ISMRM25 - AI-Powered Analysis in Cardiovascular MRI

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

AI-Powered Analysis in Cardiovascular MRI

Digital Poster

AI & Machine Learning

Thursday, 15 May 2025

Exhibition Hall

09:15 - 10:15

Session Number: D-26

No CME/CE Credit

		Computer Number: 1 4269. Assessing Deep Learning's Ability to Segment Aortic Valve Calcifications in Contrast-Free Cardiac MRI View Presentation Video E. Almar-Munoz, C. Kremser, M. Haltmeier, A. Mayr Medical University of Innsbruck, Innsbruck, Austria Impact: Calcifications are not visible on contrast-free CMR to the human eye, so we tested AI segmentation using CTA-registered labels, achieving low results (DSC 0.309). While a larger dataset or improved registration may help, the task's physical feasibility remains uncertain.
		Computer Number: 2 4270. AI-Driven Quantification of Aortic Diameters from Contrast-Enhanced MRA of the Thoracic Aorta View Presentation Video C. Apostolidis, E. Johnson, H. Berhane, D. Dushfunian, S. Cohn, B. Allen, A. Katsaggelos, M. Markl Northwestern University, Chicago, United States Impact: Aortic disease risk assessment relies on imaging-based aortic diameter surveillance. We have developed an automated, AI-driven diameter quantification pipeline, aiming to improve manual processing speed and reproducibility. We have achieved moderate agreement with manual ground truth.
		Computer Number: 3 4271. Deep learning-based segmentation of left ventricular myocardium in cardiac magnetic resonance elastography View Presentation Video V. Atamaniuk, M. Anders, M. Obrzut, A. Pozaruk, Ł. Hańczyk, B. Obrzut, M. Skoczylas, I. Sack, M. Cholewa University of Rzeszow, Rzeszow, Poland Impact: This study demonstrates deep learning’s capability to automate left ventricular myocardium segmentation in cardiac MR elastography, enabling faster and more consistent myocardial stiffness assessments. Such advancements could enhance cardiovascular disease diagnostics, paving the way for improved clinical decision-making in cardiology.
		Computer Number: 4 4272. Automated carotid artery atherosclerotic inflammation segmentation on PET-MRI: Mitigating partial volume effect View Presentation Video R. Li, A. Jha, P. Woodard, J. Zheng Washington University in St. Louis, St. Louis, United States Impact: By improving the assessment of carotid inflammation, this methodology has the potential to inform clinical decisions and interventions, ultimately reducing cardiovascular risk and mortality.
		Computer Number: 5 4273. Automated left atrial function analysis using AI is a stronger predictor of survival than physician-measured left ventricular ejection fraction View Presentation Video H. C. Cheung, S. Zaman, K. Vimalesvaran, K. Chow, P. Kellman, H. Xue, R. Davies, G. D. Cole, C. Manisty, J. C. Moon, J. P. Howard Imperial College London, London, United Kingdom Impact: Atrial function measured automatically using inline AI is a strong and incremental predictor of patient survival. This enables new biomarkers to be easily translated into clinical workflow for improved patient care.
		Computer Number: 6 4274. Deep Learning-Based Quantification of Intraplaque Hemorrhage Reveals Impacts on Long-Term Carotid Plaque Progression View Presentation Video Y. Guo, D. Hippe, X. Wang, G. Canton, K. Zhang, A. Tang, M. Ferguson, M. Mossa-Basha, N. Balu, T. Hatsukami, C. Yuan University of Washington, Seattle, United States Impact: This study underscores the significance of IPH in carotid plaque progression, offering a precise deep learning-based tool for monitoring. It enables more effective risk assessment and personalized management strategies, sparking new research into long-term IPH effects on asymptomatic patients.
		Computer Number: 7 4275. Artificial Intelligence for Gadolinium-Free CMR Tissue Charactization Using Deep Learning–Based Virtual Native Enhancement X. Jiang, Y. Wu, Y. Wang, T. Liu The First Hospital of China Medical University, Shenyang, China Impact: Virtual LGE has substantial potential to replace LGE in diagnosing various cardiovascular diseases, providing a more rapid, cost-effective scan and eliminating contrast agent risks.
		Computer Number: 8 4276. Cardiac landmark localization on axial stacks without dedicated ground truth View Presentation Video G. Delso, E. Ali, J. Names, D. Rettmann, M. Janich GE HealthCare, Barcelona, Spain Impact: The proposed approach eliminates the need for manual annotations for the training of some cardiac models. Using outputs from existing long-axis models as surrogate ground truth simplifies the creation and maintenance of the training database.
		Computer Number: 9 4277. Unsupervised phase unwrapping for aortic 4D flow MRI using deep image prior View Presentation Video Y. Ren, H. Hong, Z. Zhou, P. Hu ShanghaiTech University, Shanghai, China Impact: PUDIP outperforms the conventional methods and can yield accurate flow velocity quantifications for 4D flow MRI.
		Computer Number: 10 4278. Co-Evolution of CNNs and Learning Environments for Automated Post-Processing of T2 Parametric Mapping in Cardiovascular Magnetic Resonance View Presentation Video T. Hadler, C. Ammann, P. Reisdorf, S. Lange, J. Schulz-Menger Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Berlin, Germany, Berlin, Germany Impact: The evolutionary algorithm improves CNN performance in cardiovascular MRI by dynamically optimizing learning environments, enhancing adaptability across varied imaging conditions. This approach strengthens model generalizability and reliability, making it a promising method for advancing robust AI tools in clinical practice.
		Computer Number: 11 4279. Automated Deep Learning Pipeline for Pulse Wave Velocity Measurement in UK Biobank MRI Data View Presentation Video Y. Jiang, T. Yao, K. Punjabi, D. Knight, J. Steeden, R. Davies, V. Muthurangu University College London, London, United Kingdom Impact: Our model enables automatic aortic arch PWV measurement for UK Biobank subjects, which can be used in the investigation of arterial stiffness and prediction of cardiovascular disease for a large population.
		Computer Number: 12 4280. Physics-informed Model Selection for Robust Deep Learning Segmentation of Multi-center Perfusion CMR: Initial Findings from the SCMR Registry View Presentation Video D. M. Yalcinkaya, A. M. Sohi, K. Youssef, L. Zamudio, M. Elliott, V. Polsani, R. Dharmakumar, R. Judd, M. Tong, D. Shah, O. Simonetti, B. Sharif Purdue University, West Lafayette, United States Impact: The proposed hybrid approach has the potential to improve the reliability of fully automated stress/rest FPP analysis in clinical settings and in multi-center clinical trials.
		Computer Number: 13 4281. Contrast-Free CMR-based Transcatheter Aortic Valve Implantation Planning: Aortic Annulus and Coronary Ostias detection View Presentation Video E. Almar-Munoz, M. Pamminger, C. Kremser, M. Haltmeier, A. Mayr Medical University of Innsbruck, Innsbruck, Austria Impact: Our CMR based TAVI planning algorithm advances by automating pre-procedural assessments without iodinated contrast. For the AA segmentation, the double-network architecture and the unwrapping technique effectively address the challenge of segmenting a virtual plane. Blind tests reafirm our results.
		Computer Number: 14 4282. Arbitrary Factor Super-Resolution for 3D Whole-Heart MRI Using a Frequency-Domain Informed Neural Network C. Maciel, Q. Zou University of Texas Southwestern Medical Center, Fort Worth, United States Impact: By implementing frequency-domain regularization inform network training and arbitrary factor super-resolution, the proposed method offers the potential to decrease acquisition time in 3D whole-heart MRI, while maintaining fine image detail important for diagnostic utility.
		Computer Number: 15 4283. Semi-supervised 3D Myocardial Segmentation for Whole-Heart Joint T1/T2 mapping with Self-trained nnUNet View Presentation Video C. Rivera, A. Hua, R. Botnar, C. Prieto IMPACT, Center of Interventional Medicine for Precision and Advanced Cellular Therapy, Santiago, Chile Impact: Semi-supervised 3D nnUNet enables accurate myocardial segmentation in 3D whole-heart joint T1/T2 mapping, even with limited labeled data. This could improve efficiency, reduce manual segmentation effort, and accelerate the diagnosis of myocardial diseases.
		Computer Number: 16 4284. Improve the Accuracy of Right Ventricle Segmentation in Cardiac Magnetic Resonance Images by UNet++ View Presentation Video C-W Lin, H-H Peng National Tsing Hua University, New Taipei City, Taiwan Impact: The UNet++ model combined with special encoders provided a robust solution to RV segmentation, with the potential to enhance cardiac imaging analysis and support clinical decision-making by reducing manual intervention and increasing accuracy in heart function assessment.

Similar Session(s)