Cardiac ultrasound image segmentation via multi-expert consensus diffusion model_Chinese Journal of Clinical Thoracic and Cardiovascular Surgery

Authors：

ZHANG Han ,  LI Kang

West China Biomedical Big Data Center, West China Hospital, Sichuan University, Chengdu, 610041, P. R. China;

Corresponding?author：

LI Kang, Email: likang@wchscu.cn

Keywords：

Echocardiographic image segmentation; diffusion models; annotation diversity; annotation consensus; left ventricle segmentation; uncertainty-aware segmentation; inter-observer variability; probabilistic consensus strategy

DOI：

10.7507/1007-4848.202601043

Video：

Export PDF Favorites Scan Get Citation

Abstract Full text Figures/Tables Video References Cited by

Objective To address the inter-observer annotation variability in echocardiographic image segmentation caused by image boundary ambiguity and differences in experts’ clinical backgrounds. Methods We proposed an uncertainty-aware consensus segmentation framework based on a conditional diffusion model. Conditioned on the input image, the framework employed a probabilistic consensus strategy to dynamically integrate annotations from multiple experts, jointly modeling stable anatomical consensus and clinically plausible annotation diversity. To compensate for the lack of real multi-expert annotations in publicly available datasets (e.g., CAMUS), we constructed a synthetic multi-expert annotation system using morphological operations to emulate three representative clinical labeling styles including conservative, moderate, and aggressive, providing a reliable foundation for method validation. Results The proposed model significantly outperformed state-of-the-art methods in both left ventricular endocardium (LVEndo) and left ventricular epicardium (LVEpi) segmentation tasks. For LVEndo, the Generalized Energy Distance () at the end-diastolic (ED) phase reached 0.073 1,representing a 43.9% reduction compared to the D-Persona model. For LVEpi, the decreased by 42.0% and 39.0% at the ED and end-systolic phases, respectively, relative to D-Persona. Furthermore, the structural fidelity improved by 2.7-3.5 percentage points for LVEndo and 1.4-2.2 percentage points for LVEpi compared to single-expert models, indicating a superior ability to capture diverse expert preferences. Conclusion By jointly modeling population-level consensus and annotation diversity, this work enables a unified characterization of anatomical structures and their associated annotation uncertainty in cardiac ultrasound images, offering a novel approach toward robust and interpretable segmentation in clinical settings.

1.	Ronneberger O, Fischer P, Brox T. U-net: convolutional networks for biomedical image segmentation. Navab N, Hornegger J, Wells W, et al, editors. Medical Image Computing and Computer-assisted Intervention-MICCAI 2015. Cham: Springer, 2015: 234-241.
2.	Isensee F, Jaeger PF, Kohl SAA, et al. nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nat Methods, 2021, 18(2): 203-211.
3.	Cao H, Wang Y, Chen J, et al. Swin-unet: unet-like pure transformer for medical image segmentation. Karlinsky L, Michaeli T, Nishino K, et al, editors. European Conference on Computer Vision. Cham: Springer Nature Switzerland, 2022: 205-218.
4.	Ma J, He Y, Li F, et al. Segment anything in medical images. Nat Commun, 2024, 15(1): 654.
5.	Lei Y, Fu Y, Roper J, et al. Echocardiographic image multi-structure segmentation using Cardiac-SegNet. Med Phys, 2021, 48(5): 2426-2437.
6.	Noble JA, Boukerroui D. Ultrasound image segmentation: a survey. IEEE Trans Med Imaging, 2006, 25(8): 987-1010.
7.	Nagata Y, Kado Y, Onoue T, et al. Impact of image quality on reliability of the measurements of left ventricular systolic function and global longitudinal strain in 2D echocardiography. Echo Res Pract, 2018, 5(1): 27-39.
8.	汪天富, 鄭昌瓊, 李德玉, 等. 用神經網絡進行超聲醫學圖像分割. 生物醫學工程學雜志, 1998, 15(4): 397-399.Wang TF, Zheng CQ, Li DY, et al. Segmentation of ultrasound medical images using neural networks. J Biomed Eng, 1998, 15(4): 397-399.
9.	Pombo JF, Troy BL, Russell RO. Left ventricular volumes and ejection fraction by echocardiography. Circulation, 1971, 43(4): 480-490.
10.	Lang RM, Badano LP, Mor-Avi V, et al. Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. Eur Heart J Cardiovasc Imaging, 2015, 16(3): 233-270.
11.	Begoli E, Bhattacharya T, Kusnezov D. The need for uncertainty quantification in machine-assisted medical decision making. Nat Mach Intell, 2019, 1(1): 20-23.
12.	Yu L, Wang S, Li X, et al. Uncertainty-aware self-ensembling model for semi-supervised 3D left atrium segmentation. Shen DG, Liu TM, Peters TM, et al, editors. International Conference on Medical Image Computing and Computer-assisted Intervention. Cham: Springer International Publishing, 2019: 605-613.
13.	Lu Z, Yang L, Li J, et al. Curating a knowledge base for patients with neurosyphilis: a study protocol of a DEep learning Framework for pErsonalized prediction of Adverse prognosTic events in NeuroSyphilis (DEFEAT-NS). BMJ Open, 2025, 15(7): e092248.
14.	Xu M C, Zhou Y K, Jin C, et al. Learning morphological feature perturbations for calibrated semi-supervised segmentation. Ender K, Bjoern M, Archana V, et al, editors. International Conference on Medical Imaging with Deep Learning. PMLR, 2022: 1413-1429.
15.	Ho J, Jain A, Abbeel P. Denoising diffusion probabilistic models. Adv Neural Inf Process Syst, 2020, 33: 6840-6851.
16.	Zbinden L, Doorenbos L, Pissas T, et al. Stochastic segmentation with conditional categorical diffusion models. Proc IEEE/CVF Int Conf Comput Vis, 2023: 1119-1129.
17.	Rahman A, Valanarasu JMJ, Hacihaliloglu I, et al. Ambiguous medical image segmentation using diffusion models. Proc IEEE/CVF Conf Comput Vis Pattern Recognit, 2023: 11536-11546.
18.	Zhang H, Luo X, Chen Y, et al. DiffOSeg: omni medical image segmentation via multi-expert collaboration diffusion model. Gee JC, AlexanderDC, Hong J, et al, editors. International Conference on Medical Image Computing and Computer-assisted Intervention. Cham: Springer Nature Switzerland, 2025: 128-138.
19.	Kohl S, Romera-Paredes B, Meyer C, et al. A probabilistic u-net for segmentation of ambiguous images. Adv Neural Inf Process Syst, 2018, 31.
20.	Leclerc S, Smistad E, Pedrosa J, et al. Deep learning for segmentation using an open large-scale dataset in 2D echocardiography. IEEE Trans Med Imaging, 2019, 38(9): 2198-2210.
21.	Gilbert A, Marciniak M, Rodero C, et al. Generating synthetic labeled data from existing anatomical models: an example with echocardiography segmentation. IEEE Trans Med Imaging, 2021, 40(10): 2783-2794.
22.	Goceri E. Medical image data augmentation: techniques, comparisons and interpretations. Artif Intell Rev, 2023: 1-45.
23.	Hoogeboom E, Nielsen D, Jaini P, et al. Argmax flows and multinomial diffusion: learning categorical distributions. Adv Neural Inf Process Syst, 2021, 34: 12454-12465.
24.	Kullback S, Leibler RA. On information and sufficiency. Ann Math Stat, 1951, 22(1): 79-86.
25.	Zhang L, Tanno R, Xu MC, et al. Disentangling human error from ground truth in segmentation of medical images. Adv Neural Inf Process Syst, 2020, 33: 15750-15762.
26.	Székely GJ, Rizzo ML. Energy statistics: a class of statistics based on distances. J Stat Plan Inference, 2013, 143(8): 1249-1272.
27.	Jaccard P. étude comparative de la distribution florale dans une portion des Alpes et des Jura. Bull Soc Vaudoise Sci Nat, 1901, 37: 547-579.
28.	Everingham M, Van Gool L, Williams CKI, et al. The pascal visual object classes (VOC) challenge. Int J Comput Vis, 2010, 88(2): 303-338.
29.	Wu Y, Luo X, Xu Z, et al. Diversified and personalized multi-rater medical image segmentation. Proc IEEE/CVF Conf Comput Vis Pattern Recognit, 2024: 11470-11479.
30.	Baumgartner CF, Tezcan KC, Chaitanya K, et al. Phiseg: capturing uncertainty in medical image segmentation. Shen D, Liu TM, Peters TM, et al, editors. Medical Image Computing and Computer-assisted Intervention-MICCAI 2019. Cham: Springer International Publishing, 2019: 119-127.

1. Ronneberger O, Fischer P, Brox T. U-net: convolutional networks for biomedical image segmentation. Navab N, Hornegger J, Wells W, et al, editors. Medical Image Computing and Computer-assisted Intervention-MICCAI 2015. Cham: Springer, 2015: 234-241.
2. Isensee F, Jaeger PF, Kohl SAA, et al. nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nat Methods, 2021, 18(2): 203-211.
3. Cao H, Wang Y, Chen J, et al. Swin-unet: unet-like pure transformer for medical image segmentation. Karlinsky L, Michaeli T, Nishino K, et al, editors. European Conference on Computer Vision. Cham: Springer Nature Switzerland, 2022: 205-218.
4. Ma J, He Y, Li F, et al. Segment anything in medical images. Nat Commun, 2024, 15(1): 654.
5. Lei Y, Fu Y, Roper J, et al. Echocardiographic image multi-structure segmentation using Cardiac-SegNet. Med Phys, 2021, 48(5): 2426-2437.
6. Noble JA, Boukerroui D. Ultrasound image segmentation: a survey. IEEE Trans Med Imaging, 2006, 25(8): 987-1010.
7. Nagata Y, Kado Y, Onoue T, et al. Impact of image quality on reliability of the measurements of left ventricular systolic function and global longitudinal strain in 2D echocardiography. Echo Res Pract, 2018, 5(1): 27-39.
8. 汪天富, 鄭昌瓊, 李德玉, 等. 用神經網絡進行超聲醫學圖像分割. 生物醫學工程學雜志, 1998, 15(4): 397-399.Wang TF, Zheng CQ, Li DY, et al. Segmentation of ultrasound medical images using neural networks. J Biomed Eng, 1998, 15(4): 397-399.
9. Pombo JF, Troy BL, Russell RO. Left ventricular volumes and ejection fraction by echocardiography. Circulation, 1971, 43(4): 480-490.
10. Lang RM, Badano LP, Mor-Avi V, et al. Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. Eur Heart J Cardiovasc Imaging, 2015, 16(3): 233-270.
11. Begoli E, Bhattacharya T, Kusnezov D. The need for uncertainty quantification in machine-assisted medical decision making. Nat Mach Intell, 2019, 1(1): 20-23.
12. Yu L, Wang S, Li X, et al. Uncertainty-aware self-ensembling model for semi-supervised 3D left atrium segmentation. Shen DG, Liu TM, Peters TM, et al, editors. International Conference on Medical Image Computing and Computer-assisted Intervention. Cham: Springer International Publishing, 2019: 605-613.
13. Lu Z, Yang L, Li J, et al. Curating a knowledge base for patients with neurosyphilis: a study protocol of a DEep learning Framework for pErsonalized prediction of Adverse prognosTic events in NeuroSyphilis (DEFEAT-NS). BMJ Open, 2025, 15(7): e092248.
14. Xu M C, Zhou Y K, Jin C, et al. Learning morphological feature perturbations for calibrated semi-supervised segmentation. Ender K, Bjoern M, Archana V, et al, editors. International Conference on Medical Imaging with Deep Learning. PMLR, 2022: 1413-1429.
15. Ho J, Jain A, Abbeel P. Denoising diffusion probabilistic models. Adv Neural Inf Process Syst, 2020, 33: 6840-6851.
16. Zbinden L, Doorenbos L, Pissas T, et al. Stochastic segmentation with conditional categorical diffusion models. Proc IEEE/CVF Int Conf Comput Vis, 2023: 1119-1129.
17. Rahman A, Valanarasu JMJ, Hacihaliloglu I, et al. Ambiguous medical image segmentation using diffusion models. Proc IEEE/CVF Conf Comput Vis Pattern Recognit, 2023: 11536-11546.
18. Zhang H, Luo X, Chen Y, et al. DiffOSeg: omni medical image segmentation via multi-expert collaboration diffusion model. Gee JC, AlexanderDC, Hong J, et al, editors. International Conference on Medical Image Computing and Computer-assisted Intervention. Cham: Springer Nature Switzerland, 2025: 128-138.
19. Kohl S, Romera-Paredes B, Meyer C, et al. A probabilistic u-net for segmentation of ambiguous images. Adv Neural Inf Process Syst, 2018, 31.
20. Leclerc S, Smistad E, Pedrosa J, et al. Deep learning for segmentation using an open large-scale dataset in 2D echocardiography. IEEE Trans Med Imaging, 2019, 38(9): 2198-2210.
21. Gilbert A, Marciniak M, Rodero C, et al. Generating synthetic labeled data from existing anatomical models: an example with echocardiography segmentation. IEEE Trans Med Imaging, 2021, 40(10): 2783-2794.
22. Goceri E. Medical image data augmentation: techniques, comparisons and interpretations. Artif Intell Rev, 2023: 1-45.
23. Hoogeboom E, Nielsen D, Jaini P, et al. Argmax flows and multinomial diffusion: learning categorical distributions. Adv Neural Inf Process Syst, 2021, 34: 12454-12465.
24. Kullback S, Leibler RA. On information and sufficiency. Ann Math Stat, 1951, 22(1): 79-86.
25. Zhang L, Tanno R, Xu MC, et al. Disentangling human error from ground truth in segmentation of medical images. Adv Neural Inf Process Syst, 2020, 33: 15750-15762.
26. Székely GJ, Rizzo ML. Energy statistics: a class of statistics based on distances. J Stat Plan Inference, 2013, 143(8): 1249-1272.
27. Jaccard P. étude comparative de la distribution florale dans une portion des Alpes et des Jura. Bull Soc Vaudoise Sci Nat, 1901, 37: 547-579.
28. Everingham M, Van Gool L, Williams CKI, et al. The pascal visual object classes (VOC) challenge. Int J Comput Vis, 2010, 88(2): 303-338.
29. Wu Y, Luo X, Xu Z, et al. Diversified and personalized multi-rater medical image segmentation. Proc IEEE/CVF Conf Comput Vis Pattern Recognit, 2024: 11470-11479.
30. Baumgartner CF, Tezcan KC, Chaitanya K, et al. Phiseg: capturing uncertainty in medical image segmentation. Shen D, Liu TM, Peters TM, et al, editors. Medical Image Computing and Computer-assisted Intervention-MICCAI 2019. Cham: Springer International Publishing, 2019: 119-127.

Chinese Journal of Clinical Thoracic and Cardiovascular Surgery

Latest ArticlesCardiac ultrasound image segmentation via multi-expert consensus diffusion model

Abstract Full text Figures/Tables Video References Cited by

Format

Content