RPS 903 - Hot Topic: low-field MRI

Author Block: K. Aisikaier; Beijing/CN
Purpose: This study aims to noninvasively assess the prevalence of LM using cardiac magnetic resonance (CMR) and explore its association with sudden cardiac death (SCD) in patients with prior myocardial infarction (PMI).
Methods or Background: In this single-center retrospective study, 896 patients with PMI were included. PMI was confirmed by CMR evidence of subendocardial or transmural late gadolinium enhancement (LGE) in a coronary artery-specific distribution. LM was confirmed using frequency-based sequences (chemical shift or fat-water separation sequence) or T1 contrast-based sequences (short T1 inversion recovery [STIR] or pre-contrast T1-weighted sequence). The primary endpoint was a composite of SCD, aborted SCD, or appropriate implantable cardioverter-defibrillator (ICD) therapy.
Results or Findings: LM was detected in 127 (14.2%) patients. These patients had larger infarcts (23.4% vs. 19.9%, p < 0.001) and higher SCD event rates (11.8% vs. 6.1%, p = 0.035). Over 64 months median follow-up, 62 (6.9%) reached the endpoint. LM independently predicted SCD (adjusted HR: 2.28; 95% CI: 1.27–4.11; p = 0.006) after adjusting for after adjusting for age, sex, LVEF≤35%, gray zone, LGE extent, and end-diastolic volume index (EDVi). LM improved SCD risk prediction beyond LVEF≤35% (C-statistic: 0.58 to 0.62; log-likelihood: -398 to -394, p = 0.007) , especially in patients with LVEF >35% or LGE≤24.4%(optimal cutoff 24.4%; AUC=0.751, 95% CI 0.697–0.806) .
Conclusion: In patients with PMI, the presence of LM was independently associated with SCD risk, demonstrating incremental prognostic value beyond conventional risk markers, including LVEF. Furthermore, LM maintained significant predictive value in patients with preserved LVEF (>35%), suggesting its particular utility for risk stratification in the current guideline-defined lower-risk population.
Limitations: As a large retrospective cohort study conducted at a tertiary referral center, this investigation carries an inherent risk of referral bias.
Funding for this study: This study is supported by the National Key R&D Program of China (Nos. 2021YFF0501400 and 2021YFF0501404) and Key Project of National Natural Science Foundation of China (No. 82430066).
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: J-k. Li, Y. Zhigang, R. Shi, W-r. Li; Chengdu/CN
Purpose: To investigate the nonlinear associations between left ventricular end-diastolic diameter (LVEDD) and left ventricular strain parameters derived from cardiac magnetic resonance (CMR) in patients with metabolic syndrome (Mets), and to determine the relative impact of metabolic factors including diabetes mellitus (DM), hypertension (HTN), and body mass index (BMI) on strain behavior.
Methods or Background: We retrospectively analyzed cine CMR data from August 2019 to November 2021 from patients with Mets. Global myocardial strain parameters—LV radial peak strain (LVPRS), circumferential peak strain (LVPCS), longitudinal peak strain (LVPLS), and early diastolic radial strain rate (LV0PDSR-L0)—were quantified. Nonlinear polynomial regression models were constructed to examine associations between LVEDD (short-axis view) and strain parameters, adjusting for sex, age, DM, HTN, and BMI.
Results or Findings: A total of 105 patients with Mets (51.64 ± 12.49 years, 70.5% male) were recruited. LVEDD exhibited a significant nonlinear association with all strain parameters, including LVPRS (R² = 0.507, p<0.001), U-shaped trend; LVPCS (R² = 0.491, p<0.001), inverted U-shaped; LVPLS (R² = 0.315, p=0.017), mildly inverted U-shaped; LV0PDSR-L0 (R² = 0.204, p=0.058), weak U-shaped trend. After adjusting for sex, age, DM, HTN, and BMI in Mets patients, LVEDD and its quadratic term were the only two significant predictors in LVPRS and LVPCS models(p<0.05). Among the covariates, only sex showed a significant independent association with LVPLS(p=0.033).
Conclusion: In patients with Mets, LVEDD is nonlinearly associated with myocardial strain parameters, reflecting complex remodeling dynamics. CMR-derived LV strain analysis may serve as a sensitive marker for subclinical myocardial dysfunction beyond traditional metabolic risk factors and early treatments could be given.
Limitations: Firstly, this is a retrospective single-center study with a limited sample size. Secondly, the causality and temporal sequence relationship between LVEDD and strain parameters cannot to be comfirmed.
Funding for this study: None.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This study was approved by the Ethics Committee of West China Hospital of Sichuan University. Written informed consent was obtained. This study complied with the Declaration of Helsinki.

Author Block: N. Xu, Y. Xu; Jiangsu Province/CN
Purpose: To explore the value of four-dimensional flow cardiovascular magnetic resonance (4D Flow CMR) in connective tissue disease-associated pulmonary arterial hypertension (CTD-PAH) risk stratification by comparing its diagnostic efficacy and parameter consistency with two-dimensional phase-contrast magnetic resonance imaging (2D PC-MRI).
Methods or Background: Eighty-five CTD-PAH patients were retrospectively enrolled and categorized into two groups per 2022 ESC/ERS guidelines: low/intermediate-low risk (Group 1, n=67) and intermediate-high/high risk groups (Group 2, n=18). All underwent 2D PC-MRI and 4D Flow CMR. Right ventricular (RV) function parameters and pulmonary artery [main (MPA), left (LPA), right (RPA)] hemodynamic parameters were extracted. Inter-group comparisons, logistic regression, ROC analysis (DeLong test), and consistency (intraclass correlation coefficient, ICC) analyses were performed.
Results or Findings: Compared with Group1, Group2 showed increased RV end-systolic volume index (RVESVI), RV end-diastolic volume index (RVEDVI), RV myocardial mass index (RVMMI) and decreased RV ejection fraction (RVEF) (all P<0.05). 2D PC-MRI revealed lower relative area change (RAC), peak velocity (PV) and mean velocity (MV) of MPA (all P<0.01). 4D Flow CMR revealed prolonged MPA vortex duration (24% vs 12%, P<0.01), and lower PV/MV in MPA/LPA/RPA (all P<0.05). Multivariate Logistic analysis showed MPA PV and vortex duration from 4D Flow CMR and MPA PV from 2D PC-MRI were independent factors for distinguishing medium-high/high-risk patients. The consistency of MPA PV between the two techniques was good (ICC=0.765), while that of MPA MV was moderate (ICC=0.562).
Conclusion: 4D-derived MPA PV and vortex duration are independent predictors for CTD-PAH risk stratification, the combined model based on 4D parameters has superior diagnostic efficacy over 2D PC-MRI, improving CTD-PAH risk stratification accuracy.
Limitations: Not applicable
Funding for this study: Not applicable
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: M. F. Kazanbaş, B. Tütüncüoğlu, T. Banaz, M. Kadıoğlu, F. Köse, T. Y. Kılıç, N. Okumuş, A. Arslan, S. Ozkök; Istanbul/TR
Purpose: This study aimed to evaluate the effect of contrast administration on volumetric and functional parameters
of the left ventricle(LV) measured by cardiac magnetic resonance cine imaging, with particular emphasis
on paired comparison and Bland–Altman agreement analysis.
Methods or Background: Cardiac cine MRI is a reference method for quantifying LV volumes and ejection fraction (EF). However,
the influence of contrast administration on volumetric assessment remains debated. In this retrospective
analysis, 42 patients undergoing both pre- and post-contrast cine underwent 1.5 Tesla CMR were
included. LV end-diastolic volume index(EDVI), end-systolic volume index(ESVI), stroke volume index
(SVI), ejection fraction(EF), and heart rate(BPM) were measured. ECG/Holter were used to assessed
arrhythmia. Results were expressed as median(25th–75th percentile) and compared using paired
analysis with Bland–Altman evaluation. Significance was set at p<0.05.
Results or Findings: Pre-contrast cine measurements demonstrated a median LV-EDVI of 72.5 mL/m2, LV-ESVI 27.5 mL/m2, LV-SVI 41.0 mL/m2, LV-EF 60.5%, and heart rate 86.5 bpm(68.0–101.8). Post-contrast, LV-EDVI was 69.0 mL/m2, LV-ESVI 36.5 mL/m2, LV-SVI 33.5 mL/m2, LV-EF 52.5%, and heart rate 87.5 bpm(76.8–111.3). Compared with pre-contrast, post-contrast cine demonstrated a significant reduction in LV-EF(Δ–10.4 ± 9.6%, p<0.01) and LV-SVI (Δ–9.0 ± 12.3 mL/m2, p=0.02), while changes in LV-EDVI and LV-ESVI were not significant. Heart rate significantly increased (Δ +12.8±21.3, p=0.04). Bland–Altman analysis confirmed systematically lower post-contrast values, with a mean bias of –10.4%(LoA –35.2 to +14.5) for LV-EF and –9.0 mL/m2(LoA –56.9 to +38.9) for LV-SVI, indicating wide inter-individual variability.
Conclusion: Contrast administration may lead to systematic underestimation of EF and SVI on cine MRI, whereas
EDVI and ESVI remain stable. Awareness of this effect is important for consistent longitudinal follow-up
and clinical decision-making.
Limitations: The small sample size of only 42 patients restricts the generalizability of the findings.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: An application was made to the ethics committee of Başakşehir Çam and Sakura City Hospital.

Author Block: K. XU, W. Liu, Y. Xu; Nan Jing/CN
Purpose: To evaluate the value of multiparametric cardiac magnetic resonance (CMR) in assessing myocardial involvement in patients with antiphospholipid syndrome (APS).
Methods or Background: This retrospective study enrolled 30 APS patients and 30 healthy controls who underwent CMR between January 2022 and June 2025. CMR assessment encompassed biventricular structural and functional parameters, and myocardial tissue characterization parameters. Subgroup analyses were performed based on secondary APS (SAPS) or primary APS (PAPS), and positive or negative late gadolinium enhancement (LGE). Spearman correlation analysis was used to explore the correlation between CMR parameters and laboratory indicators.
Results or Findings: Compared with the control group,APS group showed reduced biventricular ejection fractions and increased left ventricular end systolic volume index (LVESVI), native T1, T2 values and extracellular volume (ECV) (all P < 0.01). Subgroup analysis revealed that right ventricular end diastolic volume index (RVEDVI), stroke volume index (SVI), cardiac index (CI), native T1, and ECV (32.55 ± 4.45% vs. 27.94 ± 1.86%, P = 0.020) in SAPS group were significantly higher than those in PAPS group (all P < 0.05). In addition, LGE in SAPS group was more frequently localized in the left ventricular anterior wall, inferior wall, and interventricular septum, and subendocardial involvement was more common. The LGE-positive group exhibited lower left ventricular ejection fraction (LVEF) and higher left ventricular end systolic volume index (LVESVI) and mass index (MI) (all P < 0.05). LGE burden positively correlated with logNT-proBNP (r = 0.493, p = 0.006).
Conclusion: CMR reveals biventricular dysfunction and myocardial fibrosis in APS. SAPS patients demonstrated more severe involvement, characterized by elevated LGE burden and distinct subendocardial patterns. The correlation between LGE and NT-proBNP indicates that CMR has important value in risk stratification and prognostic assessment in APS.
Limitations: Not applicable
Funding for this study: Not applicable
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: C. Y. Chen, F. Y. Yang, S-J. Chen, W-J. Lee, Y-S. Huang, Y-F. Tsai, P-Y. Chiang, Y. W. GUO, S. Y. Huang; Taipei/TW
Purpose: To develop and validate an AI platform integrating Hybrid CNN–SVM classification, U-Net segmentation, and 3D scar visualization for automated myocardial scar analysis in LGE cardiac MRI, aiming to reduce manual annotation workload and improve diagnostic consistency.
Methods or Background: Ninety LGE MRI datasets were analyzed. A convolutional neural network (CNN) with four convolutional layers was trained for feature extraction, followed by a support vector machine (SVM) classifier with five-fold cross-validation to distinguish ischemic from non-ischemic cardiomyopathy. Scar segmentation was performed using a U-Net and compared with manual annotations (2SD, 3SD, 5SD, FWHM). Performance metrics included accuracy, AUC, Dice coefficient, and concordance with manual references. Grad-CAM was applied for interpretability.
Results or Findings: The Hybrid CNN–SVM achieved an accuracy of 0.851 and an AUC of 0.920. U-Net segmentation demonstrated strong concordance with manual contours (Dice = 0.801 ± 0.021) while substantially reducing annotation workload. Automated analysis markedly shortened labeling time compared with manual delineation and improved reproducibility across observers. In addition, the platform generated 3D scar visualizations, providing intuitive assessment of scar burden and transmural extent.
Conclusion: This AI-based platform demonstrates robust performance for automated classification and scar segmentation in LGE MRI. By improving workflow efficiency, reducing inter-observer variability, and integrating 3D visualization, it offers strong potential for adoption into routine radiology practice for standardized CMR interpretation.
Limitations: While the platform demonstrated robust performance, limitations include the relatively small, single-center dataset and occasional segmentation errors in blood pool regions. Future studies with larger, multi-center cohorts are warranted.
Funding for this study: This research was funded by National Science and Technology Council of Taiwan (no. MOST 114-2314-B-031-001)
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the Research Ethics Committee of the National Taiwan University Hospital (Reference number: 202509157RINE).