The joint learning of multi-resolution feature for multi-class retinal vessel segmentation

The task of multi-class vessel segmentation on retinal images is the basis for the arteriovenous quantitative analysis, which plays an important role in the diagnosis and treatment of cerebrovascular diseases. Due to the intricate details and intertwining of the retinal vessels, traditional feature learning networks based on single-level resolution images are prone to the troubles from arteriovenous confusion and vascular edge discontinuity. To this end, we develop a paradigm of multi-level image resolution joint learning, which makes up for the defects of feature modeling arising from single-level image resolution by combining the feature learning ability of the network across multiple resolution images. Specifically, we design a cross-scale feature fusion network, which uses structure of a global-local dual-branch network to extract the features of retinal images at multiple resolutions, so as to supplement the missing vascular features arising from the single-resolution branch network. This framwork not only corrects the intra-segment misclassification, but also improves continuity by supplementing the details of vascular edge. Furthermore, the cross-scale fusion process of the network at multiple stages is conducive to its optimization and enhances the collaborative learning ability of dual-branch. Meanwhile, we use the generative adversarial structure as the backbone to supervise and constrain the aforementioned feature fusion results. Finally, a large number of experiments are conducted on three publicly available datasets, DRIVE-AV, LES-AV, and HRF-AV. It is shown that the proposed scheme outperforms the current state-of-the-art methods significantly. The source code is available at https://github.com/Tang9867/Multi-Resolution-Learning.