Multi-Resolution Feature Stem for Diabetic Retinopathy Lesion Segmentation

Publication Date

1-1-2026

Document Type

Conference Proceeding

Publication Title

2026 International Conference on Advances in Artificial Intelligence and Machine Learning Aaiml 2026

DOI

10.1109/AAIML67890.2026.11498215

First Page

497

Last Page

502

Abstract

Diabetic Retinopathy (DR) is a leading cause of preventable blindness worldwide, requiring automated lesion segmentation using deep learning models for early detection and monitoring. However, DR lesions vary dramatically in size from tiny microaneurysms to large hemorrhages and exudates. This variability creates conflicting demands on the model architecture and input resolution, posing a challenge for effective design. This work investigates the impact of input resolution on different lesion types. Through systematic experimentation with multiple architectures (U-Net, UNet++, Vision Transformers, DeepLabV3+) at 512 × 512 and 1024 × 1024 resolutions, we identify a critical, counter-intuitive phenomenon where increasing input resolution has opposing effects on different lesion types. We demonstrate that while higher resolution is essential for resolving fine-grained microaneurysms, it can unexpectedly degrade performance on larger hemorrhages. This finding challenges the common assumption that higher resolution is uniformly beneficial. To address this, we propose a novel Multi-Resolution Feature Stem, an input-level pyramid integrated with a UNet++ backbone. This architecture processes multiple scales in parallel, capturing fine-grained details without sacrificing contextual information. This work contributes crucial empirical evidence of this complex, resolution-dependent behavior and a practical, parameterefficient architecture that successfully resolves this trade-off. Our code is available at https://github.com/taeheej/Multi-Resolution-Feature-Stem-for-Diabetic-Retinopathy-lesion-segmentation.

Keywords

Diabetic retinopathy, Lesion segmentation, Multi-Resolution, Pyramid integration

Department

Applied Data Science

Link to Full Text

Share

COinS