Learning to Address Intra-segment Misclassification in Retinal Imaging

TL;DR

This paper introduces a novel deep learning approach for multi-class retinal vessel segmentation that reduces intra-segment misclassification by decomposing the task into binary classifications and using a fusion network with adversarial training.

Contribution

It proposes a new multi-stage network architecture that improves differentiation between arteries and veins in retinal images, addressing intra-segment misclassification issues.

Findings

Achieved up to 5.1% higher F1-score compared to state-of-the-art methods.

Effectively reduces intra-segment misclassification in retinal vessel segmentation.

Demonstrated improved performance across multiple retinal image datasets.

Abstract

Accurate multi-class segmentation is a long-standing challenge in medical imaging, especially in scenarios where classes share strong similarity. Segmenting retinal blood vessels in retinal photographs is one such scenario, in which arteries and veins need to be identified and differentiated from each other and from the background. Intra-segment misclassification, i.e. veins classified as arteries or vice versa, frequently occurs when arteries and veins intersect, whereas in binary retinal vessel segmentation, error rates are much lower. We thus propose a new approach that decomposes multi-class segmentation into multiple binary, followed by a binary-to-multi-class fusion network. The network merges representations of artery, vein, and multi-class feature maps, each of which are supervised by expert vessel annotation in adversarial training. A skip-connection based merging process explicitly maintains class-specific gradients to avoid gradient vanishing in deep layers, to favor the discriminative features. The results show that, our model respectively improves F1-score by 4.4\%, 5.1\%, and 4.2\% compared with three state-of-the-art deep learning based methods on DRIVE-AV, LES-AV, and HRF-AV data sets. Code: https://github.com/rmaphoh/Learning-AVSegmentation