Evaluating semi-supervision methods for medical image segmentation: applications in cardiac magnetic resonance imaging.

Autor:	Hooper SM; Stanford University, Department of Electrical Engineering, Stanford, California, United States., Wu S; Stanford University, Department of Computer Science, Stanford, California, United States., Davies RH; Barts Health NHS Trust, Barts Heart Centre, London, United Kingdom.; University of College London, Institute of Cardiovascular Sciences, London, United Kingdom.; University of College London, MRC Centre for Lifelong Health and Ageing, London, United Kingdom., Bhuva A; Barts Health NHS Trust, Barts Heart Centre, London, United Kingdom.; University of College London, Institute of Cardiovascular Sciences, London, United Kingdom., Schelbert EB; United Hospital, St. Paul, Minnesota, and Abbott Northwestern Hospital, Minneapolis Heart Institute, Minneapolis, Minnesota, United States.; UPMC Cardiovascular Magnetic Resonance Center, UPMC, Pittsburgh, Pennsylvania, United States.; University of Pittsburgh School of Medicine, Department of Medicine, Pittsburgh, Pennsylvania, United States., Moon JC; Barts Health NHS Trust, Barts Heart Centre, London, United Kingdom.; University of College London, Institute of Cardiovascular Sciences, London, United Kingdom., Kellman P; National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, Maryland, United States., Xue H; National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, Maryland, United States., Langlotz C; Stanford University, Department of Radiology, Department of Biomedical Informatics, Stanford, California, United States., Ré C; Stanford University, Department of Computer Science, Stanford, California, United States.
Jazyk:	angličtina
Zdroj:	Journal of medical imaging (Bellingham, Wash.) [J Med Imaging (Bellingham)] 2023 Mar; Vol. 10 (2), pp. 024007. Date of Electronic Publication: 2023 Mar 30.
DOI:	10.1117/1.JMI.10.2.024007
Abstrakt:	Purpose: Neural networks have potential to automate medical image segmentation but require expensive labeling efforts. While methods have been proposed to reduce the labeling burden, most have not been thoroughly evaluated on large, clinical datasets or clinical tasks. We propose a method to train segmentation networks with limited labeled data and focus on thorough network evaluation. Approach: We propose a semi-supervised method that leverages data augmentation, consistency regularization, and pseudolabeling and train four cardiac magnetic resonance (MR) segmentation networks. We evaluate the models on multiinstitutional, multiscanner, multidisease cardiac MR datasets using five cardiac functional biomarkers, which are compared to an expert's measurements using Lin's concordance correlation coefficient (CCC), the within-subject coefficient of variation (CV), and the Dice coefficient. Results: The semi-supervised networks achieve strong agreement using Lin's CCC ( > 0.8 ), CV similar to an expert, and strong generalization performance. We compare the error modes of the semi-supervised networks against fully supervised networks. We evaluate semi-supervised model performance as a function of labeled training data and with different types of model supervision, showing that a model trained with 100 labeled image slices can achieve a Dice coefficient within 1.10% of a network trained with 16,000+ labeled image slices. Conclusion: We evaluate semi-supervision for medical image segmentation using heterogeneous datasets and clinical metrics. As methods for training models with little labeled data become more common, knowledge about how they perform on clinical tasks, how they fail, and how they perform with different amounts of labeled data is useful to model developers and users. (© 2023 Society of Photo-Optical Instrumentation Engineers (SPIE).)
Databáze:	MEDLINE
Externí odkaz:	Zobrazit plný text záznamu