-scan of Temporal-Superior-Nasal-Inferior-Temporal (TSNIT) peripapillary retinal profile is widely used to diagnose and monitor glaucoma. However, raw OCT images can be marred by noise and artifacts, especially vitreoretinal interface opacity: this can lead to segmentation error, misinterpretation of retinal thickness measurements and possibly inappropriate glaucoma management. In this study, we designed and trained a U-Net model on OCT B-scans with artifacts, and their corresponding 'artifact-free B-scans'. The U-Net was able to remove the artifacts successfully with better performance in terms of PSNR and SSIM values. The SNR of the OCT scans with speckle noise associated with artifacts has also been improved. To the best of our knowledge, this is the first study where automated vitreous opacity artifact removal has been applied to the TSNIT profile. The performance of the U-net model on measures such as PSNR, SSIM, MAE, and MSE is compared with the state-of-the-art image denoising models. It is observed that the proposed U-Net model performs better as compared to the other models on both parametric and visual evaluations. In the future, this U-Net model could be used to solve automatic retinal layer segmentation errors and assist clinicians in interpreting OCT images in glaucoma diagnosis and monitoring.

Symptomatic vitreous opacity image enhancement using linear and nonlinear transformations

“…In recent years, research for symptomatic vitreous opacity image enhancement has made significant progress in several areas. These studies typically use traditional methods [11][12][13][14][15] and deep learning methods [16][17][18][19][20] to improve the quality and clarity of symptomatic vitreous opacity images. Some studies use traditional methods, such as filtering, to improve the quality of symptomatic vitreous opacity images.…”

“…Although conventional methods perform well in some aspects, they may not be able to fully utilize the advantages of advanced techniques such as deep learning. In the case of deep learning methods [16][17][18][19][20], despite significant progress in image enhancement, a large amount of labeled data is usually required to train the model. Especially for symptomatic vitreous opacity image enhancement, obtaining a sufficient amount of labeling data can be a challenging task.…”

Glaucoma Detection and Feature Visualization from OCT Images Using Deep Learning

“…Machine-based segmentation also often failed due to the presence of speckle noise, [20][21][22] low signal strength, 23 motion artifacts, 24 and vitreous opacity or vitreoretinal interface effects. [25][26][27] Furthermore, mis-segmentation occurs in eyes with advanced glaucoma or retinal pathology (e.g., diabetic retinopathy). 28 The benefit of using CNNs is their feature-agnostic ability, i.e., the ability to classify images by learning the position and scaling the variant structures within the data.…”

“…Machine-based segmentation also often failed due to the presence of speckle noise, 20-22 low signal strength, 23 motion artifacts, 24 and vitreous opacity or vitreoretinal interface effects. 25-27 Furthermore, mis-segmentation occurs in eyes with advanced glaucoma or retinal pathology (e.g., diabetic retinopathy). 28…”

Analysis of Noise Removal Techniques on Retinal Optical Coherence Tomography Images

“…Nahida Akter et al proposed deep learning based method to remove noise from OCT images. The authors created and trained a U-Net model using OCT artifact-filled and artifactfree B-scans for investigation and shown that the U-Net performed better in terms of SSIM and PSNR values in removing the artefacts [11].…”