GVC-Net:Global Vascular Context Network for Cerebrovascular Segmentation Using Sparse Labels

Cerebrovascular disease is a serious threat to human health. Because of its high mortality and disability rate, early diagnosis and prevention are very important. The performance of existing cerebrovascular segmentation methods based on deep learning depends on the integrity of labels. However, manual labels are usually of low quality and poor connectivity at small blood vessels, which directly affects the cerebrovascular segmentation results. In this paper, we propose a new segmentation network to segment cerebral vessels from MRA images by using sparse labels. The long-distance dependence between vascular structures is captured by the global vascular context module, and the topology is constrained by the hybrid loss function to segment the cerebral vessels with good connectivity. Experiments show that our method performed with a sensitivity, precision, dice similarity coefficient, intersection over union and centerline dice similarity coefficient of 61.24%, 75.58%, 67.66%, 51.13% and 83.79% respectively. The obtained results reveal that the proposed cerebrovascular segmentation network has better segmentation performance for cerebrovascular segmentation under sparse labels, and can suppress the noise of background to a certain extent. We adapted hybrid loss to preserve the topology of segmentation results and proposed Global Vascular Context Module (GVCM) captures the long-distance dependence between plane features, to help the network understand the structure of blood vessels and improve the anti-interference ability of segmentation model to skull information noise and the connectivity of blood vessels. • Extracting the continuous vascular structure information from two planes. • Combining GVC-Net and hybrid loss against sparse labels and background. • Suppressing over segmentation of the blood vessels in the skull. • Improving the connectivity of segmented cerebral vessels.