Artifact Reduction in Fundus Imaging using Cycle Consistent Adversarial Neural Networks

2021/12/25 by S, Sai Koushik S, Srinivasa, K. G.
#Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #FOS: Electrical engineering #Image and Video Processing (eess.IV) #Machine Learning (cs.LG) #electronic engineering #information engineering

paper · doi:10.48550/arxiv.2112.13264

Abstract

Fundus images are very useful in identifying various ophthalmic disorders. However, due to the presence of artifacts, the visibility of the retina is severely affected. This may result in misdiagnosis of the disorder which may lead to more complicated problems. Since deep learning is a powerful tool to extract patterns from data without much human intervention, they can be applied to image-to-image translation problems. An attempt has been made in this paper to automatically rectify such artifacts present in the images of the fundus. We use a CycleGAN based model which consists of residual blocks to reduce the artifacts in the images. Significant improvements are seen when compared to the existing techniques.

Related

• Modeling and Enhancing Low-quality Retinal Fundus Images
• RFormer: Transformer-based Generative Adversarial Network for Real Fundus Image Restoration on A New Clinical Benchmark
• Optimal Transport Guided Unsupervised Learning for Enhancing low-quality Retinal Images
• Retinal Image Restoration and Vessel Segmentation using Modified Cycle-CBAM and CBAM-UNet
• Analyzing an Imitation Learning Network for Fundus Image Registration\n Using a Divide-and-Conquer Approach