The Topology-Overlap Trade-Off in Retinal Arteriole-Venule Segmentation
This work addresses the challenge of accurate and topologically correct vessel segmentation in retinal images, which is crucial for efficient diagnosis of diseases like hypertension or diabetes, though it represents an incremental improvement over existing methods.
The paper tackles the problem of automatic segmentation of retinal arterioles and venules, where convolutional neural networks often produce topologically incorrect predictions despite high overlap with expert annotations. By incorporating a topology-preserving loss term and an orientation score guided convolutional module, the model achieves results on par with state-of-the-art in overlap metrics while improving topological accuracy.
Retinal fundus images can be an invaluable diagnosis tool for screening epidemic diseases like hypertension or diabetes. And they become especially useful when the arterioles and venules they depict are clearly identified and annotated. However, manual annotation of these vessels is extremely time demanding and taxing, which calls for automatic segmentation. Although convolutional neural networks can achieve high overlap between predictions and expert annotations, they often fail to produce topologically correct predictions of tubular structures. This situation is exacerbated by the bifurcation versus crossing ambiguity which causes classification mistakes. This paper shows that including a topology preserving term in the loss function improves the continuity of the segmented vessels, although at the expense of artery-vein misclassification and overall lower overlap metrics. However, we show that by including an orientation score guided convolutional module, based on the anisotropic single sided cake wavelet, we reduce such misclassification and further increase the topology correctness of the results. We evaluate our model on public datasets with conveniently chosen metrics to assess both overlap and topology correctness, showing that our model is able to produce results on par with state-of-the-art from the point of view of overlap, while increasing topological accuracy.