Multi-scale Sparse Representation-Based Shadow Inpainting for Retinal OCT Images


Shadows artifacts in retinal optical coherence tomography (OCT) images might cause errors for accurate machine analysis and clinical diagnosis. However, traditional methods often fail to deal with wide shadows, while deep learning-based methods require large datasets and laborious labelling. Here, we propose a multi-scale sparse representation-based shadow inpainting framework. Specifically, shadows are inpainted by sparse representation at different resolution scales and are later concatenated with the aid of a convolutional neural network-based super resolution (SR) module. Experiments are conducted to compare our proposal versus both traditional and deep learning-based techniques on synthetic and real-world shadows. Results demonstrate that our proposed method achieves favorable image inpainting in terms of visual quality and quantitative metrics, especially when wide shadows are presented.



Download our collected dataset at this link:


Visual results of synthetic shadow and real shadow on the collected dataset:

Retinal OCT image segmentation results before and after inpainting:

For more results, please refer to our paper (


If you have any questions, please contact