Cape Town - 2026 ISMRM-ISMRT Annual Meeting and Exhibition

Automated Cardiovascular MRI

Digital Poster

Cardiovascular

Wednesday, 13 May 2026

Digital Posters Row A

14:35 - 15:30

Session Number: 560-04

No CME/CE Credit

Similar Sessions

Session on automated techniques to enable push button cardiovascular MRI from acquisition to analysis

		560-04-001. Deep Learning-Based Automation of Aortic Flow View Planning in Cardiac MRI Gaspar Delso, Ada Doltra, Collin Buelo, Fara Nikbeh, Jane Names, Willian Cordeiro, Martin Janich GE HealthCare (ES), Spain Impact: This model advances cardiac MRI by automating aortic flow view prescription, reducing operator variability, improving scan efficiency, and enabling broader clinical adoption.
		560-04-002. Contrast-Agnostic Deep-Learning Segmentation of Free-Running Whole-Heart MRI Léonard Treil, Shuailong Zhu, Angela Rocca, Jérôme Yerly, Robert Holtackers, Matthias Stuber, Roger Hullin, David Rotzinger, Augustin Ogier, Ruud van Heeswijk Lausanne University Hospital (CHUV) and University of Lausanne (UNIL), Lausanne, Switzerland Impact: This study enables contrast-agnostic automated segmentation of free-running whole-heart MRI, allowing consistent cardiac function quantification across scanners and protocols. The proposed framework reduces reliance on sequence-specific models, facilitating clinical adoption of free-running cardiac MRI.
		560-04-003. Automated Streaking Artifact Suppression via Signal Processing Reduces Spurious Values in Pixel-Wise MBF Quantification Dima Bishara Radiology, Feinberg School of Medicine, Northwestern Medicine, Chicago, United States of America Impact: This automated streak artifact suppression method enables reliable pixel-wise myocardial blood flow quantification in radial cardiac perfusion MRI without field-of-view truncation or signal-to-noise ratio loss. The technique produces resting flow values consistent with PET literature, potentially improving diagnostic confidence.
		560-04-004. Automated Quantification of Peri-Cardiac Fat on Non-Gated Thoracic MRI Using nnU-Net for Opportunistic Screening Ahmed Gouda, Saqib Basar, Kristin Mae Esteban, Javad Khaghani, Daniel Daly-Grafstein, Duc Nguyen, Saurabh Garg, Siavash Khallaghi, Yuntong Ma, Sam Hashemi Prenuvo, Inc, San Francisco, United States of America Impact: Deep learning-based cardiac fat quantification reduces reliance on time- and resource-intensive gated MR sequences. This faster, automated approach enables opportunistic cardiometabolic disease screening and risk stratification on both new and existing scans, supporting earlier detection and prevention strategies.
		560-04-005. 5D Free-running CMR Reconstruction with Motion-Resolved Convolutional Gated Recurrent Network (MR-CGRNet) Yitong Yang, Jérôme Yerly, Davide Piccini, Matthias Stuber, John Oshinski Biomedical Engineering, Georgia Institute of Technology and Emory University, Atlanta, United States of America Impact: The proposed MR-CGRNet for 5D free-running CMR reconstruction significantly reduced computational time from an hours-long reconstruction to approximately 10 minutes while preserved image features capturing cardiac and respiratory motion. These results demonstrate the potential for in-line reconstruction 5D free-running CMR.
		560-04-006. Free-Running 3D Cardiac T1 Mapping using Look-Locker and Variable Flip Angle Imaging with Low-Rank Modeling Paul Han, Thibault Marin, Wonil Lee, Didi Chi, Guodong Weng, Ismaël Mounime, Jaehun Lee, John Stendahl, Georges El Fakhri, Chao Ma Yale Biomedical Imaging Institute, Yale University School of Medicine, New Haven, United States of America Impact: The proposed method utilizing Look-Locker and variable flip angle imaging with low-rank modeling allows free-breathing, whole-heart cardiac T₁ mapping with improved through-plane spatial resolution compared to the existing method. The method is potentially useful for cardiovascular research and clinical applications.
		560-04-007. Diffusion-based k-space inpainting for improved 5D free-running CMR reconstruction Thomas Coudert, Amit Rand, Xinran Gao Xinran Gao, Zhengyang Ming, J. Paul Finn, Dan Ruan, Kim-Lien Nguyen David Geffen School of Medicine at UCLA and VA Greater Los Angeles Healthcare System, Los Angeles, United States of America Impact: A diffusion-based k-space inpainting strategy could accelerate free-running CMR reconstruction while preserving fidelity and motion detail. By improving generalization across contrasts and varying anatomy, this framework may enable fast, data-consistent 5D cardiac MRI for pediatric and adult clinical imaging.
		560-04-008. Optimized System-calibration-based B0-shimming Procedure with Global Solver for T2-weighted Cardiac MRI at 7T Maxim Terekhov, Istvan Homolya, Marwan Hamid, Rebekka Grampp, Anja Stadtmueller, Anna Frey, Wolfgang Bauer, Ulrich Hofmann, Laura Schreiber Comprehensive Heart Failure Center, University Hospital Würzburg, Würzburg, Germany Impact:* The proposed pipeline for 3rd-order B₀-shimming may substantially reduce susceptibility-induced artefacts in the heart at 7 T. The combination of shim-coils calibration, robust phase-unwrapping, and global optimization establishes a transferable framework to control B₀-field homogeneity in the ultra-high-field cardiac MRI.
		560-04-009. A Preliminary Evaluation of Optimal Pilot Tone Frequency for Cardiac Motion Monitoring in MRI Siyuan Liu, Jianmin Wang Magnetic resonance, Siemens Shenzhen Magnetic Resonance Ltd., Shenzhen, China Impact: We found that pilot tones near 100 MHz offer optimal modulation for cardiac motion tracking while maintaining spectral separation from MRI frequencies. This frequency region is potentially the best pilot tone range for robust, interference-free motion sensing in MRI systems.
		560-04-010. MRI in Clinical Practice: A-LIKNet Single-Breath-Hold CINE for LV/RV Function When Breath-Holds Are Limited Patrick Krumm, Jens Kübler, Stanislau Chekan, Siying Xu, Aya Ghoul, Jan Brendel, Konstantin Nikolaou, Thomas Küstner University Hospital Tuebingen, Tuebingen, Germany Impact: A-LIKNet single-breath-hold CINE cuts cine time to 8–12 s while preserving LV/RV volumes and EF, enabling reliable exams in dyspneic or pediatric patients and supporting same-session management when multi-breath-hold imaging is not feasible.
		560-04-011. Echo Planar Imaging Enables Single Breath-hold 3D Late Gadolinium Enhancement Imaging Oscar Goldman, Karl Kunze, Tevfik Ismail, Amedeo Chiribiri, John Whitaker, Radhouene Neji King's College London, London, United Kingdom Impact: This work enables the acquisition of 3D LGE images in a single breath-hold. This could reduce the scan time for cardiac MR exams.
		560-04-012. AdaptCMR: Adaptive Spectrally Guided Mixture-of-Experts Network for Efficient Multi-Contrast Cardiac MRI Reconstruction Mahan Veisi, Kian Anvari Hamedani, Shahabedin Nabavi Shahid Beheshti University, Tehran, Iran (Islamic Republic of) Impact: AdaptCMR is a parameter-efficient, detail-preserving deep learning framework for cardiac MRI reconstruction. It uses a spectrally guided mixture of experts to generalize across different views, contrasts, and acceleration factors, preserving fine anatomical detail with fewer parameters and faster inference.
		560-04-013. Extended Phase Graph-Informed Deep Learning for Accelerated and Improved Joint T1 and T2 Cardiac Mapping Catarina Carvalho, Andreia Gaspar, Rita Nunes, Teresa Correia Center of Marine Sciences (CCMAR), Faro, Portugal Impact: A deep learning network performs accelerated joint cardiac T1 and T2 estimation directly from k-space, where signals follow Extended Phase Graph modelling, greatly improving the estimated T1 map accuracy compared to methods that disregard T2, and simultaneously shortening protocol times.
		560-04-014. Optimizing Efficiency & Patient Comfort in ferumoxytol-enhanced Whole-body MRA: Outside-Scanner vs. On-Table Protocols Jing Liang, Guangxiang Si, Ruijing Xin, Xiance Zhao, Bing Zhang Nanjing Drum Tower Hospital, Affiliated Hospital of Medical School, Nanjing University, Nanjing, China Impact: The Outside-scanner FE-WB-MRA protocol cuts room time, eases scheduling bottlenecks, improves patient anxiety, and validates ferumoxytol’s value for efficient, quality MRI in patients.
		560-04-015. Supervised Residual U-Net for Pixel-wise Principal Strain Mapping from Cine SSFP Using Physics-Based Training Data Ahmed Hassan, Osama Mahmoud, Muhannad Abdallah, Ali Badran, Mishkat Habib, Tamer Basha, Ahmed Gharib, Ahmed Abdelfadeel, Khaled Abd-Elmoniem Norwegian University of Science and Technology NTNU, Norway Impact: This work provides a supervised benchmark for pixel‑wise principal strain from cine SSFP MRI with strong quantitative validation, reducing reliance on manual or boundary-related feature-tracking, supporting reproducible analysis and informing future unsupervised/hybrid approaches for scalable clinical deployment.