# Learning Mutually Local-global U-nets For High-resolution Retinal Lesion Segmentation in Fundus Images

arXiv: https://arxiv.org/abs/1901.06047 · HTML: https://mlsanity.com/paper/1901.06047/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 18 Jan 2019.

## Built on by (most cited first, 9 of 9)

- [Applications of Deep Learning in Fundus Images: A Review](https://mlsanity.com/paper/2101.09864/) (2021)
- [Global-Local Transformer for Brain Age Estimation](https://mlsanity.com/paper/2109.01663/) (2021)
- [Progressive Multi-scale Consistent Network for Multi-class Fundus Lesion Segmentation](https://mlsanity.com/paper/2205.15720/) (2022)
- [OUCopula: Bi-Channel Multi-Label Copula-Enhanced Adapter-Based CNN for Myopia Screening Based on OU-UWF Images](https://mlsanity.com/paper/2403.11974/) (2024)
- [Automated Detection and Diagnosis of Diabetic Retinopathy: A Comprehensive Survey](https://mlsanity.com/paper/2107.00115/) (2021)
- [Coarse Retinal Lesion Annotations Refinement via Prototypical Learning](https://mlsanity.com/paper/2208.14294/) (2022)
- [Cross-Dataset Generalization For Retinal Lesions Segmentation](https://mlsanity.com/paper/2405.08329/) (2024)
- [Multi-style conversion for semantic segmentation of lesions in fundus images by adversarial attacks](https://mlsanity.com/paper/2410.13822/) (2024)
- [L2GNet: Optimal Local-to-Global Representation of Anatomical Structures for Generalized Medical Image Segmentation](https://mlsanity.com/paper/2502.05229/) (2025)

## Abstract (arXiv)

Diabetic retinopathy is the most important complication of diabetes. Early diagnosis of retinal lesions helps to avoid visual loss or blindness. Due to high-resolution and small-size lesion regions, applying existing methods, such as U-Nets, to perform segmentation on fundus photography is very challenging. Although downsampling the input images could simplify the problem, it loses detailed information. Conducting patch-level analysis helps reaching fine-scale segmentation yet usually leads to misunderstanding as the lack of context information. In this paper, we propose an efficient network that combines them together, not only being aware of local details but also taking fully use of the context perceptions. This is implemented by integrating the decoder parts of a global-level U-net and a patch-level one. The two streams are jointly optimized, ensuring that they are enhanced mutually. Experimental results demonstrate our new framework significantly outperforms existing patch-based and global-based methods, especially when the lesion regions are scattered and small-scaled.