Journal of Cardiovascular Magnetic Resonance最新文献

Background: Sotatercept is a novel drug therapy for pulmonary arterial hypertension (PAH) that significantly reduces clinical worsening and improves exercise capacity, functional class and pulmonary haemodynamics. This study aims to use cardiovascular magnetic resonance (CMR) to monitor the cardiac response to sotatercept in patients with PAH.

Methods: Patients with PAH at intermediate-high or high mortality risk and already on background maximal triple PAH therapies, including parenteral prostacyclin analogue for at least 3 months, were offered enrolment to a patient access program for sotatercept at the Royal Free Hospital National Pulmonary Hypertension Service. Baseline and follow-up CMR studies were performed at a median interval of 24 (interquartile range 6) weeks.

Results: 18 out of 23 patients enrolled to the sotatercept access program underwent baseline and follow-up CMR studies (3 patients paused or discontinued sotatercept prior to 12 weeks; 2 patients were unable to undergo CMR). All 18 patients were stratified as being of intermediate-high mortality risk. Significant improvements were observed in right ventricular (RV) size (RV end-diastolic volume -34±30mL, p = 0.0023; RV end-systolic volume -27±29mL, p = 0.0023), RV mass (Z = -2.63, p = 0.016), RV function (RV ejection fraction 5±8%, p = 0.034) and right atrial size (-6±4cm², p = 0.0023). End-systolic interventricular septal curvature also significantly improved at follow-up (Z = 2.07, p = 0.046), suggesting improvement in RV afterload. Five (28%) patients also had a new-onset (1) or larger (4) pericardial effusion without haemodynamic compromise at follow-up, despite improvements in clinical, biochemical and CMR metrics of PAH. The median increase in pericardial effusion volume was 69% (full range 33%-203%, Z = 2.0, p = 0.043).

Conclusion: CMR tracks improvements in right heart chamber size, mass and function along with metrics of RV afterload in patients with PAH receiving sotatercept. The improvements in RV size and function met or exceeded the clinically relevant minimally important differences for these CMR-derived metrics in patients with PAH. Routine interval surveillance with CMR in patients receiving sotatercept will also enable surveillance for the off-target finding of new-onset or worsening pericardial effusions.

Background: Ferumoxytol, an intravenous iron supplement that can be used off-label as a contrast agent in cardiovascular magnetic resonance (CMR), has been proposed to enhance the amplitude and dynamic range of myocardial T₁ vasoreactivity. Its combined use with vasodilators during stress testing may pose additional risks for hypotension. We aim to evaluate the hemodynamic profile and myocardial T₁ response of ferumoxytol and regadenoson for ferumoxytol-enhanced stress CMR (FE-CMR).

Methods: Seventy-two participants (ischemic heart disease [IHD]: N=44, healthy: N=28) underwent FE-CMR under continuous hemodynamic monitoring. All IHD patients underwent clinically-indicated regadenoson stress cardiac positron emission tomography (PET) between 6 days and 31 weeks before FE-CMR. Ferumoxytol was administered in cumulative doses of 3.0 or 4.0mg/kg, followed by regadenoson. Hemodynamic responses were compared with gadobutrol alone and with regadenoson alone. Post-contrast T₁ maps were segmented to quantify FE-T₁ vasoreactivity in healthy, remote, ischemic, and infarcted tissues. Linear mixed-effects models evaluated the effects of ferumoxytol and regadenoson on mean arterial pressure (MAP) and heart rate (HR), and compared FE-T₁ vasoreactivity across tissue types.

Results: No severe or life-threatening adverse events occurred. Five patients had mild symptoms. One study was terminated early due to moderate hypotension. Post-ferumoxytol MAP increased slightly (0.1-3.4%) at rest, without statistical significance. MAP decreased modestly post-regadenoson (-5.3 to -2.8%; all P<0.05). HR remained stable after ferumoxytol administration at rest and increased transiently after regadenoson, consistent with its pharmacologic effect. Unlike the modest MAP increase with ferumoxytol at rest, gadobutrol produced minor MAP decreases (-2.1 to -0.8%), with no significant between-agent differences. FE-T₁ vasoreactivity demonstrated a consistently blunted response across tissue types (ΔFE-T₁ in remote: -7.6±0.2%, ischemic: -4.6±0.3%, scar: -1.6±0.4%, healthy: -9.5±0.3%; all P<0.05). Receiver operating characteristic analysis showed strong remote versus scar discrimination (AUC 0.86) and modest remote vs. ischemia discrimination (0.69).

Conclusions: Regadenoson stress FE-CMR is well-tolerated. Hemodynamic changes from combined ferumoxytol and regadenoson were small in magnitude and generally not clinically significant. Relative to healthy myocardium, blunted responses were observed in remote, ischemic, and infarcted tissues. Despite early promise, FE-T₁ vasoreactivity requires further dedicated study to fully evaluate its diagnostic performance as a gadolinium-free imaging biomarker in IHD.

Background: Following myocardial infarction, late gadolinium-enhancement (LGE) assessed by cardiovascular magnetic resonance (CMR) provides a reliable metric for risk stratification and therapeutic planning. However, conventional segmentation methods are time-consuming and labor-intensive, with high interobserver variability and inconsistent performance in routine clinical practice. This study sought to develop an interactive deep learning system for scar segmentation and quantification.

Methods: The framework was developed and evaluated using LGE-CMR images from 348 patients with chronic myocardial infarction (244 training, 51 validation, 53 test). The model incorporates prompt-guided segmentation and leverages a vision foundation model adapted for medical imaging, integrated into a clinician-facing interface for real-time interaction, and automated quantification. Training used a composite loss function combining Dice overlap, voxel-wise cross-entropy, and Kullback-Leibler divergence against soft labels to address annotation uncertainty. Performance was evaluated on a held-out test set using expert manual annotations as the reference standard, with assessment of segmentation accuracy, repeatability, and agreement with the conventional full-width at half-maximum method (FWHM).

Results: The framework achieved expert-level segmentation performance on the test set (Dice similarity coefficient=0.74±0.10; Hausdorff distance=5.87±6.79mm) with median scar mass error of 1.28g (IQR 0.74-2.34), corresponding to 1.4% (IQR 0.81-2.47) of left ventricular mass. Repeatability analysis (n=41) demonstrated excellent agreement, with both inter- and intra-observer concordance correlation coefficients of 0.999 (compared with 0.737 and 0.952, respectively, for the conventional FWHM). Segmentation time was substantially reduced when using the interactive tool compared with the conventional workflow, averaging 65 ± 34seconds per patient. Performance and repeatability remained high across the test set with differing levels of image quality.

Conclusions: The proposed framework for scar segmentation with a human-in-the-loop design enables fast, accurate, and highly reproducible myocardial scar quantification from LGE-CMR. This may provide more consistent performance in routine clinical workflows.

Background: Left ventricular pressure-volume (PV) loop analysis via cardiovascular magnetic resonance (CMR) offers noninvasive assessment of cardiac thermodynamic efficiency. We aimed to evaluate the clinical relevance and prognostic value of noninvasive PV-loop parameters in patients with cardiac light-chain (AL) amyloidosis.

Methods: This prospective, single-center study enrolled patients with cardiac AL amyloidosis (AL-CA) who underwent CMR between November 2011 and September 2023. PV-loop parameters were derived from CMR cine images and brachial blood pressure. The primary endpoint was all-cause mortality. Cox regression analysis assessed associations between PV-loop parameters and outcomes, with incremental prognostic value evaluated using C statistics and likelihood ratio tests.

Results: The single-center, prospective study included 267 consecutive AL-CA patients (mean age 58.8 years±9.8 [SD]; 168 [62.9%] males) and 30 healthy controls with similar age and sex (mean age, 59.9±year 8.2 [SD]; 15 [50%] males). Work efficiency (WE) showed moderate to strong correlations with cardiac function, volumes, decease activity and amyloid burden (all P < 0.05). During a median follow-up of 42 months (IQR: 35-49), 185 patients (69.3%) died. Univariable Cox analysis showed AL-CA patients with WE < 72.2% were at higher mortality risk (hazard ratio 2.45, 95% CI: 1.68-3.57; P < 0.001). After multivariable adjustment, WE < 72.2% remained independent prognostic factor. The integration of WE with Mayo 2004 stage enhanced prognostic discrimination and calibration (C-statistic 0.65, χ² 36.28) relative to Mayo 2004 stage alone (C-statistic 0.58, χ² 25.79), and performed comparably to the combination of extracellular volume fraction (ECV) with Mayo 2004 stage (C-statistic 0.66, χ² 44.49).

Conclusion: Myocardial work impairment can be detected early, even with preserved LVEF. WE derived from noninvasive and non-enhanced PV-loop analysis via CMR, not only reflects disease severity as quantified by ECV, but also serves as an alternative marker to ECV in risk stratification.

Background: Myocardial T2 mapping enables non-invasive assessment of inflammation and oedema. However, in patients with implantable cardiac devices, such as pacemakers or defibrillators (ICDs), off-resonance effects often cause severe image artefacts and inaccurate T2 values.

Purpose: The aim of this study was to develop and evaluate a wideband T2-prepared gradient-echo (GRE) myocardial T2 mapping sequence combined with an advanced patch-based denoising approach, designed to reduce artefacts and improve image quality in device-implanted patients at 1.5T.

Methods: A T2 preparation with wideband adiabatic refocusing pulses (5.0kHz bandwidth) was integrated into a breath-held 2D GRE T2 mapping sequence (TE = 0/27/55 ms). Patch-based denoising was applied after image reconstruction. The sequence was tested in a phantom, eight healthy volunteers with and without ICDs placed on their chests, thirteen patients without devices, seven patients with ICDs or pacemakers, and one sheep scanned before and after induced myocardial infarction with and without external ICD. The proposed sequence was compared against reference conventional GRE and balanced steady-state free-precession (bSSFP) T2 mapping. Patch-based denoising was optimized in patients without devices and impact on T2 precision and accuracy was assessed. Phantom studies included Bland-Altman and correlation analyses between the sequences. In-vivo performance was assessed through global and segmental T2 quantification, coefficient of variation (COV), artefact scoring, and oedema detection. ANOVA with Bonferroni correction and pairwise testing were used for statistical comparisons.

Results: In subjects without devices, wideband and conventional GRE T2 mapping yielded comparable T2 values (P=0.60). With ICDs, conventional GRE T2 mapping underestimated global T2 by 16% (P<0.001) and increased segmental COV up to 30%. In contrast, wideband GRE T2 mapping provided accurate T2 values (P=0.56) and preserved oedema detection, showing relative T2 elevations of 44% comparable to bSSFP. Patch-based denoising significantly improved precision (P=0.006) without biasing mean values (P=0.999). Results were consistent across phantom, volunteer, patient, and animal experiments, including animal ex-vivo histology confirmation.

Conclusion: Wideband GRE T2 mapping substantially reduced device-related artefacts, provided accurate T2 values, and allowed oedema detection, offering a clinically feasible solution for patients with cardiac implants in this initial study.

Background: Abdominal aortic aneurysm (AAA) progression is driven by extracellular matrix (ECM) proteolysis and vascular smooth muscle cell loss, processes not captured by diameter-based surveillance. A disintegrin and metalloproteinase with thrombospondin motifs 4 (ADAMTS4) is upregulated during early ECM remodeling, making it a compelling target for molecular imaging.

Methods: Eighteen female German Landrace swine were enrolled. Eight animals underwent AAA induction and four served as controls; six animals were excluded due to procedural complications. Molecular 3T MRI was performed at two and four weeks after induction following intravenous administration of an ADAMTS4-specific gadolinium probe (~0.03mmol/kg). Contrast-to-noise ratio (CNR) was calculated pre- and post-contrast. Aortic diameter was monitored by serial ultrasound. Ex vivo analysis included immunofluorescence, western blotting, and laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS). Spearman correlation was used to evaluate relationships between MRI signal and molecular markers of ECM remodeling.

Results: ΔCNR increased at two weeks (3.18 ± 0.56) and four weeks (4.35 ± 0.84) relative to controls (-0.14 ± 0.57; p = 0.017 and p = 0.046). ADAMTS4 immunofluorescence increased to 43.91 ± 9.48% at two weeks and 46.97 ± 4.50% at four weeks (control: 0.39 ± 0.03%). ΔCNR strongly correlated with ADAMTS4 (r_s = 0.92, p < 0.001) and Galectin-3 (r_s = 0.87, p < 0.001), and inversely with alpha-smooth muscle actin (r_s = -0.90, p < 0.001), indicating that MRI signal reflects active proteolytic remodeling and smooth muscle cell depletion. LA-ICP-MS confirmed focal probe accumulation in regions of high ADAMTS4 expression.

Conclusion: ADAMTS4-targeted molecular MRI enables early, non-invasive detection of ongoing ECM remodeling in AAA and provides activity-based disease characterization beyond diameter-based assessment.

Background: Flow quantification by cardiac magnetic resonance (CMR) is essential for diagnosis-making and follow-up in patients with congenital heart disease (CHD). Standard 2D phase contrast (2D flow) sequences require, however, precise image plane prescription resulting in long exam times. Free-breathing 3D whole-heart flow sequences (4D flow) offer faster acquisition but require respiratory navigation rendering exam duration unpredictable. Free-running radial, fully self-gated, respiratory and cardiac motion-resolved 5D flow promises ease-of-use and stable scan duration. This study compared net flow volumes (NFV) between 2D, 4D and 5D flow and maximum velocity (Vmax) to transthoracic echocardiography (TTE).

Methods: CHD patients were scanned on a 1.5T scanner. 2D flow was performed in the ascending (AAo), descending aorta, main (MPA), right, and left pulmonary arteries, superior vena cava, and pulmonary veins. By a prototype whole-heart free-running 3D radial phase contrast CMR sequence end-expiratory 5D flow images were obtained. Finally, 4D flow data were collected. 3D vessel segmentation was performed for 4D and 5D flow datasets. NFV were compared between 2D, 4D and 5D flow and Vmax to transthoracic echocardiography (TTE).

Results: 59 CHD patients (median age 28, IQR 17.5, 42% women) with right- and/or left-sided CHD were included. Scan duration did not differ significantly between methods. 4D and 5D flow underestimated NFV relative to 2D flow by 4-17ml per beat which was more prominent for 4D flow and smaller vessels. Internal agreement was best for 4D flow, followed by 5D flow, while 2D flow showed the largest variability: MPA vs AAo (bias ml, 95% LOA): 2D flow: -4.84 (-34.18, 24.5), 4D flow -0.84 (-17.85, 16.18), 5D flow -6.4 (-28.53, 15.73). 2D and 4D but not 5D flow underestimated Vmax compared to TTE: AAo: TTE 120±46; 2D 108±36; 4D 107±36; 5D 115±40cm/s; p<0.03 TTE vs 2D and 4D flow; p=0.253 TTE vs 5D flow).

Conclusion: 4D and 5D flow provided measurements comparable to 2D flow. Despite systematic underestimation of NFV, internal agreement of whole heart methods was superior while scan time did not increase significantly. 5D flow yielded the most accurate Vmax compared to TTE, supporting its potential for comprehensive non-invasive hemodynamic evaluation.

Background: Accurate assessment of cardiac flow and volumes is essential in children with heart disease. Cardiovascular magnetic resonance (CMR) enables comprehensive quantitative evaluation but is technically demanding and often requires sedation or contrast administration in neonates. Free-breathing four-dimensional (4D) flow CMR simplifies image acquisition and enables imaging without general anesthesia or controlled ventilation. However, most prior neonatal 4D flow studies have relied on sedation and/or contrast, limiting evaluation in healthy populations. Establishing normative data and assessing feasibility and reproducibility in unsedated neonates are therefore important for clinical translation. Accordingly, this study aimed to evaluate the feasibility, accuracy, and interobserver reproducibility of free-breathing, non-contrast 4D flow CMR in healthy unsedated neonates and to characterize postnatal development of cardiac flow.

Methods: One hundred healthy term neonates (mean age: 18.6 ± 7.6 days; 53% male) with normal echocardiograms underwent free-breathing feed-and-wrap axial 4D flow cardiovascular magnetic resonance imaging (ViosWorks) with ECG gating, followed by a transverse cine Steady-State Free Precession (FIESTA) stack. Cardiac output (CO), cardiac index (CI), pulmonary blood flow (Qp), systemic blood flow (Qs), and Qp/Qs were quantified. Interobserver reproducibility was assessed by three independent observers.

Results: Free-breathing 4D flow acquisition without sedation was feasible in 70% of neonates, with a mean scan time of 9.3 ± 1.9minutes. Mean CO was 196 ± 39ml·min⁻¹·kg⁻¹. CI increased by 80% during the first 31 days of life (1.95 vs. 3.50L·min⁻¹·m^-², p<0.001), driven primarily by a 72% increase in stroke volume. Pulmonary blood flow was on average 17% higher than systemic flow (Qp/Qs = 1.17 ± 0.17). Inter-observer reproducibility was excellent, with intraclass correlation coefficients of 0.93 for the ascending aorta and 0.91 for the main pulmonary artery. 47 neonates remained calm after the 4D flow scan, allowing a transverse cine stack sequence to be acquired as well. Cardiac output derived from 4D flow showed good agreement with volumetric measurements with no significant difference (p=0.1).

Conclusion: Free-breathing, non-contrast 4D flow CMR was feasible in the majority of neonates, with rapid acquisition, excellent interobserver reliability, and good agreement with volumetric measurements. Neonatal CO increased markedly after birth, driven primarily by rising stroke volume, and mean Qp/Qs was modestly elevated.

Background: Right atrial (RA) phasic function is impaired in individuals with pulmonary arterial hypertension (PAH), evidence for the clinical significance of the RA strain in risk stratification for PAH patients is limited.

Methods: Participants with PAH from June 2013 to December 2022 were prospectively recruited. C-index, 1-year mortality, and annual event rates were used to evaluate prognostic performance. Risk groups were defined as low (<5%), intermediate (5-20%), and high (>20%) 1-year mortality.

Results: A total of 348 PAH patients were included (mean age: 40.0 ± 14.0 years; 93 males), with a median follow-up of 41.5 months (interquartile range: 24.9-61.9 months). RA reservoir strain independently predicted all-cause mortality (HR = 0.950, 95% CI: 0.922-0.979; P < 0.001). Based on cutoff values of 16.1% and 36.8%, RA reservoir strain stratified patients into high-, intermediate-, and low-risk groups with 1-year mortality rates of 24.0%, 5.5%, and 0.0%, demonstrating the highest discriminative ability among CMR-derived metrics (C-index: 0.79, 95% CI: 0.71-0.87). Incorporating RA reservoir strain significantly improved the performance of the REVEAL Lite 2 model (P = 0.03), with comparable prognostic value to that of the combined CMR parameters (both P > 0.05). Similar findings were observed in PAH patients without shunts.

Conclusion: RA reservoir strain outperformed RV functional parameters in stratifying PAH patients into low-, intermediate-, and high-risk strata. RA reservoir strain has the potential to be part of the multiparametric evaluation of patients with PAH.

Trial registration: This study was registered in the Chinese clinical trial registry (ChiCTR1800019314 and ChiCTR1900025518). URL: https://www.chictr.org.cn/index.html.

Background: Late gadolinium enhancement (LGE) cardiovascular magnetic resonance (CMR) is the reference standard for assessing myocardial scar and microvascular obstruction (MVO), strong predictors of post-acute myocardial infarction (AMI) outcomes. However, manual segmentation is time-consuming and subject to inter-observer variability, limiting clinical scalability. This study develops and validates LGE-CMRnet, an end-to-end deep learning pipeline for automated scar and MVO segmentation on LGE CMR, and evaluates its prognostic value in AMI patients.

Methods: A total of 3,874 LGE images from 567 AMI patients (409 for training/internal stress-test cohort; 158 for external testing) were analyzed. LGE-CMRnet integrates YOLOv8 for heart localization and nnU-Net for simultaneous segmentation of myocardium, scar, and MVO. Performance was evaluated using Dice similarity coefficient (DSC), correlation, and Bland-Altman analysis against expert annotations. Prognostic value was assessed using Cox regression for major adverse cardiac events (MACE) over a median follow-up of 24.4 months.

Results: LGE-CMRnet achieved rapid processing (0.05seconds per image) and high segmentation accuracy. In the external validation cohort, the model achieved mean DSC of 0.83±0.11 for scar and 0.88±0.11 for MVO at the patient level, with strong volumetric correlations to expert reference segmentations (scar: r=0.90; MVO: r=0.98, both P<0.0001). Bland-Altman analysis showed minimal bias in volumetric measurements (scar: 2.5±8.9 cm³; MVO: -0.20±0.89 cm³). Among the 158 patients in the external validation cohort (age 57±10 years, 80% male), 35 (22.2%) experienced MACE. LGE-CMRnet-derived %MVO (hazard ratio [HR], 1.06; 95% confidence interval [CI]: 1.02 to 1.09; P=0.003) and %Scar (HR, 1.05; 95% CI: 1.02 to 1.08; P=0.001) were independent predictors of MACE after adjustment for established risk factors. Furthermore, LGE-CMRnet-derived metrics demonstrated non-inferior discrimination for MACE prediction compared with expert analysis. The differences in C-index were 0.02 for %MVO and 0.01 for %Scar, with the lower bounds of the 95% CIs remaining above the pre-specified non-inferiority margin.

Conclusions: LGE-CMRnet enables fast and accurate scar and MVO quantification, with prognostic performance comparable to expert analysis, supporting its potential for automated clinical risk stratification after AMI.