Vessel segmentation is a fundamental task in medical image analysis and plays an important role in computer-aided diagnosis, lesion localization, vascular morphology analysis, and subsequent three-dimensional reconstruction. However, blood vessels usually exhibit elongated shapes, complex branching patterns, significant scale variations, and locally low contrast. Under challenging conditions such as complex backgrounds, noise interference, and blurred boundaries, tiny vessels are prone to missed detection, while background textures and spurious edges are easily misclassified as vessels, resulting in increased false positives. To address these issues, this paper proposes a prior-guided vessel segmentation method with false positive suppression. The proposed method adopts an encoder-decoder architecture as the backbone and introduces a Vessel False Positive Suppression Gate (VFPSGate) into the skip connections during the decoding stage. By integrating vessel region priors and edge priors, the shallow features are recalibrated in a suppression-oriented manner, thereby reducing the interference of background noise and non-vascular high responses on segmentation results. In addition, a Vessel False Positive Suppression Loss (VFPSLoss) is designed to impose extra constraints on abnormally high responses in background regions that are not supported by the priors, thus enhancing the model’s targeted suppression ability against false positives at the optimization level. Experimental results on the DCA dataset demonstrate that the proposed method achieves competitive performance, with IoU, DSC, ACC, and SEN reaching 66.49%, 79.72%, 97.88%, and 85.16%, respectively. Overall, the proposed method can more effectively distinguish real vessels from pseudo-vessels, providing a feasible solution for vessel segmentation under complex background conditions.