“Error factory”: a framework for simulating errors in left ventricle segmentation in cardiac images

• Bruno F. Raquel USP
• Matheus A. O. Ribeiro USP
• Fátima L. S. Nunes USP

DOI: https://doi.org/10.5753/sbcas_estendido.2025.7846

Abstract

Left ventricle segmentation in Cardiac Magnetic Resonance exams is important for medical diagnosis. Deep learning methods have excelled in obtaining segmentations similar to those of experts. However, one of the current limitations is the arbitrary production of anatomical errors that can compromise the diagnosis. Given this problem, this work proposes a framework for categorization and controlled automatic simulation of different anatomical errors. The framework favors the development of methods aimed at detecting and correcting these errors. Results indicate that the proposed framework is capable of generating errors of several categories and replicating the same errors produced by networks (Dice > 0.8).

References

Bernard, O. et al. (2018). Deep learning techniques for automatic MRI cardiac multistructures segmentation and diagnosis: Is the problem solved? IEEE Transactions on Medical Imaging, 37(11):2514–2525.

Guan, S., Samala, R. K., Arab, A., and Chen, W. (2023). Miss-tool: medical image segmentation synthesis tool to emulate segmentation errors. In Iftekharuddin, K. M. and Chen, W., editors, Medical Imaging 2023: Computer-Aided Diagnosis, page 41. SPIE.

Landman, B. and Warfield, S. (2013). 2013 cardiac atlas project standard challenge - participant project.

Lohr, D. et al. (2024). Precision imaging of cardiac function and scar size in acute and chronic porcine myocardial infarction using ultrahigh-field mri. Communications Medicine, 4(1).

Radau, P. et al. (2009). Evaluation framework for algorithms segmenting short axis cardiac MRI. The MIDAS Journal.

Ribeiro, M. A. O. and Nunes, F. L. S. (2021). Evaluating the pre-processing impact on the generalization of deep learning networks for left ventricle segmentation. In 2021 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), page 3505–3512. IEEE.

Ribeiro, M. A. O. and Nunes, F. L. S. (2022). Left ventricle segmentation in cardiac MR: A systematic mapping of the past decade. ACM Comput. Surv., 54(11s).

Tajbakhsh, N., Lai, B., Ananth, S. P., and Ding, X. (2020). Errornet: Learning error representations from limited data to improve vascular segmentation. In 2020 IEEE 17th International Symposium on Biomedical Imaging (ISBI), page 1364–1368. IEEE.

Waite, S. et al. (2017). Tired in the reading room: The influence of fatigue in radiology. Journal of the American College of Radiology, 14(2):191–197.

Wang, Y. et al. (2021). Deep learning based fully automatic segmentation of the left ventricular endocardium and epicardium from cardiac cine MRI. Quantitative Imaging in Medicine and Surgery, 11(4):1600–1612.

Yuan, T. et al. (2018). Fully automatic segmentation of the left ventricle using multi-scale fusion learning. In 2018 24th International Conference on Pattern Recognition (ICPR), page 3838–3843. IEEE.