Resumen
Traditional approaches to image reconstruction uses physics-based loss with data-efficient inference, although the difficulty to properly model the inverse solution precludes learning the reconstruction across a distribution of data. Modern deep learning approaches enable expressive modeling but rely on a large number of reconstructed images (labeled data) that are often not available in practice. To combine the best of the above two lines of works, we present a novel label-free image reconstruction network that is supervised by physics-based forward operators rather than labeled data. We further present an expressive yet disentangled spatial-temporal modeling of the inverse solution, where its latent dynamics is modeled by neural ordinary differential equations and its emission over non-Euclidean geometrical domains by graph convolutional neural networks. We applied the presented method to reconstruct electrical activity on the heart surface from body-surface potential. In simulation and real-data experiments in comparison to both traditional physics-based and modern data-driven reconstruction methods, we demonstrated the ability of the presented method to learn how to reconstruct using observational data without any corresponding labels.
Idioma original | English |
---|---|
Título de la publicación alojada | Medical Image Computing and Computer Assisted Intervention – MICCAI 2021 - 24th International Conference, Proceedings |
Editores | Marleen de Bruijne, Marleen de Bruijne, Philippe C. Cattin, Stéphane Cotin, Nicolas Padoy, Stefanie Speidel, Yefeng Zheng, Caroline Essert |
Editorial | Springer Science and Business Media Deutschland GmbH |
Páginas | 361-371 |
Número de páginas | 11 |
ISBN (versión impresa) | 9783030872304 |
DOI | |
Estado | Published - 2021 |
Evento | 24th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2021 - Virtual, Online Duración: sep. 27 2021 → oct. 1 2021 |
Serie de la publicación
Nombre | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
---|---|
Volumen | 12906 LNCS |
ISSN (versión impresa) | 0302-9743 |
ISSN (versión digital) | 1611-3349 |
Conference
Conference | 24th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2021 |
---|---|
Ciudad | Virtual, Online |
Período | 9/27/21 → 10/1/21 |
Nota bibliográfica
Funding Information:Acknowledgement. This work is supported by the National Institutes of Health
Funding Information:
This work is supported by the National Institutes of Health (NIH) under Award Number R01HL145590 and R01NR018301.
Publisher Copyright:
© 2021, Springer Nature Switzerland AG.
ASJC Scopus Subject Areas
- Theoretical Computer Science
- General Computer Science