Display options
Cite
Burwinkel H, Matz H, Saur S, et al. Physics-aware learning and domain-specific loss design in ophthalmology. Med Image Anal. 2021;76:102314doi: 10.1016/j.media.2021.102314.
Burwinkel, H., Matz, H., Saur, S., Hauger, C., Trost, M., Hirnschall, N., Findl, O., Navab, N., & Ahmadi, S. A. (2021). Physics-aware learning and domain-specific loss design in ophthalmology. Medical image analysis, 76102314. https://doi.org/10.1016/j.media.2021.102314
Burwinkel, Hendrik, et al. "Physics-aware learning and domain-specific loss design in ophthalmology." Medical image analysis vol. 76 (2021): 102314. doi: https://doi.org/10.1016/j.media.2021.102314
Burwinkel H, Matz H, Saur S, Hauger C, Trost M, Hirnschall N, Findl O, Navab N, Ahmadi SA. Physics-aware learning and domain-specific loss design in ophthalmology. Med Image Anal. 2021 Nov 27;76:102314. doi: 10.1016/j.media.2021.102314. Epub 2021 Nov 27. PMID: 34891109.
Copy Download .nbib Format: NLM
Share it on

Med Image Anal. 2021 Nov 27;76:102314. doi: 10.1016/j.media.2021.102314. Epub 2021 Nov 27.

Physics-aware learning and domain-specific loss design in ophthalmology.

Medical image analysis

Hendrik Burwinkel, Holger Matz, Stefan Saur, Christoph Hauger, Michael Trost, Nino Hirnschall, Oliver Findl, Nassir Navab, Seyed-Ahmad Ahmadi

Affiliations

  1. Computer Aided Medical Procedures, Technische Universität München, Boltzmannstraße 3, Garching bei München 85748, Germany; Carl Zeiss Meditec AG, Rudolf-Eber-Str. 11, Oberkochen 73447, Germany. Electronic address: [email protected].
  2. Carl Zeiss Meditec AG, Rudolf-Eber-Str. 11, Oberkochen 73447, Germany.
  3. Carl Zeiss Meditec AG, Göschwitzer Str. 51-52, Jena 07745, Germany.
  4. Vienna Institute for Research in Ocular Surgery, A Karl-Landsteiner Institute, Hanusch Hospital, Vienna, Austria.
  5. Computer Aided Medical Procedures, Technische Universität München, Boltzmannstraße 3, Garching bei München 85748, Germany; Computer Aided Medical Procedures, Johns Hopkins University, 3400 North Charles Street, Baltimore, MD 21218, USA.
  6. German Center for Vertigo and Balance Disorders, Ludwig-Maximilians Universität München, Marchioninistr. 15, München 81377, Germany.

PMID: 34891109 DOI: 10.1016/j.media.2021.102314

Abstract

The human cataract, a developing opacification of the human eye lens, currently constitutes the world's most frequent cause for blindness. As a result, cataract surgery has become the most frequently performed ophthalmic surgery in the world. By removing the human lens and replacing it with an artificial intraocular lens (IOL), the optical system of the eye is restored. In order to receive a good refractive result, the IOL specifications, especially the refractive power, have to be determined precisely prior to surgery. In the last years, there has been a body of work to perform this prediction by using biometric information extracted from OCT imaging data, recently also by machine learning (ML) methods. Approaches so far consider only biometric information or physical modelling, but provide no effective combination, while often also neglecting IOL geometry. Additionally, ML on small data sets without sufficient domain coverage can be challenging. To solve these issues, we propose OpticNet, a novel optical refraction network based on an unsupervised, domain-specific loss function that explicitly incorporates physical information into the network. By providing a precise and differentiable light propagation eye model, physical gradients following the eye optics are backpropagated into the network. We further propose a new transfer learning procedure, which allows the unsupervised pre-training on the optical model and fine-tuning of the network on small amounts of surgical patient data. We show that our method outperforms the current state of the art on five OCT-image based data sets, provides better domain coverage within its predictions, and achieves better physical consistency.

Copyright © 2021 Elsevier B.V. All rights reserved.

Keywords: Domain prior incorporation; IOL calculation; Physics-based learning

Conflict of interest statement

Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this pa

Publication Types