Journal article
Employing texture loss to denoise OCT images using generative adversarial networks
Biomedical Optics Express, Vol.15(4), pp.2262-2280
2024
PMID: 38633090
Abstract
OCT is a widely used clinical ophthalmic imaging technique, but the presence of speckle noise can obscure important pathological features and hinder accurate segmentation. This paper presents a novel method for denoising optical coherence tomography (OCT) images using a combination of texture loss and generative adversarial networks (GANs). Previous approaches have integrated deep learning techniques, starting with denoising Convolutional Neural Networks (CNNs) that employed pixel-wise losses. While effective in reducing noise, these methods often introduced a blurring effect in the denoised OCT images. To address this, perceptual losses were introduced, improving denoising performance and overall image quality. Building on these advancements, our research focuses on designing an image reconstruction GAN that generates OCT images with textural similarity to the gold standard, the averaged OCT image. We utilize the PatchGAN discriminator approach as a texture loss to enhance the quality of the reconstructed OCT images. We also compare the performance of UNet and ResNet as generators in the conditional GAN (cGAN) setting, as well as compare PatchGAN with the Wasserstein GAN. Using real clinical foveal-centered OCT retinal scans of children with normal vision, our experiments demonstrate that the combination of PatchGAN and UNet achieves superior performance (PSNR = 32.50) compared to recently proposed methods such as SiameseGAN (PSNR = 31.02). Qualitative experiments involving six masked clinical ophthalmologists also favor the reconstructed OCT images with PatchGAN texture loss. In summary, this paper introduces a novel method for denoising OCT images by incorporating texture loss within a GAN framework. The proposed approach outperforms existing methods and is well-received by clinical experts, offering promising advancements in OCT image reconstruction and facilitating accurate clinical interpretation.
Details
- Title
- Employing texture loss to denoise OCT images using generative adversarial networks
- Authors
- Maryam Mehdizadeh (Corresponding Author) - The University of Western AustraliaSajib Saha - Commonwealth Scientific and Industrial Research OrganisationDavid Alonso-Caneiro - University of the Sunshine Coast, Queensland, School of Science, Technology and EngineeringJason Kugelman - Queensland University of TechnologyCara MacNish - The University of Western AustraliaFred Chen - The University of Western Australia
- Publication details
- Biomedical Optics Express, Vol.15(4), pp.2262-2280
- Publisher
- Optica
- Date published
- 2024
- DOI
- 10.1364/BOE.503868
- ISSN
- 2156-7085
- PMID
- 38633090
- Copyright note
- © 2024 Optica Publishing Group under the terms of the Open Access Publishing Agreement. Users may use, reuse, and build upon the article, or use the article for text or data mining, so long as such uses are for noncommercial purposes and appropriate attribution is maintained. All other rights are reserved.
- Data Availability
- he code implementations used in this study were based on the SiameseGAN and pix2pix frameworks. The SiameseGAN code was sourced from the public repository SiameseGAN [43], and the keras implementation of pix2pix code was obtained from pix2pix [44]. These frameworks provided the essential foundations for the development and training of the deep learning models employed in our experiments. The datasets employed in this study were acquired from the Queensland University of Technology (QUT) under the terms of an ethics agreement that ensures compliance with all relevant ethical guidelines and regulations. Unfortunately, due to privacy and ethical considerations, the datasets cannot be made publicly available. Researchers interested in accessing the datasets can initiate the necessary ethical approval process through QUT.
- Organisation Unit
- School of Science, Technology and Engineering
- Language
- English
- Record Identifier
- 991020798702621
- Output Type
- Journal article
Metrics
39 Record Views