Vol2Flow: Segment 3D Volumes Using a Sequence of Registration Flows
Bitarafan A, Azampour MF, Bakhtari K, Baghshah MS, Keicher M, Navab N (2022)
Publication Type: Conference contribution
Publication year: 2022
Journal
Publisher: Springer Science and Business Media Deutschland GmbH
Book Volume: 13434 LNCS
Pages Range: 609-618
Conference Proceedings Title: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Event location: Singapore, SGP
ISBN: 9783031164392
DOI: 10.1007/978-3-031-16440-8_58
Abstract
This work proposes a self-supervised algorithm to segment each arbitrary anatomical structure in a 3D medical image produced under various acquisition conditions, dealing with domain shift problems and generalizability. Furthermore, we advocate an interactive setting in the inference time, where the self-supervised model trained on unlabeled volumes should be directly applicable to segment each test volume given the user-provided single slice annotation. To this end, we learn a novel 3D registration network, namely Vol2Flow, from the perspective of image sequence registration to find 2D displacement fields between all adjacent slices within a 3D medical volume together. Specifically, we present a novel 3D CNN-based architecture that finds a series of registration flows between consecutive slices within a whole volume, resulting in a dense displacement field. A new self-supervised algorithm is proposed to learn the transformations or registration fields between the series of 2D images of a 3D volume. Consequently, we enable gradually propagating the user-provided single slice annotation to other slices of a volume in the inference time. We demonstrate that our model substantially outperforms related methods on various medical image segmentation tasks through several experiments on different medical image segmentation datasets. Code is available at https://github.com/AdelehBitarafan/Vol2Flow.
Involved external institutions
Sharif University of Technology / دانشگاه صنعتی شریف (SUT)
Iran, Islamic Republic of (IR)
Technische Universität München (TUM)
Germany (DE)
Iran, Islamic Republic of (IR)
Technische Universität München (TUM)
Germany (DE)
How to cite
APA:
Bitarafan, A., Azampour, M.F., Bakhtari, K., Baghshah, M.S., Keicher, M., & Navab, N. (2022). Vol2Flow: Segment 3D Volumes Using a Sequence of Registration Flows. In Linwei Wang, Qi Dou, P. Thomas Fletcher, Stefanie Speidel, Shuo Li (Eds.), Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (pp. 609-618). Singapore, SGP: Springer Science and Business Media Deutschland GmbH.
MLA:
Bitarafan, Adeleh, et al. "Vol2Flow: Segment 3D Volumes Using a Sequence of Registration Flows." Proceedings of the 25th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2022, Singapore, SGP Ed. Linwei Wang, Qi Dou, P. Thomas Fletcher, Stefanie Speidel, Shuo Li, Springer Science and Business Media Deutschland GmbH, 2022. 609-618.
BibTeX: Download