Title: Anatomical-prior-based polar regression method for intravascular ultrasound contour delineation

Authors: Xu Yu; Xun Liu; Yutao Lin; Shengxian Tu

Addresses: School of Biomedical Engineering, Southern Medical University, No.1023, South Shatai Road, Baiyun District, Guangzhou, 510515, Guangdong, China ' School of Biomedical Engineering, Shanghai Jiao Tong University, 1954 Huashan Rd., Xuhui District, Shanghai, 200030, Shanghai, China ' School of Biomedical Engineering, Southern Medical University, No.1023, South Shatai Road, Baiyun District, Guangzhou, 510515, Guangdong, China ' School of Biomedical Engineering, Southern Medical University, No.1023, South Shatai Road, Baiyun District, Guangzhou, 510515, Guangdong, China; School of Biomedical Engineering, Shanghai Jiao Tong University, 1954 Huashan Rd., Xuhui District, Shanghai, 200030, Shanghai, China

Abstract: Accurate delineation of the lumen and external elastic lamina in intravascular ultrasound images is critical for quantitative vascular analysis but remains challenging due to noise, artefacts, contour ambiguity, and limited temporal consistency. This study proposes a contour delineation framework that is better aligned with the radial imaging mechanism of IVUS and the closed-topology geometry of vascular contours. Radius-aware circular positional encoding is used to capture the intrinsic circular vessel topology, and a dual-branch architecture with radial and global pathways enables attention-based multi-scale feature fusion. A temporal-spatial smoothing constraint is further incorporated during training to enhance inter-frame consistency. Experiments on a multi-centre IVUS dataset comprising 112 pullback examinations from 58 patients (43,867 frames) demonstrate that the proposed method outperforms state-of-the-art approaches across multiple evaluation metrics, achieving accurate and sequence-consistent contour delineation.

Keywords: intravascular ultrasound; deep learning; contour regression; atherosclerosis.

DOI: 10.1504/IJBET.2026.156072

International Journal of Biomedical Engineering and Technology, 2026 Vol.51 No.2, pp.147 - 174

Received: 09 Jan 2026
Accepted: 09 Apr 2026
Published online: 02 Sep 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article