Journal of Cardiovascular Magnetic Resonance最新文献

Background: Due to the presence of complex flow states and significant jet eccentricity in patients with congenital heart disease (CHD), accurate quantification of aortic regurgitation (AR) using standard echocardiographic or conventional cardiac magnetic resonance (CMR) imaging measures remains challenging. Four-dimensional flow (4DF) CMR permits transvalvular flow quantification under non-laminar flow states, although has not been well validated for AR quantification in CHD.

Methods: In 186 patients with moderate or complex CHD, we evaluated the agreement between different methods of AR quantification by 4DF CMR when compared to volumetry. Regurgitant flow volumes were measured (1) conventionally on time-resolved, velocity-encoded 4DF sequences at the aortic annulus, sinotubular junction (STJ), and ascending aorta (AAo), and via (2) direct regurgitant jet quantification 5mm proximal to the vena contracta.

Results: Moderate overall agreement in AR quantification was observed between study methods (ρ=0.58-0.73). Compared with conventional flow quantification at the annulus, STJ, and AAo, direct regurgitant jet measurements showed improved correlation with volumetry (ρ=0.76), especially in patients with significant aortic dilation (r=0.95-0.97). In this latter group, regurgitant flow quantification at all other aortic levels resulted in AR severity classifications that were nearly a full grade lower (mean aortic regurgitant fraction difference: 7-12% ± 10-12%; p<0.001).

Conclusions: 4DF CMR permits AR quantification in complex CHD with comparable accuracy to volumetry. Under non-laminar or complex flow states, as observed with significant aortic dilation, direct regurgitant jet measurements may be preferable to regurgitant flow quantification at all other aortic levels.

Background: In the field of cardiovascular imaging, 4D flow MRI provides non-invasive assessment of blood flow. Dual velocity encoding (dual-VENC) strategies have emerged to obtain quantitative information on both low and high blood flow velocities simultaneously. However, these strategies often encounter difficulties in coping with large velocity ranges. This work presents a dual-VENC 4D flow MRI sequence that utilizes the coprime rule to define the VENC ratio.

Methods: A dual-VENC 4D flow MRI sequence and reconstruction algorithm were developed and validated in vitro at two different field strengths, using a flow phantom generating realistic complex flow patterns. A digital twin of the phantom allowed comparison of the MRI measurements with computational fluid dynamics (CFD) simulations. Three patients with different cardiac pathologies were scanned in order to evaluate the in vivo feasibility of the proposed method.

Results: The results of the in vitro acquisitions demonstrated significant improvement in velocity-to-noise ratio (VNR) with respect to single-VENC acquisitions (110  ±  3%) and conventional dual-VENC de-aliasing approach (75  ±  3%). Furthermore, the effectiveness of aliasing correction was demonstrated even when both sets of images from the dual-VENC acquisition presented velocity aliasing artifacts. We observed a high degree of agreement between the measured and simulated velocity fields.

Conclusion: The strength of this approach lies in the fact that, unlike the conventional de-aliasing method, no data is discarded. The final image is obtained by a weighted average of the VENC_low and VENC_high datasets. Consequently, setting the value of the VENC_high to prevent aliasing is no longer necessary, and higher VNR gains are possible.

Achieving sufficient spatial and temporal resolution for dynamic applications in cardiac MRI is a challenging task due to the inherently slow nature of MR imaging. In order to accelerate scans and allow improved resolution, much research over the past three decades has been aimed at developing innovative reconstruction methods that can yield high-quality images from reduced amounts of k-space data. In this review, we describe the evolution of these reconstruction techniques, with a particular focus on those advances that have shifted the dynamic reconstruction paradigm as it relates to cardiac MRI. This review discusses and explains the fundamental ideas behind the success of modern reconstruction algorithms, including parallel imaging, spatio-temporal redundancies, compressed sensing, low-rank methods and machine learning.

Background: Fetal cardiac cine MRI is an emerging technique for evaluating the fetal heart in conditions such as congenital heart disease, but limited evidence on factors affecting image quality restricts its clinical potential. This study investigated key determinants of image quality in a multicenter cohort.

Methods: This study analyzed fetal cardiac MRI scans from April 2021 to July 2023 at three centers (University Hospital Bonn, Children's Hospital Colorado, Phoenix Children's Hospital). Cine image quality was assessed using a 5-point Likert scale (1=non-diagnostic to 5=excellent) across three criteria: contour sharpness, blood-to-structure contrast, and artifacts. Overall image quality scores were calculated by the average of all criteria. Apparent signal-to-noise (aSNR) and contrast-to-noise ratios (aCNR) were measured. Nine parameters were evaluated for their impact on image quality: gestational age, body mass index (BMI), fetal motion, patient positioning, gating signal stability, breathing technique, field strength, slice thickness, and flip angle. Comparisons were conducted using the Mann-Whitney U test.

Results: A total of 98 scans were analyzed. Higher overall image quality, aSNR, and aCNR were observed in participants with BMI <30kg/m², gestational age ≥32 weeks, low fetal motion severity, and stable gating signals (e.g., overall image quality for BMI <30kg/m² vs ≥30kg/m²: 4.4 ± 0.7 vs. 4.1 ± 0.7, p <0.001). Supine positioning resulted in better overall image quality compared to the left lateral position (4.5 ± 0.5 vs. 4.2 ± 0.8, p = 0.001). Breath-holds provided similar overall image quality but improved contour sharpness and reduced artifacts compared to free breathing (5 [4-5] vs. 4 [4-5], p = 0.042; and 4 [3-5] vs. 4 [3-5], p = 0.014, respectively). At 1.5T field strength, higher contrast and fewer artifacts were observed compared to 3T (5 [4-5] vs. 5 [4-5], p = 0.041; and 4 [4-5] vs. 4 [3-5], p = 0.010, respectively). Slice thickness showed no significant impact on image quality.

Conclusions: Various factors (e.g. BMI) influence fetal cardiac cine MRI image quality. Understanding these factors may help achieving reliable examinations and better exploit the potential of fetal cardiac MRI in clinical routine.

Background: Main pulmonary artery (mPA) dilatation has been reported in patients with isolated pulmonary valve (PV) stenosis. The aim of our study was to detect the incidence of mPA dilatation and aneurysm in patients with isolated PV stenosis and the association with PV function.

Methods: In this single-centre retrospective observational study all patients with a diagnosis of isolated PV stenosis referred to our centre were enrolled. Patients were divided into two groups (children and adults) according to age. Echocardiography, cardiac magnetic resonance imaging (MRI) and computed tomography (CT) were reviewed. MPA dilatation was defined as ≥2 Z-Score in children and ≥30mm diameters in adults, while giant (aneurysmal) mPA dilatation was defined as ≥4 Z-Score in children and ≥40mm in adults.

Results: Out of 197 patients (41.6% males, 51.8% children), 67.2% presented mPA dilatation and 16.8% giant dilatation of the mPA. The majority were adults (p<0.001). There was an underestimation of the mPA with echocardiography in 11% of patients with mPA dilatation and 50% with giant mPA. The diameter did not correlate with sex and the degree of PV stenosis. About 44% of cohort under follow-up presented a progression in mPA dilatation, however no rupture or cardiac-related death was reported.

Conclusions: There is a very high incidence of mPA dilatation in both adults and paediatric patients with isolated PV stenosis. Nevertheless, there were not recorded dissections even in patients with the largest diameters suggesting a more benign lesion. Echocardiography often underestimates the mPA measurement compared to MRI/CT which might be indicated in selected patients as a baseline 3D cross-sectional imaging evaluation, even in the presence of mild PV disease, especially when the mPA is not well visualised.

Introduction: 3D water-fat separated LGE imaging is a cardiac magnetic resonance imaging technique allowing simultaneous assessment of and discrimination between cardiac fibrosis and myocardial fatty infiltration. The aim of this study is to systematically analyze the image quality of a 3D water-fat separated LGE research sequence and identify confounders of image quality.

Methods: In total 126 patients and 12 healthy volunteers were included. Patients were included with inflammatory bowel disease (n=35), muscular dystrophy (n=38), hypertrophic cardiomyopathy (n=23) and paroxysmal atrial fibrillation (n=30). 3D water-fat separated LGE images were acquired at 1.5T (n=122) or 3T (n=16). Image quality was subjectively rated (4-point Likert scale) in six categories (overall image quality, blood-myocardium border sharpness, LGE-remote/healthy myocardium border sharpness, fat suppression, myocardial nulling, anatomical structures), additionally the contrast ratio was calculated. Cardiac function, acquisition conditions, and demographic data were investigated as potential confounders for image quality and contrast ratio.

Results: Fat suppression had the highest quality score (2.54 ± 0.72), followed by anatomical structures (2.11 ± 0.94) and myocardial nulling (2.01 ± 0.78). In total, 18 parameters showed a significant correlation with multiple image quality categories, most of which related to cardiac function, such as the cardiac index, which significantly correlated with overall image quality (Wald Chi-squared=4.35; p<0.05), LGE-remote/healthy myocardium border sharpness (Wald Chi-squared=5.03; p<0.05), and anatomical structures (Wald Chi-square=16.00; p<0.001). Left ventricular mass index to height showed significant correlation with overall image quality (Wald Chi-squared=7.57; p<0.01), blood-myocardium border sharpness (Wald Chi-squared=7.35; p<0.01), and contrast ratio (Wald Chi-squared=5.50; p<0.05). Furthermore, demographic parameters, such as body mass index (BMI), were identified as significant confounders, showing a notable correlation between BMI and the depiction of anatomical structures. (Wald Chi-square=11.14; p<0.01).

Conclusion: In this study, 3D water-fat separated LGE imaging shows satisfying image quality, especially for fat separation. However, image quality may be affected by several surrounding parameters such as patient obesity, high myocardial mass, and cardiac function.

Trial registration: 3000339.

Background: The coronary sinus reducer (CSR) is a novel percutaneous treatment for patients with refractory angina. Increasing evidence supports its clinical efficacy in patients with advanced epicardial coronary artery disease. However, its mechanism of action and its effects on myocardial perfusion remain undefined. Using quantitative stress perfusion cardiac magnetic resonance (CMR), this study assessed changes in myocardial perfusion in patients with refractory angina undergoing CSR implantation.

Methods: This single-centre retrospective observational cohort study included 16 patients. Rest and adenosine stress perfusion CMR was performed before and at median 5 months after CSR implantation. Perfusion images were acquired using a dual-sequence quantitative protocol with automated generation of myocardial blood flow (MBF; mL/min/g). In addition to visual assessment of ischaemic segments, changes in absolute MBF across myocardial segments and between myocardial layers were analysed.

Results: A high proportion of myocardial segments had visually adjudicated ischaemia at baseline (208 out of 254: 81.9%), which significantly reduced after CSR implantation (175 out of 254: 68.9%; P=0.001). There were no changes in global MBF or strain values. Changes in myocardial perfusion reserve (MPR) correlated with baseline MPR with more ischaemic segments at baseline improving to a greater extent at follow-up. Similar patterns were observed in both the left and right coronary artery territories. Changes in endocardial/epicardial MBF ratio at stress were similarly dependent on baseline values.

Conclusion: In patients with refractory angina undergoing CSR implantation, quantitative stress perfusion CMR demonstrated redistribution of myocardial perfusion across segments, from less ischaemic to more ischaemic myocardium, and across myocardial layers with greatest improvements in endocardial perfusion observed in the most ischaemic myocardium. Further studies are needed to validate the different patterns MBF redistribution that may occur after CSR implantation and correlate with clinical outcomes.

Background: Arrhythmogenic cardiomyopathy (ACM) related to Desmoplakin (DSP) mutations is a distinct condition associated with particularly severe outcomes, more frequent left ventricular (LV) involvement including fibrosis, dysfunction and inflammatory episodes. Whether DSP-ACM is associated with specific imaging features remains elusive.

Purpose: To provide a comprehensive description of cardiac magnetic resonance (CMR) findings in patients with DSP-ACM and to compare them to RV-dominant ACM with LV involvement (LV+ right-dominant-ACM).

Methods: Patients with DSP-ACM matched with patients with ACM related toa non-DSP desmosomal mutation and ≥1 feature of LV involvement underwent CMR in two institutions. Biventricular metrics and segmental wall motion abnormalities (WMA) were assessed. LV late gadolinium enhancement (LGE) was assessed both qualitatively and quantitatively after semi-automated segmentation.

Results: Overall, 70 ACM patients were analyzed; 37 with DSP-ACM and 33 in the LV+ right-dominant-ACM group. LVEF was significantly lower in the DSP-ACM group (46±12%) than in the LV+ right-dominant-ACM group (56±10%, P=0.001). Conversely, RVEF was significantly higher in the DSP-ACM group (45±11% vs. 40±12%, P=0.04) and both RV end-diastolic (100±24 vs 130±44mL/m², P=0.002) and end-systolic (56±21 vs 81±45mL/m², P=0.007) indexed volumes were significantly smaller in DSP-ACM as compared to the LV+ right-dominant-ACM group. The LV to RV end-systolic volume ratio (0.96[IQR0.70-1.27] vs. 0.59[IQR0.48-0.69]) was significantly higher in the DSP-ACM group (P<0.0001), and had a good performance in differentiating both groups (area under the ROC curve 0.86, optimal threshold 0.8). Patients in the DSP-ACM group had significantly more LV and less RV WMA than those in the LV+ right-dominant-ACM group. The amount of LGE was significantly higher in the DSP group (14±16 vs. 2±3%, P<0.0001) and present in the majority of LV segments, particularly in the lateral and inferior walls, as compared to LV+ right-dominant-ACM patients. Transmural LGE and the presence of a ring-like pattern corresponding to circumferential subepicardial LGE involving ≥3contiguous LV basal segments were highly suggestive of DSP-ACM.

Conclusions: The presence of LV to RV end-systolic volume ratio>0.8, global LGE>5%, transmural and/or a ring-like LGE pattern are highly suggestive of DSP-ACM and should prompt careful diagnostic assessment considering the severe associated outcomes.

Background: Three-dimensional (3D) tagged magnetic resonance (MR) imaging enables in vivo quantification of cardiac motion. While deep learning methods have been developed to analyze these images, they have been restricted to two-dimensional datasets. We present a deep learning approach specifically designed for displacement analysis of 3D cardiac tagged MR images.

Methods: We developed two neural networks to predict left-ventricular motion throughout the cardiac cycle. Networks were trained using synthetic 3D tagged MR images, generated by combining a biophysical left-ventricular model with an analytical MR signal model. Network performance was initially validated on synthetic data, including assessment of signal-to-noise ratio (SNR) sensitivity. The networks were then retrospectively evaluated on an in vivo external validation human datasets and a in vivo porcine study.

Results: For the external validation dataset, predicted displacements deviated from manual tracking by median(IQR) values of 0.72(1.51), 0.81(1.64) and 1.12(4.17) mm in x, y and z directions, respectively. In the porcine dataset, strain measurements showed median(IQR) differences from manual annotations of 0.01(0.04), 0.01(0.06) and  - 0.01(0.18) for circumferential, longitudinal, and radial components. These strain values are within physiological ranges and demonstrate superior performance of the network approach compared to existing 3D tagged image analysis methods.

Conclusions: The method enables rapid analysis times of approximately 10 seconds per cardiac phase, making it suitable for large cohort investigations.