TY - GEN
T1 - ICoNIK
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
AU - Spieker, Veronika
AU - Huang, Wenqi
AU - Eichhorn, Hannah
AU - Stelter, Jonathan
AU - Weiss, Kilian
AU - Zimmer, Veronika A.
AU - Braren, Rickmer F.
AU - Karampinos, Dimitrios C.
AU - Hammernik, Kerstin
AU - Schnabel, Julia A.
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.
PY - 2024
Y1 - 2024
N2 - Motion-resolved reconstruction for abdominal magnetic resonance imaging (MRI) remains a challenge due to the trade-off between residual motion blurring caused by discretized motion states and undersampling artefacts. In this work, we generate blurring-free motion-resolved abdominal reconstructions by learning a neural implicit representation directly in k-space (NIK). Using measured sampling points and a data-derived respiratory navigator signal, we train a network to generate continuous signal values. To aid the regularization of sparsely sampled regions, we introduce an additional informed correction layer (ICo), which leverages information from neighboring regions to correct NIK’s prediction. The proposed generative reconstruction methods, NIK and ICoNIK, outperform standard motion-resolved reconstruction techniques and provide a promising solution to address motion artefacts in abdominal MRI.
AB - Motion-resolved reconstruction for abdominal magnetic resonance imaging (MRI) remains a challenge due to the trade-off between residual motion blurring caused by discretized motion states and undersampling artefacts. In this work, we generate blurring-free motion-resolved abdominal reconstructions by learning a neural implicit representation directly in k-space (NIK). Using measured sampling points and a data-derived respiratory navigator signal, we train a network to generate continuous signal values. To aid the regularization of sparsely sampled regions, we introduce an additional informed correction layer (ICo), which leverages information from neighboring regions to correct NIK’s prediction. The proposed generative reconstruction methods, NIK and ICoNIK, outperform standard motion-resolved reconstruction techniques and provide a promising solution to address motion artefacts in abdominal MRI.
KW - MRI Reconstruction
KW - Motion-Resolved Abdominal MRI
KW - Neural Implicit Representations
KW - Parallel Imaging
UR - http://www.scopus.com/inward/record.url?scp=85187660449&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-53767-7_18
DO - 10.1007/978-3-031-53767-7_18
M3 - Conference contribution
AN - SCOPUS:85187660449
SN - 9783031537660
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 183
EP - 192
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
Y2 - 8 October 2023 through 12 October 2023
ER -