Citation

BibTex format

@article{Wang:2025:10.1109/TBME.2025.3574090,
author = {Wang, Z and Xiao, M and Zhou, Y and Wang, C and Wu, N and Li, Y and Gong, Y and Chang, S and Chen, Y and Zhu, L and Zhou, J and Cai, C and Wang, H and Jiang, X and Guo, D and Yang, G and Qu, X},
doi = {10.1109/TBME.2025.3574090},
journal = {IEEE Trans Biomed Eng},
pages = {3642--3654},
title = {Deep Separable Spatiotemporal Learning for Fast Dynamic Cardiac MRI.},
url = {http://dx.doi.org/10.1109/TBME.2025.3574090},
volume = {72},
year = {2025}
}

RIS format (EndNote, RefMan)

TY  - JOUR
AB - OBJECTIVE: Dynamic magnetic resonance imaging (MRI) plays an indispensable role in cardiac diagnosis. To enable fast imaging, the k-space data can be undersampled but the image reconstruction poses a great challenge of high-dimensional processing. This challenge necessitates extensive training data in deep learning reconstruction methods. In this work, we propose a novel and efficient approach, leveraging a dimension-reduced separable learning scheme that can perform exceptionally well even with highly limited training data. METHODS: We design this new approach by incorporating spatiotemporal priors into the development of a Deep Separable Spatiotemporal Learning network (DeepSSL), which unrolls an iteration process of a 2D spatiotemporal reconstruction model with both temporal low-rankness and spatial sparsity. Intermediate outputs can also be visualized to provide insights into the network behavior and enhance interpretability. RESULTS: Extensive results on cardiac cine datasets demonstrate that the proposed DeepSSL surpasses state-of-the-art methods both visually and quantitatively, while reducing the demand for training cases by up to 75%. Additionally, its preliminary adaptability to unseen cardiac patients has been verified through a blind reader study conducted by experienced radiologists and cardiologists. Furthermore, DeepSSL enhances the accuracy of the downstream task of cardiac segmentation and exhibits robustness in prospectively undersampled real-time cardiac MRI. CONCLUSION: DeepSSL is efficient under highly limited training data and adaptive to patients and prospective undersampling. SIGNIFICANCE: This approach holds promise in addressing the escalating demand for high-dimensional data reconstruction in MRI applications.
AU - Wang,Z
AU - Xiao,M
AU - Zhou,Y
AU - Wang,C
AU - Wu,N
AU - Li,Y
AU - Gong,Y
AU - Chang,S
AU - Chen,Y
AU - Zhu,L
AU - Zhou,J
AU - Cai,C
AU - Wang,H
AU - Jiang,X
AU - Guo,D
AU - Yang,G
AU - Qu,X
DO - 10.1109/TBME.2025.3574090
EP - 3654
PY - 2025///
SP - 3642
TI - Deep Separable Spatiotemporal Learning for Fast Dynamic Cardiac MRI.
T2 - IEEE Trans Biomed Eng
UR - http://dx.doi.org/10.1109/TBME.2025.3574090
UR - https://www.ncbi.nlm.nih.gov/pubmed/40434852
VL - 72
ER -

Contact


For enquiries about the MRI Physics Collective, please contact:

Mary Finnegan
Senior MR Physicist at the Imperial College Healthcare NHS Trust

Pete Lally
Assistant Professor in Magnetic Resonance (MR) Physics at Imperial College

Jan Sedlacik
MR Physicist at the Robert Steiner MR Unit, Hammersmith Hospital Campus