Back to Search
Start Over
Learning deep abdominal CT registration through adaptive loss weighting and synthetic data generation
- Source :
- PLoS ONE, Vol 18, Iss 2 (2023)
- Publication Year :
- 2023
- Publisher :
- Public Library of Science (PLoS), 2023.
-
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.
Details
- Language :
- English
- ISSN :
- 19326203
- Volume :
- 18
- Issue :
- 2
- Database :
- Directory of Open Access Journals
- Journal :
- PLoS ONE
- Publication Type :
- Academic Journal
- Accession number :
- edsdoj.f54e9ab7b4c041559965718de6b193de
- Document Type :
- article