Журнал «Труды Института системного анализа Российской академии наук» - N. Ismail, N.F. Gusarova "Vesselness-Based Semi-Automatic Carotid Segmentation in CTA"

Cervical carotid artery segmentation in contrast-enhanced CTA supports stenosis assessment and vessel geometry analysis, but routine scans are challenging due to noise, variable enhancement, calcified plaque, and complex bifurcation/skull-base anatomy. This paper presents a semi-automatic approach that combines multiscale Hessian vesselness with slice-wise mask propagation and local intensity-based refinement. Segmentation is initialized from a small set of manual lumen contours, propagated axially using vesselness thresholding and connected-component selection with simple area/centroid constraints, and refined by HU-based region growing within the vesselness neighborhood. Tested on 10 CTA scans, the method tracks the CCA and ICA over several centimeters in most cases, with generally stable behavior at the bifurcation and only occasional switching to adjacent vessels. Across the 8 cases without ICA/ECA switching (Table 1), the segmentation achieved Dice = 0.925 ± 0.015 and HD95 = 1.13 ± 0.21 mm.

Полная версия статьи в формате pdf.

EDN: EEDPPA

PP. 77-86.

References

1. Isensee F., Jaeger P.F., Kohl S.A.A., Petersen J., Maier-Hein K.H. nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nat Methods. 2021;18:203-211. https://doi.org/10.1038/s41592-020-01008-z

2. Frangi A.F., Niessen W.J., Vincken K.L., Viergever M.A. Multiscale vessel enhancement filtering. In: Medical Image Computing and Computer-Assisted Intervention (MICCAI). 1998. p. 130-137. https://doi.org/10.1007/BFb0056195

3. Maier A. Multiscale vessel enhancement filtering: what made the Frangi filter so popular? Medium. 2024. Available from: https://akmaier.medium.com/multiscale-vessel-enhancement-filtering-whatmade-the-frangi-filter-so-popluar-02240cf54046 [Accessed 2025].

4. Sato Y., Nakajima S., Atsumi H., et al. Threedimensional multi-scale line filter for segmentation and visualization of curvilinear structures in medical images. Med Image Anal. 1998;2(2):143-168. https://doi.org/10.1016/S1361-8415(98)80009-1

5. Xie H., Gu H., Li M., Zhu L., Wang T., Li Z., Wu H.. Carotid artery segmentation in computed tomography angiography (CTA) using multi-scale deep supervision with Swin-UNet and advanced data augmentation. Quant Imaging Med Surg. 2025;15(4):3161-3175. https://doi.org/10.21037/qims-24-2087

6. Luo X., Hu B., Zhou S., Wu Q., Geng C., Zhao L., Li Y., Di R., Pu J., Geng D., Yang L. CAP-Net: carotid artery plaque segmentation system based on computed tomography angiography. Acad Radiol. 2025;32(10):6194-6204. https://doi.org/10.1016/j.acra.2025.07.009

7. Zhou T., Wang H., Li J., et al. Fully automatic deep learning trained on limited data for 3D carotid bifurcation segmentation in CT angiography (CarotidNet). Quant Imaging Med Surg. 2021;11(1):249-262. Available from: https://qims.amegroups.org/article/view/50255 [Accessed 2025].

8. Florack L.M.J., et al. Scale and the differential structure of images. Image Vis Comput. 1992;10(6):376-388. https://doi.org/10.1016/0262-8856(92)90024-W

9. Chapman B.E., et al. 3D multi-scale vessel enhancement filtering based on Hessian eigenvalues. Med Image Anal. 2005;9(5):491-502. https://doi.org/10.1016/j.media.2005.03.004

10. Zhai D., Liu R., Liu Y., Yin H., Tang W., Yang J., Liu K., Fan G., Ju S., Cai W. Deep learningbased fully automatic screening of carotid artery plaques in computed tomography angiography: a multicenter study. Clin Radiol. 2024;79(8):e994-e1002. https://doi.org/10.1016/j.crad.2024.04.015

11. Yoo S.W., et al. CACSNet for automatic robust classification and segmentation of carotid atherosclerotic plaques. Sci Rep. 2024;14. https://doi.org/10.1038/s41598-024-64265-4

12. Guo Z., Liu Y., Xu J., Huang C., Zhang F., Miao C., Zhang Y., Li M., Shan H., Gu Y. A deep learning model for carotid plaques detection based on CTA images: a two stepwise early-stage clinical validation study. Front Neurol. 2024;15:1480792. https://doi.org/10.3389/fneur.2024.1480792