RPS 703 - Cardiac function: advanced imaging techniques

Purpose:

Feature-tracking (FT) derived strain imaging of the left atrium is a promising method for the detection of atrial function and (subclinical) dysfunction in cardiomyopathies (REF). FT relies on sufficient spatial and temporal resolution that allows corrects tracking of endo- and epicardial contours. Compressed sensing (CS) is a relatively new method to accelerate image acquisition by incoherent sub-sampling of k-space, transformation of sparsity, and iterative reconstructions. The aim of this study was to compare left atrial strain by FT between a conventional (Conv) and a CS-based CINE sequence.

This prospective study included 32 healthy volunteers. Imaging was performed at 3T (MAGNETOM Prisma, Siemens Healthineers, Erlangen, Germany). CINE images (Conv and CS) in typical long axis orientations were used to calculate left atrial FT strain. Both sequences were comparable in regard to temporal and spatial resolution. Post-processing was performed with a dedicated cardiac imaging software (cvi42, Circle, Calgary, Canada).

Left atrial reservoir, booster, and conduit strain differed significantly between CONV and CS (e.g. reservoir strain (Conv vs CS): 42.9 vs 27.6%, p<0.0001 ) (Table 2). Nevertheless, there was an almost linear correlation for reservoir and conduit strain, and a weaker correlation for booster strain.

FT strain parameters of the left atrium differ significantly between Conv and CS-based CINE images. These differences could be explained by different diastolic reference phases for FT and thin atrial walls for contour detection. Further research is needed to establish a potential correction factor and validate this in a larger cohort with normal and impaired atrial function.

A single-centre study.

All volunteers provided written informed consent to participate. The study was approved by the local ethics committee (2018-13520).

No funding was received for this work.

Purpose:

The mitral valve (MV) vortex ring is a swirling blood flow structure formed during diastole that is thought to store the kinetic energy of blood entering the left ventricle (LV). Cardiac magnetic resonance four-dimensional phase-contrast (4D-flow) imaging allows for the analysis of the MV vortex ring. The aim of this study was to investigate the relation of absolute and relative vortex ring kinetic energy with systolic and diastolic LV function parameters.

21 healthy subjects (male/female 7/14; age 59±7 years) underwent realtime-cine and 4D-flow imaging at 3T. LV volumetric function parameters (ejection fraction, ventricular volumes, and peak filling rates) were determined from realtime-cine series with routine software. Pre-processing of velocity data as well as the measurement of peak early-(E) and late-(A)-diastolic transmitral velocities were performed by prototype software (4DFlow, Siemens Healthcare). MV vortex rings were automatically detected and evaluated by in-house software. Dependencies of early- and late-diastolic peak absolute and relative kinetic energy (Ekin_abs,E, Ekin_rel,E, Ekin_abs,A, and Ekin_rel,A) on LV function parameters were assessed by correlation analysis.

Early- and late-diastolic MV vortex rings were detected in each subject. While LV volumetric function parameters did not correlate significantly with vortex ring kinetic energy, early- and late-diastolic peak transmitral velocities (E=74±12cm/s, A=67±13cm/s) showed strong correlations with the respective peak kinetic energies (Ekin_abs,E=1.48±0.66mJ, r=0.86; Ekin_rel,E=67.7±24.4J/m³, r=0.90; Ekin_abs,A=1.09±0.55mJ, r=0.89; Ekin_rel,A=57.8±23.3J/m³, r=0.93). Furthermore, the E/A ratio (E/A=1.14±0.23) correlated strongly with the ratio of peak kinetic energies (Ekin_abs,E/Ekin_abs,A=1.61±0.87, r=0.80; Ekin_rel,E/Ekin_rel,A=1.32±0.64, r=0.84).

MV vortex ring kinetic energy demonstrated a strong relation with peak transmitral inflow velocities, suggesting the MV vortex ring as a promising structure for studying LV diastolic function.

A small number of subjects.

Medical University Graz (24-126ex11/12).

OeNB-Anniversary-Fund No.17934.

Purpose:

To evaluate biventricular strain in the early diagnosis of AC.

We retrospectively evaluated 38 patients who underwent cardiac magnetic resonance and myocardial biopsy suggestive for AC between 2016 and 2019.

MR images were transferred to an off-line dedicated workstation and global longitudinal (GLS), circumferential (GCS), and radial (GRS) strain were calculated using a dedicated tool. After the evaluation of volumes and function, the population was divided based on MR imaging criteria for the diagnosis of AC (Model 1). Model 2 divided the population into two parts, using EF of left ventricle (LV EF)≥55% as a cut-off value. Model 3 identified patients with a positive biopsy and the absence of imaging criteria for AC plus LV EF≥55%.

All the MR parameteres were compared with a healthy population who underwent cardiac MR for other purposes. A p value < 0.05 was considered statistically significant.

In Model 1, a statistically significant difference (p<0.01) was observed for GLS, GCS, and GRS of LV between the normal population and the population with a positive biopsy. Similarly, significant results of Model 1 were observed for RV in terms of GCS and GRS.

In Model 2, a significant difference of left ventricle GLS and GRS was observed between patients with a positive biopsy and LV EF ≥ or < 55%.

In Model 3, the GRS (p: 0.02) and GCS (p: 0.003) resulted in statistically significant differences compared to the normal population.

An analysis of strain could be extremely helpful for the identification of patients that do not satisfy the criteria for diagnosis of AC.

The population was composed exclusively by patients with a positive biopsy. It would be interesting to include patients with a negative biopsy.

In process.

No funding was received for this work.

Purpose:

Cardiac magnetic resonance four-dimensional (4D) flow imaging is a potential technique for the characterisation of left ventricular (LV) diastolic function, which could substantially benefit from a reduction of scan times. The purpose of the present study was to investigate if the acceleration of whole-heart 4D flow imaging by compressed sensing (CS) could be employed to derive standard parameters of LV diastolic function.

11 adult cardiac patients with regular heart rhythm were prospectively recruited for 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany), ECG- and navigator-gated whole-heart 4D flow imaging with CS (acceleration-factor=7.6), and without CS (parallel acquisition acceleration-factor=3). Protocols of the prototype sequence were matched for spatial resolution and the number of reconstructed frames. For both data sets, early- (E) and late diastolic (A) transmitral peak velocities, systolic (S) and diastolic (D) pulmonary venous peak velocities, as well as early-diastolic peak mitral annular velocity (e’), were evaluated using a prototype software (4DFlow, Siemens Healthcare, Erlangen, Germany). The corresponding parameters of LV diastolic function E/A, S/D, and E/e’ were calculated. The relationships of the velocity-ratios derived with and without CS were investigated by means of correlation and Bland-Altman analysis.

CS acquisition reduced scan time significantly (369±92 s vs. 618±119 s, p<0.01). Velocity-ratios from CS measurements (E/A=1.35±0.31; S/D=1.11±0.33; E/e’=11.7±4.7) exhibited no significant bias compared to respective results without CS (E/A=1.44±0.31, p=0.15; S/D=1.16±0.29, p=0.13; E/e’=11.8±4.8, p=0.92), demonstrated small standard deviations of errors (0.17, 0.10 and 2.6 for E/A, S/D and E/e’, respectively), and correlated strongly (r=0.85, 0.96 and 0.85 for E/A, S/D and E/e’, respectively).

Compressed sensing allows substantial acceleration of whole-heart 4D flow imaging without a significant impact on the evaluation of standard parameters of LV diastolic function.

A small subject number.

Medical University Graz (25-044ex12/13).

OeNB Anniversary-Fund 17934.

Purpose:

Correction of feature tracking (FT) derived strain values to myocardial mass lead to an improvement in the diagnostic power in acute myocarditis and hypertensive heart disease. The purpose of this study was to evaluate this concept in a large cohort of various cardiomyopathies in comparison to healthy volunteers.

This study incorporated 444 data sets of healthy volunteers (n=127) and various cardiac diseases with and without depressed ejection fraction (n=317). CINE images in typical short- and long axis orientation were used to calculate traditional parameters of cardiac function and volumes, as well as FT derived strain values. Subsequently, strain values were corrected to myocardial mass (strain-ratio). For post-processing, a dedicated cardiovascular software tool was used (cvi42, Circle, Calgary, Canada). ROC-Analysis was used to compare the diagnostic power to differentiate between health and disease.

Global longitudinal rStrain showed the highest diagnostic power to differentiate between health and disease with an AUC of 0.904, resulting in a sensitivity and specificitiy of 82.8% and 82.6%. For differentiation of health and disease, rStrain clearly outperformed EF (ΛAUC: 0.20), EDV (ΛAUC: 0.43), myocardial mass (ΛAUC: 0.04), and “naked” longitudinal strains (ΛAUC: 0.15).

Myocardial mass corrected strains are a powerful method to discriminate between health and various acute and chronic cardiomyopathies.

A single-centre study and retrospective study design.

All patients provided written informed consent. The study was approved by the local ethics committee (2018-13520/837.196.13/837.477.14).

No funding was received for this work.

Purpose:

Pulmonary hypertension (PH) is associated with left (LV) and right ventricular (RV) functional and myocardial alterations. Cardiac magnetic resonance (CMR) cine-imaging is used not only for the assessment of volumetric function but also for the determination of myocardial strain parameters.

The aim of the study was to compare associations of feature-tracking derived myocardial strain parameters with mean pulmonary arterial pressure (mPAP) and volumetric function in patients with PH.

34 patients with PH underwent right heart catheterisation for mPAP assessment (mPAP=44±11mmHg) and CMR cine realtime free-breathing imaging at 3T (Skyra, Siemens Healthcare) within 1±3 days. LV and RV volumetric parameters were evaluated from short-axis slices (Argus, Siemens Healthcare). Global radial (GRS), circumferential (GCS), and longitudinal (GLS) LV and RV strains and strain rates (GRS_rate, GCS_rate, and GLS_rate, respectively) were assessed from short-and four-chamber series (cvi42, Circle Cardiovascular Imaging) using a two-dimensional approach. Relations between parameters were analysed by regression and correlation analysis.

Significant correlation with mPAP was found for the following strain parameters: RV-GRS (r=-0.43), systolic RV-GRS_rate (r=-0.41), RV-GCS (r=0.41), systolic LV-GRS_rate (r=0.42), and systolic LV-GCS_rate (r=-0.42).

All strain parameters correlated more strongly to respective ejection fraction (EF) than to mPAP: RV-GRS (r=0.62), systolic RV-GRS_rate (r= 0.45), and RV-GCS (r=-0.62) with RV-EF; systolic LV-GRS_rate(r=0.63) and systolic LV-GCS_rate (r=-0.54) with LV-EF. With the respective EF as the covariate, only partial correlations of LV strain parameters with mPAP remained significant (r=0.46 for systolic LV-GRS_rate, and r=-0.46 for systolic LV-GCS_rate).

In PH patients, global LV and RV strain parameters are more strongly associated with ventricular function than with the severity of PH. Consistent correlations of LV systolic strain rates with mPAP might be related to septal bowing.

A single-centre study.

NCT01725763.

OeNB-Anniversary-FundNr17934.

Purpose:

To assess the feasibility of fully-automated myocardial strain estimation in a large epidemiological study since manual segmentation and strain estimation is time-consuming and prone to interoperator-variability.

From the study of health in Pomerania (SHIP), 1,519 subjects from the general population (median age 52 years, 54.9% men) underwent steady-state-free-precession cine-CMR. A U-Net convolutional neural network (U-Net) was trained with 4-chamber, 2-chamber, and short-axis images of 991 probands from SHIP-TREND and used to predict left ventricular (LV) segmentation in end-diastole for the independent SHIP-CORE cohort. Time-series images were registered using deformable B-spline registration. In this manner, a set of indicator-points, derived from segmentation, was warped through the cardiac cycle. The movement of the indicator-points was visually assessed by two CMR experts. Global and segmental circumferential and longitudinal and radial myocardial strain curves were calculated following Green-Lagrangian formulation.

Predicted segmentation was robust for LV myocard and cavity except for very apical and basal slices, with minor errors in the presence of banding-artefacts. In the vast majority of the assessed time-series, the indicator point-set followed wall motion and biologically plausible strain curves were calculated. Minor segmental tracking errors occurred with underlying motion and chemical-shift artefacts.

Fully-automated myocardial strain estimation combining deformable B-Spline registration with U-Net is feasible in a large population-based CMR study and presents a promising approach for even larger CMR studies like the German National Cohort and UK Biobank. Future research is needed for validation and comparison to manual software.

No quantitative comparison with a reference-standard was undertaken. Comparison with commercially available software is ongoing.

All individuals gave written informed consent. The study was approved by the Ethics Committee of the University of Greifswald.

This study is supported by the German Federal State of Mecklenburg-West-Pomerania.

Purpose:

The variation between rest and peak stress end-systolic pressure-volume relation (ESPVR) is obtained during routine stress echocardiography and is an index of myocardial contractile performance.

This is the first study assessing the delta rest-stress ESPVR (DESPVR) by dipyridamole stress cardiovascular magnetic resonance (CMR).

100 consecutive patients (24 females, 63.76±10.17 years) who underwent dipyridamole stress-CMR were considered.

The ESPVR was evaluated at rest and peak-stress from raw measurement of systolic arterial pressure by sphygmomanometer and end-systolic volume by the biplane Simpson method. Wall motion and perfusion at rest and after dipyridamole were analysed. Macroscopic fibrosis was detected by the late gadolinium enhancement technique.

The mean ESPVR at rest and peak stress was, respectively, 4.84±2.47 and 5.33±3.16 mmHg/mL/m2; mean ΔESPVR was 0.48±1.45 mmHg/mL/m2.

ΔESPVR was significantly lower in males.

ΔESPVR was not correlated to baseline left ventricular end-diastolic volume index or ejection fraction.

43 patients had myocardial fibrosis and showed significantly lower ΔESPVR values.

An abnormal stress CMR was found in 25 patients. The ΔESPVR was significantly lower in patients with abnormal stress CMR.

During a mean follow-up of 56.34±30.04 months, 24 cardiovascular events occurred. At receiver-operating characteristic curve analysis, a ΔESPVR<0.02 predicted the presence of future cardiac events with a sensitivity of 0.79 and a specificity of 0.68.

The noninvasive assessment of the ΔESPVR index during a dipyridamole stress-CMR exam was feasible. The ΔESPVR was independent from baseline LV dimensions and function, so it can be used for a comparative assessment of patients with different diseases. ΔESPVR by CMR can be a useful marker for additional prognostic stratification.

n/a

Ethics committee approval obtained.

No funding was received for this work.

Purpose:

Strain parameters reflect cardiac deformation by analysing longitudinal and circumferential shortening as well as radial thickening of the myocardium. Theoretically, optimal myocardial contraction should be a result of all three deformation patterns. Therefore, we propose a new method (combined strain) to link strain values to cardiac function.

In this retrospective cross-sectional study, we evaluated 444 subjects, both healthy (n=127) and diseased (n=317), with varying degrees of ejection fraction (EF) using CMR imaging at 3T (MAGNETOM Prisma, Siemens Healthineers, Erlangen, Germany). Left ventricular volumes and EF, as well as global strain values [radial (RS), circumferential (CS), and longitudinal (LS)], were acquired for the left ventricle using dedicated software (cvi42, Circle, Calgary, Canada). Combined strain (CoS) was calculated by the following equitation: CoS=RS+Absolute(CS+LS).

Overall, absolute combined strain was found to have a higher correlation to EF than absolute longitudinal and radial strain, and a comparable correlation as absolute circumferential strain. The correlation was much stronger in patients with depressed EF than in those with normal EF. Sensitivity and specificity to predict reduced EF was higher in combined strain than in longitudinal or radial strain (AUC for CoS: 0.91 vs LS: 0.86 vs RS: 0.86).

In our large collective of healthy volunteers and magnitude of cardiomyopathies with and without depression of left ventricular function, the proposed concept of a combined strain outperformed radial and longitudinal strain in the evaluation of left ventricular function. The prognostic implications of these parameters have to be determined in prospective studies.

A single-centre, retrospective study.

No funding was received for this work.