dc.contributor.author | Perez de Frutos, Javier | |
dc.contributor.author | Pedersen, Andre | |
dc.contributor.author | Pelanis, Egidijus | |
dc.contributor.author | Bouget, David Nicolas Jean-Mar | |
dc.contributor.author | Survarachakan, Shanmugapriya | |
dc.contributor.author | Langø, Thomas | |
dc.contributor.author | Elle, Ole Jakob | |
dc.contributor.author | Lindseth, Frank | |
dc.date.accessioned | 2023-09-29T08:16:40Z | |
dc.date.available | 2023-09-29T08:16:40Z | |
dc.date.created | 2023-02-26T12:29:21Z | |
dc.date.issued | 2023 | |
dc.identifier.issn | 1932-6203 | |
dc.identifier.uri | https://hdl.handle.net/11250/3092971 | |
dc.description.abstract | Purpose This study aims to explore training strategies to improve convolutional neural network-based image-to-image deformable registration for abdominal imaging. Methods Different training strategies, loss functions, and transfer learning schemes were considered. Furthermore, an augmentation layer which generates artificial training image pairs on-the-fly was proposed, in addition to a loss layer that enables dynamic loss weighting. Results Guiding registration using segmentations in the training step proved beneficial for deep-learning-based image registration. Finetuning the pretrained model from the brain MRI dataset to the abdominal CT dataset further improved performance on the latter application, removing the need for a large dataset to yield satisfactory performance. Dynamic loss weighting also marginally improved performance, all without impacting inference runtime. Conclusion Using simple concepts, we improved the performance of a commonly used deep image registration architecture, VoxelMorph. In future work, our framework, DDMR, should be validated on different datasets to further assess its value. | en_US |
dc.description.abstract | Learning deep abdominal CT registration through adaptive loss weighting and synthetic data generation | en_US |
dc.language.iso | eng | en_US |
dc.publisher | Public Library of Science | en_US |
dc.rights | Navngivelse 4.0 Internasjonal | * |
dc.rights.uri | http://creativecommons.org/licenses/by/4.0/deed.no | * |
dc.title | Learning deep abdominal CT registration through adaptive loss weighting and synthetic data generation | en_US |
dc.title.alternative | Learning deep abdominal CT registration through adaptive loss weighting and synthetic data generation | en_US |
dc.type | Peer reviewed | en_US |
dc.type | Journal article | en_US |
dc.description.version | publishedVersion | en_US |
dc.source.volume | 18 | en_US |
dc.source.journal | PLOS ONE | en_US |
dc.source.issue | 2 | en_US |
dc.identifier.doi | 10.1371/journal.pone.0282110 | |
dc.identifier.cristin | 2129318 | |
dc.relation.project | EC/H2020/722068 | en_US |
cristin.ispublished | true | |
cristin.fulltext | original | |
cristin.qualitycode | 1 | |