ISMRM25 - AI-Based Acquisition & Reconstruction: Part IV

ISMRM & ISMRT Annual Meeting & Exhibition • 10-15 May 2025 • Honolulu, Hawai'i

AI-Based Acquisition & Reconstruction: Part IV

Digital Poster

AI & Machine Learning

Tuesday, 13 May 2025

Exhibition Hall

14:30 - 15:30

Session Number: D-25

No CME/CE Credit

		Computer Number: 49 2778. Expectation-Maximization inspired training for End-to-End MRI reconstruction without fully sampled data View Presentation Video W. Shang, Y. Liu, W. Liu, Z. Zhou, P. Hu ShanghaiTech University, shanghai, China Impact: This novel self-supervised training method does not require splitting the undersampled k-space and enables end-to-end MRI reconstruction without pre-estimated coil sensitivity maps. It streamlines self-supervised reconstruction workflows and paves way for further advances in self-supervised reconstruction.
		Computer Number: 50 2779. Deep Learning-based Joint Image Reconstruction and Biomarker Estimation for Highly-accelerated Multi-shot Intravoxel Incoherent Motion DWI View Presentation Video C. Yuan, S. Chen, L. Liang, X. Xu, H. Xiong, Y. Li, T. Liu, W. Y. A. Cheung, E. S. Hui, H-C Chang The Chinese University of Hong Kong, Hong Kong, China Impact: Our proposed DL-based technique is capable of precisely reconstructing IVIM-DWI and producing IVIM-related biomarker maps within a clinically feasible acquisition time, potentially improving the quantitative evaluation and analysis of IVIM-DWI-based assessment of cerebrovascular disease.
		Computer Number: 51 2780. Feasibility of a deep learning-accelerated T1-w Dixon sequence for kidney MRI View Presentation Video T. Scheef, J. Fingerhut, H. Engel, A. Rau, L. Kolbe, C. Wilpert, M. Russe, R. Strecker, M. Nickel, F. Bamberg, J. Weiss, N. Verloh University Medical Center Freiburg, Freiburg im Breisgau, Germany Impact: The study demonstrates that deep learning-accelerated MRI improves image quality and renal cyst and microstructure detection while reducing scan time. Can it help detecting more renal lesions than conventional T1-w sequences? Can it lead to a re-classification of Bosniak?
		Computer Number: 52 2781. SUPREM: A Super-Resolution Network for Through-Plane Structure Enhancement in Multi-Contrast MRI Using Disentangled Representation Learning View Presentation Video Y. Choi, S. Jung, M. Al-masni, D. Kim, D-H Kim Yonsei University, Seoul, Korea, Republic of Impact: The proposed method has the potential to reduce discomfort during brain MRI assessments by enhancing through-plane super-resolution, thereby improving structural detail recovery and supporting more accurate diagnostics in clinical practice.
		Computer Number: 53 2782. Memory-Efficient Image Reconstruction using Diffusion Models for Accelerated 3D Non-Cartesian UTE imaging View Presentation Video J. Petersen, D. Grodzki, T. Küstner, S. Sommer Medical Image and Data Analysis (MIDAS.lab), Department of Diagnostic and Interventional Radiology, University of Tübingen, Tübingen, Germany Impact: We propose a memory-efficient diffusion model for reconstructing accelerated 3D radial UTE acquisitions, enabling high-quality, and reliable reconstructions while reducing the reconstruction time of common deep-learning methods. The model generalizes well across body parts, supporting various applications and acceleration factors.
		Computer Number: 54 2783. Robust Cardiac Cine MRI Reconstruction with Spatiotemporal Diffusion Model View Presentation Video Z. Wang, J. Huang, C. Wang, G. Yang, X. Qu Xiamen University, Xiamen, China Impact: Given the amazing robustness and generalizability, we believe that our spatiotemporal diffusion model can be an important framework for cardiac cine MRI. Furthermore, this spatiotemporal diffusion approach can be extended to inverse problems in other medical modalities involving dynamic acquisitions.
		Computer Number: 55 2784. Accelerated T2WI for Liver MRI with Deep Learning Reconstruction: A Prospective Comparison on Image Quality and Respiration Factors Y. Yang, C. Hu, H. Li, Q. Liu, R. Tang, X. Song, Z. Li Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China Impact: Respiratory-gating and breath-hold DL-T2WI sequences can be used in daily practice as standard sequences, and parameters derived from the breathing curve can aid in developing a personalized workflow for the imaging process.
		Computer Number: 56 2785. Learned K-Space Partitioning for Improved Dual-Domain Self-Supervised Image Reconstruction View Presentation Video B. Kadota, C. Millard, M. Chiew Sunnybrook Research Institute, Toronto, Canada Impact: Learned k-space partitioning enhances reconstruction quality better utilizing acquired data for reconstructions. Learned k-space partitioning provides a framework for optimizing self-supervised partitioning to diverse k-space trajectories which previously were hand-picked or sub-optimal.
		Computer Number: 57 2786. KOMET: Kspace Optimal Masking for Efficient Training of Zero-shot MRI Reconstruction View Presentation Video H. Kim, J. Joo, D. Lee, D. Hwang, T. Eo Yonsei University, Seoul, Korea, Republic of Impact: KOMET establishes a novel framework for stable zero-shot MRI reconstruction by adapting masked modeling to k-space domain. This advancement enables robust acceleration without fully sampled reference data, paving the way for broader clinical application of accelerated MRI.
		Computer Number: 58 2787. High-Resolution Deep Learning Reconstruction (HR-DLR) with Compressed Sensing for a Highly Accelerated Acquisition View Presentation Video H. Kutsuna, M. Bekku, K. Shinoda Canon Medical Systems Corporation, Tochigi, Japan Impact: High-Resolution Deep Learning Reconstruction (HR-DLR) re-defined the trade-off between scan time, image SNR and sharpness that previously limited Compressed Sensing (CS) acceleration. Combination of HR-DLR and CS makes yet higher acceleration practical, benefiting throughput and patient comfort.
		Computer Number: 59 2788. Generative Diffusion Model for Dynamic MRI Reconstruction with Temproal State Space Representation View Presentation Video Z. Zhang, C. Qin Imperial College London, London, United Kingdom Impact: Our work will enable faster and higher-quality dynamic CMR imaging for improving the MR imaging workflow and aiding in clinical diagnosis.
		Computer Number: 60 2789. Guided MRI Reconstruction via Schrödinger Bridge View Presentation Video Y. Wang, T. Zhou, J. Xie, Z. cui, B. huang, H. Zheng, Y. Zhou, H. Wang, D. Liang, Y. Zhu Shenzhen Institute of Advanced Technology，Chinese Academy of Sciences, Shenzhen, China Impact: A novel approach combines structural similarity modeling with difference compensation, introducing image editing techniques to MRI reconstruction. This paves the way for integrating MRI-guided reconstruction and image editing methods.
		Computer Number: 61 2790. Low-Rank Representation Enhanced Implicit Neural Representation for Zero-Shot Reconstruction in Highly Accelerated 3D Multi-parametric MRI H. Zhang, G. Lao, H. Wei Shanghai Jiao Tong University, Shanghai, China Impact: The proposed framework could reconstruct whole-brain T1, T2, T2* maps within just a 2.5-minute scan. This advancement holds clinical promise for tissue characterization and pathological assessment in neuroscience, and enables probability for higher resolution and additional contrasts in MRI sequences.
		Computer Number: 62 2791. Simultaneous CT-MR imaging using sparse X-ray projections View Presentation Video O. Pastor-Serrano, L. Xing Stanford University , Stanford, United States Impact: The proposed framework enables efficient, low-dose imaging with exceptional tissue contrast, which could transform fields like radiation therapy by reducing scan requirements without sacrificing image quality. This can potentially enable treatment guidance (MR) with simultaneous dose calculation (CT).
		Computer Number: 63 2792. A deep learning model to enhance temporal information for the cardiac cine imaging View Presentation Video T. C. Chao, J. Browne, S. Waddle, D. Wang, T. Leiner Mayo Clinic, Rochester, United States Impact: The proposed deep-learning based temporal interpolation for the cine CMR is independent of and compatible with existing acceleration strategies such as SENSE, Compressed Sensing and GRAPPA allowing a further increase in acceleration rate without the need for the raw data.
		Computer Number: 64 2793. Clinical AI-based MR Image Reconstruction: UK National Stakeholder Engagement View Presentation Video A. Peplinski, D. Adams, M. Miquel, P. Martin Barts Health NHS Trust, London, United Kingdom Impact: Coherent approaches for implementation of the software can be achieved after understanding how different sites have rolled out the software, what worked well alongside the challenges. Recommendations can be provided for how to inform patients of the use of AI.

Similar Session(s)