TY - GEN
T1 - Towards Generalised Neural Implicit Representations for Image Registration
AU - Zimmer, Veronika A.
AU - Hammernik, Kerstin
AU - Sideri-Lampretsa, Vasiliki
AU - Huang, Wenqi
AU - Reithmeir, Anna
AU - Rueckert, Daniel
AU - Schnabel, Julia A.
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.
PY - 2024
Y1 - 2024
N2 - Neural implicit representations (NIRs) enable to generate and parametrize the transformation for image registration in a continuous way. By design, these representations are image-pair-specific, meaning that for each signal a new multi-layer perceptron has to be trained. In this work, we investigate for the first time the potential of existent NIR generalisation methods for image registration and propose novel methods for the registration of a group of image pairs using NIRs. To exploit the generalisation potential of NIRs, we encode the fixed and moving image volumes to latent representations, which are then used to condition or modulate the NIR. Using ablation studies on a 3D benchmark dataset, we show that our methods are able to generalise to a set of image pairs with a performance comparable to pairwise registration using NIRs when trained on N=10 and N=120 datasets. Our results demonstrate the potential of generalised NIRs for 3D deformable image registration.
AB - Neural implicit representations (NIRs) enable to generate and parametrize the transformation for image registration in a continuous way. By design, these representations are image-pair-specific, meaning that for each signal a new multi-layer perceptron has to be trained. In this work, we investigate for the first time the potential of existent NIR generalisation methods for image registration and propose novel methods for the registration of a group of image pairs using NIRs. To exploit the generalisation potential of NIRs, we encode the fixed and moving image volumes to latent representations, which are then used to condition or modulate the NIR. Using ablation studies on a 3D benchmark dataset, we show that our methods are able to generalise to a set of image pairs with a performance comparable to pairwise registration using NIRs when trained on N=10 and N=120 datasets. Our results demonstrate the potential of generalised NIRs for 3D deformable image registration.
KW - Generalisation
KW - Image registration
KW - Neural implicit representation
KW - Periodic activation functions
UR - http://www.scopus.com/inward/record.url?scp=85187654407&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-53767-7_5
DO - 10.1007/978-3-031-53767-7_5
M3 - Conference contribution
AN - SCOPUS:85187654407
SN - 9783031537660
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 45
EP - 55
BT - Deep Generative Models - Third MICCAI Workshop, DGM4MICCAI 2023, Held in Conjunction with MICCAI 2023, Proceedings
A2 - Mukhopadhyay, Anirban
A2 - Oksuz, Ilkay
A2 - Engelhardt, Sandy
A2 - Zhu, Dajiang
A2 - Yuan, Yixuan
PB - Springer Science and Business Media Deutschland GmbH
T2 - 3rd Workshop on Deep Generative Models for Medical Image Computing and Computer Assisted Intervention, DGM4MICCAI 2023 Held in Conjunction with 26th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2023
Y2 - 8 October 2023 through 12 October 2023
ER -