SVVD-NET: A framework with relative position constraints for vertebral vertex detection

Yongkang Xu, Lianhong Duan, Zhicheng Zhang, Tiansheng Sun, Yang Zhang, Lixia Tian

Abstract: Highlights•Overlapping and deviated vertebrae are common with existing methods.•SVVD-Net is proposed to tackle the issues via a landmark sequence generation framework.•It employs anatomy-aware encoder to handle vertebrae similarities among individuals.•It adopts structure mask to address occasional subtle misalignments between two adjacent vertebrae.