Decom-UNet3+: A Retinal Vessel Segmentation Method Optimized with Decomposed Convolutions

The intricate and highly branched structure of retinal blood vessels, along with the fragility of fine vessels, makes segmentation a challenging task. To address this issue, we proposed Decom-UNet3+, a model that optimizes the encoders by using decomposed convolutions. Specifically, the encoders replace standard convolutional layers with asymmetric convolutions and depth-wise separable convolutions, reducing the number of parameters while enhancing feature extraction capability. Additionally, a spatial attention mechanism is integrated to enhance focus on vessel regions and suppress background noise interference. The model is evaluated on high-resolution, expertly annotated datasets CHASEDB1, DRIVE, STARE and HRF, achieving an average accuracy of 97.2% on CHASEDB1, 96.4% on DRIVE, 94.3% on STARE and 97.6% on HRF, outperforming the original UNet3+ model. The result demonstrates that Decom-UNet3+ effectively improves vascular segmentation performance with lower computational cost and parameter overhead, offering a more efficient and reliable solution for automated retinal disease screening.