Deep Learning-Based Analysis of Aortic Morphology From Three-Dimensional MRI.

BACKGROUND: Quantification of aortic morphology plays an important role in the evaluation and follow-up assessment of patients with aortic diseases, but often requires labor-intensive and operator-dependent measurements. Automatic solutions would help enhance their quality and reproducibility. PURPOSE: To design a deep learning (DL)-based automated approach for aortic landmarks and lumen detection derived from three-dimensional (3D) MRI. STUDY TYPE: Retrospective. POPULATION: Three hundred ninety-one individuals (female: 47%, age = 51.9 ± 18.4) from three sites, including healthy subjects and patients (hypertension, aortic dilation, Turner syndrome), randomly divided into training/validation/test datasets (N = 236/77/78). Twenty-five subjects were randomly selected and analyzed by three operators with different levels of expertise. FIELD STRENGTH/SEQUENCE: 1.5-T and 3-T, 3D spoiled gradient-recalled or steady-state free precession sequences. ASSESSMENT: Reinforcement learning and a two-stage network trained using reference landmarks and segmentation from an existing semi-automatic software were used for aortic landmark detection and segmentation from sinotubular junction to coeliac trunk. Aortic segments were defined using the detected landmarks while the aortic centerline was extracted from the segmentation and morphological indices (length, aortic diameter, and volume) were computed for both the reference and the proposed segmentations. STATISTICAL TESTS: Segmentation: Dice similarity coefficient (DSC), Hausdorff distance (HD), average symmetrical surface distance (ASSD); landmark detection: Euclidian distance (ED); model robustness: Spearman correlation, Bland-Altman analysis, Kruskal-Wallis test for comparisons between reference and DL-derived aortic indices; inter-observer study: Williams index (WI). A WI 95% confidence interval (CI) lower bound >1 indicates that the method is within the inter-observer variability. A P-value <0.05 was considered statistically significant. RESULTS: DSC was 0.90 ± 0.05, HD was 12.11 ± 7.79 mm, and ASSD was 1.07 ± 0.63 mm. ED was 5.0 ± 6.1 mm. A good agreement was found between all DL-derived and reference aortic indices (r >0.95, mean bias <7%). Our segmentation and landmark detection performances were within the inter-observer variability except the sinotubular junction landmark (CI = 0.96;1.04). DATA CONCLUSION: A DL-based aortic segmentation and anatomical landmark detection approach was developed and applied to 3D MRI data for achieve aortic morphology evaluation. EVIDENCE LEVEL: 3 TECHNICAL EFFICACY: Stage 2.

Estimate of the hydraulic force in the aging heart: a cardiovascular magnetic resonance imaging study.

BACKGROUND: Coupling between left ventricle (LV) and left atrium (LA) plays a central role in the process of cardiac remodeling during aging and development of cardiac disease. The hydraulic force (HyF) is related to variation in size between LV and LA. The objectives of this study were to: (1) derive an estimate of left atrioventricular HyF using cine- Magnetic Resonance Imaging (MRI) in healthy subjects with a wide age range, and (2) study its relationship with age and conventional diastolic function parameters, as estimated by reference echocardiography. METHODS: We studied 119 healthy volunteers (mean age 44 ± 17 years, 58 women) who underwent Doppler echocardiography and MRI on the same day. Conventional transmitral flow early (E) and late (A) LV filling peak velocities as well as mitral annulus diastolic longitudinal peak velocity (E') were derived from echocardiography. MRI cine SSFP images in longitudinal two and four chamber views were acquired, and analyzed using feature tracking (FT) software. In addition to conventional LV and LA strain measurements, FT-derived LV and LA contours were further used to calculate chamber cross-sectional areas. HyF was approximated as the difference between the LV and LA maximal cross-sectional areas in the diastasis phase corresponding to the lowest LV-LA pressure gradient. Univariate and multivariate analyses while adjusting for appropriate variables were used to study the associations between HyF and age as well as diastolic function and strain indices. RESULTS: HyF decreased significantly with age (R²=0.34, p < 0.0001). In addition, HyF was significantly associated with conventional indices of diastolic function and LA strain: E/A: R²=0.24, p < 0.0001; E': R²=0.24, p < 0.0001; E/E': R²=0.12, p = 0.0004; LA conduit longitudinal strain: R²=0.27, p < 0.0001. In multivariate analysis, associations with E/A (R2 = 0.39, p = 0.03) and LA conduit strain (R2 = 0.37, p = 0.02) remained significant after adjustment for age, sex, and body mass index. CONCLUSIONS: HyF, estimated using FT contours, which are primarily used to quantify LV/LA strain on standard cardiac cine MRI, varied significantly with age in association with subclinical changes in ventricular filling. Its usefulness in cohorts of patients with left heart disease to detect LV-LA uncoupling remains to be evaluated.

New atrio-ventricular indices derived from conventional cine MRI correlate with functional capacity in patients with asymptomatic primary mitral regurgitation.

Mitral regurgitation (MR) is associated with morphological and functional alterations of left atrium (LA) and ventricle (LV), possibly inducing LA-LV misalignment. We aimed to: (1) characterize angulation between LA and mitral annulus from conventional cine MRI data and feature-tracking (FT) contours, (2) assess their associations with functional capacity in MR patients, as assessed by oxygen consumption (peak-VO2) and minute ventilation to carbon dioxide production (VE/VCO2) slope, in comparison with MRI LA/LV strain indices. Thirty-two asymptomatic primary MR patients (56 [40; 66] years, 12 women) underwent cardiac MRI resulting in LA/LV conventional FT-derived strain indices. Then, end-diastolic angles were derived from FT LA contours: (1) α, centered on the LA centre of mass and defined by mitral valve extremities, (2) γ, centered on the mitral ring anterior/lateral side, and defined by LA centre and the other extremity of the mitral ring. Cardiopulmonary exercise testing with simultaneous echocardiography were also performed; peak-VO2 and VE/VCO2 slope were measured. While peak-VO2 and VE/VCO2 slope were not correlated to LA/LV strains, they were significantly associated with angles (α: r = 0.50, p = 0.003 and r = - 0.52, p = 0.003; γ: r = - 0.53, p = 0.002 and r = 0.52, p = 0.003; respectively), independently of age and gender (R2 ≥ 0.29, p ≤ 0.03). In primary MR, the new LA/mitral annulus angles, computed directly from standard-of-care MRI, are better correlated to exercise tolerance than conventional LA/LV strain.

Feasibility of a New Regional Myocardial Strain Parameter for the Detection of Wall Motion Abnormalities in Arrhythmogenic Right Ventricular Cardiomyopathy.

Purpose: To evaluate a cardiac MRI feature tracking (FT)-derived parameter that combines right ventricular (RV) longitudinal and radial motions in detecting arrhythmogenic right ventricular cardiomyopathy (ARVC). Materials and Methods: Patients with ARVC (n = 47; median age, 46 [IQR, 30-52] years; 31 men) were compared with controls (n = 39; median age, 46 [IQR, 33-53] years; 23 men) and separated into two groups based on fulfillment of major structural 2020 International criteria. Cine data from 1.5-T cardiac MRI examinations were analyzed using FT, resulting in conventional strain parameters and a novel composite index named the longitudinal-to-radial strain loop (LRSL). Receiver operating characteristic (ROC) analysis was used to assess diagnostic performance of RV parameters. Results: Volumetric parameters differed significantly between patients in the major structural criteria group and controls but not between patients in the no major structural criteria group and controls. Patients in the major structural criteria group had significantly lower magnitudes of all FT parameters than controls, including RV basal longitudinal strain, radial motion fraction, circumferential strain, and LRSL (-15.6% ± 6.4 vs -26.7% ± 13.9; -9.6% ± 4.89 vs -13.8% ± 4.7; -6.9% ± 4.6 vs -10.1% ± 3.8; and 217.0 ± 128.9 versus 618.6 ± 356.3, respectively). Only LRSL differed between patients in the no major structural criteria group and controls (359.5 ± 195.8 vs 618.6 ± 356.3; P < .0001). Parameters with the highest area under the ROC curve values for discriminating patients in the no major structural criteria group from controls were LRSL, RV ejection fraction, and RV basal longitudinal strain (0.75, 0.70, and 0.61, respectively). Conclusion: A new parameter combining RV longitudinal and radial motions showed good diagnostic performance in ARVC, even in patients without major structural abnormalities.Keywords: Arrhythmogenic Right Ventricular Dysplasia, Strain, Wall Motion Abnormalities, Right Ventricle, MRI, Inherited Cardiomyopathy Supplemental material is available for this article. © RSNA, 2023.

Myocardial fibrosis assessed by magnetic resonance imaging in asymptomatic heterozygous familial hypercholesterolemia: the cholcoeur study.

BACKGROUND: Familial Hypercholesterolemia (FH) is an underdiagnosed condition with an increased cardiovascular risk. It is unknown whether lipid accumulation plays a role in structural myocardial changes. Cardiovascular Magnetic Resonance (CMR) is the reference technique for the morpho-functional evaluation of heart chambers through cine sequences and for myocardial tissue characterization through late gadolinium enhancement (LGE) and T1 mapping images. We aimed to assess the prevalence of myocardial fibrosis in FH patients. METHODS: Seventy-two asymptomatic subjects with genetically confirmed FH (mean age 49·24, range 40 to 60 years) were prospectively recruited along with 31 controls without dyslipidaemia matched for age, sex, BMI, and other cardiovascular risk factors. All underwent CMR including cine, LGE, pre- and post-contrast T1 mapping. Extracellular volume (ECV) and enhancement rate of the myocardium (ERM = difference between pre- and post-contrast myocardial T1, normalized by pre-contrast myocardial T1) were calculated. FINDINGS: Five FH patients and none of the controls had intramyocardial LGE (p= 0·188). While no changes in Native T1 and ECV were found, post-contrast T1 was significantly lower (430·6 ± 55ms vs. 476·1 ± 43ms, p<0·001) and ERM was higher (57·44± 5·99 % vs 53·04±4·88, p=0·005) in HeFH patients compared to controls. Moreover, low post-contrast T1 was independently associated with the presence of xanthoma (HR 5·221 [1·04-26·28], p= 0·045). A composite score combining the presence of LGE, high native T1 and high ERM (defined as ≥ mean ± 1·5 SD) was found in 20·8% of the HeFH patients vs. 0% in controls (p<0·000, after adjustment for main confounders). INTERPRETATION: CMR revealed early changes in myocardial tissue characteristics in HeFH patients, that should foster further work to better understand and prevent the underlying pathophysiological processes.