Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Scan–rescan reproducibility of diastolic left ventricular kinetic energy, viscous energy loss and vorticity assessment using 4D flow MRI: analysis in healthy subjects

Scan–rescan reproducibility of diastolic left ventricular kinetic energy, viscous energy loss and... The aim of the current study was to assess the scan–rescan reproducibility of left ventricular (LV) kinetic energy (KE), viscous energy loss (EL) and vorticity during diastole from four-dimensional flow magnetic resonance imaging (4D flow MRI) in healthy subjects. Twelve volunteers (age 27 ± 3 years) underwent whole-heart 4D flow MRI twice in one session. In-scan consistency was evaluated by correlation between KE and EL. EL was computed to measure the amount of EL index relative to KE over diastole. Scan–rescan analysis was performed to test reproducibility of volumetric measurements of KE, EL, EL and vorticity in the LV over early (E) and late (A) diastolic filling. In-scan consistency between KE and EL was index strong-excellent (E-filling scan1: r = 0.92, P < 0.001; scan2: ρ = 0.96, P < 0.001 and A-filling scan1: ρ = 0.87, P < 0.001; scan2: r = 0.99, P < 0.001). For the majority of subjects (10 out of 12), KE and EL measures showed good to strong reproducibility. However, with a wide range of agreement [intraclass correlation (ICC): 0.64–0.95] and coefficients of variation (CV) ≤ 25%. EL showed strong reproducibility for all 12 subjects with a strong ICC (0.94, P < 0.001) and a CV of 9%. Scan–rescan index reproducibility of volumetric vorticity showed good–excellent ICCs (0.83–0.95) with CVs ≤ 11%. In conclusion, the current study shows strong–excellent in-scan consistency and overall good agreement between scans for 4D flow MRI assessment of left ventricular kinetic energy, energy loss and vorticity over diastole. However, substantial differences between the scans were also found in some parameters in two out of twelve subjects. Strong reproducibility was found in the dimensionless EL , which measures the amount of viscous energy loss relative to the average kinetic energy over diastole. index Keywords 4D flow MRI · Magnetic resonance imaging · Kinetic energy · Viscous energy loss · Vorticity Introduction imaging and analysis techniques are needed to evaluate volu- metric changes in such complex hemodynamic parameters. Congenital and acquired heart diseases affect the efficacy Four-dimensional flow magnetic resonance imaging (4D of intracardiac flow patterns and energy distribution [1 , flow MRI) allows for comprehensive non-invasive assess- 2]. Given the three-dimensional (3D) time-varying nature ment of 3D time-varying blood flow properties in the heart of these flow patterns and energetics, specialized in vivo and great vessels in all three velocity encoding directions and spatial dimensions over the cardiac cycle [3]. 4D flow MRI has recently emerged as a novel tool * Vivian P. Kamphuis for in vivo quantification of intracardiac f low energetics, v.p.kamphuis@lumc.nl associated energy losses and vortical flow patterns by means of kinetic energy (KE) [4–7], viscous energy loss Division of Pediatric Cardiology, Department of Pediatrics, Leiden University Medical Center, Leiden, The Netherlands (EL) [1] and vorticity [8, 9]. KE is the energy contained in the flow of the bloodstream due to motion and EL is Netherlands Heart Institute, Utrecht, The Netherlands 3 the KE that is irreversibly lost due to viscosity-induced Department of Radiology, Leiden University Medical Center, frictional forces within the blood flow [1]. In acquired Leiden, The Netherlands 4 heart disease, remodelling occurs which can lead to alter- Division of Pediatric Cardiology, Department of Pediatrics, ations in intracardiac hemodynamics [4]. Alteration in Academic Medical Center, Amsterdam, The Netherlands Vol.:(0123456789) 1 3 906 The International Journal of Cardiovascular Imaging (2018) 34:905–920 intraventricular KE derived from 4D f low MRI has been Cardiovascular magnetic resonance acquisition used to assess left ventricular (LV) and right ventricular and data preparation (RV) (dys)function in patients with different stages of heart failure (HF) [4–7]. This is also the case in various Whole-heart 4D flow MRI was obtained on a 3 T scan- congenital heart diseases. Even after correction, patients ner (Ingenia, Philips Medical Systems, The Netherlands) may develop systolic and/or diastolic dysfunction, leading with maximal amplitude of 45 mT/m for each axis, slew to changes in intracardiac flow energetics [1]. In addition, rate of 200  T/m/s and a combination of FlexCoverage intracardiac anatomy may not be restored. In corrected Posterior coil in the table top with a dStream Torso coil, atrioventricular septal defect patients, elevated EL was providing up to 32 coil elements for signal reception. The associated with altered 3D vortex ring formation in the orientation of the acquisition of 4D flow data was identi- LV filling pattern [1]. Vorticity, the curl of velocity, is a cal to the 4-chamber orientation (usually double-oblique fundamental quantity in fluid mechanics that describes axial or coronal). Velocity-encoding of 150 cm/s in all the local spinning rate of fluid particles and can charac- three directions was used in a standard four-point encod- terize vortex flow [10]. Quantitative vortex parameters, ing scheme, spatial resolution 3.0 × 3.0 × 3.0  mm , field- such as vorticity, have been used to assess diastolic (dys) of-view 400 mm, flip angle 10°, echo time (TE) 3.7 ms, function in several patient groups [8, 9, 11, 12]. Further- repetition time (TR) 10  ms, true temporal resolution more, in patients with complex congenital intracardiac 40 ms, SENSitivity Encoding (SENSE) factor 2 in ante- deformations such as after the Fontan operation, flow rior–posterior direction and Echo Planar Imaging (EPI) collision with remaining septal structures and stagnation readout with a factor 5. Free breathing was allowed and of flow through a ventricular septal defect may result in no respiratory motion compensation was performed. Ret- altered EL and vortex formation [13]. rospective gating was used with 30 phases reconstructed Recently, good reproducibility of inf low- and outf low- to represent one cardiac cycle. Expected scan-time for assessment from 4D flow MRI was shown [14]. However, the 4D flow MRI acquisition for a patient with a heart there is a lack of studies validating the reproducibility of rate of 60 bpm and 39 slices would be 9 min and 11 s. intracardiac energy and quantitative vorticity parameters This 4D flow MRI sequence with EPI readout has been from 4D f low MRI in a scan–rescan setting. Scan–rescan validated in vivo and in vitro [15] and compared to other reproducibility is important for clinical applicability as 4D flow MRI sequences [16]. Concomitant gradient cor - it expresses the reliability in repeated quantitation, for rection was performed from standard available scanner example during serial follow-up or in case of a rest–stress software. Cine two-dimensional (2D) left 2-chamber, protocol. Therefore, the aim of this study was to assess 4-chamber, coronal and sagittal aorta views and a cine the scan–rescan reproducibility of 4D flow MRI measure- multi-2D short-axis stack of slices were acquired, using ments of kinetic energy, viscous energy loss and vorticity steady-state free-precession (SSFP) sequences with TE/TR within the LV during diastolic filling in healthy subjects. 1.5/3.0, 350 mm field-of-view, 45° flip angle, acquisition resolution 1.9 × 2.0 × 8.0  mm . Retrospective gating was used with 30 phases reconstructed to represent one cardiac cycle. Expected scan-time for the cine multi-2D short-axis Materials and methods acquisition for a patient with a heart rate of 60 bpm and 12 slices would be 1 min. Free breathing was allowed without Study population using motion suppression, three signal averages were taken to minimize effects of breathing motion, which makes the The study protocol was approved by the local Medical expected scan-time 3 min. Image analysis was performed Ethical Committee of the Leiden University Medical by one observer (VPK) with over 2 years of experience Center and informed consent was obtained from all par- in MRI and verified by one observer (JJMW) with over ticipants. Twelve healthy volunteers with no history of 15 years of experience in MRI. The endocardial border cardiac disease were included. All subjects underwent an was manually traced in all slices and phases in the multi- MRI scan including whole-heart 4D flow MRI between slice 2D cine short-axis images and ventricular volume July 2015 and April 2017. The same scanning protocol was calculated at the end-diastolic and end-systolic phases was performed twice in the same session with 10-min using in-house developed MASS software. Papillary mus- breaks between the scans and repositioning and replan- cles were disregarded and assumed to be included in the ning for every volunteer. Ten of these volunteers were ventricular volume. LV in- and outflow was assessed using included in a recent study [14]. That study did not report the 4D flow MRI data with retrospective valve tracking assessment of KE, EL or vorticity. of the mitral and aortic valve, as shown in a recent study 1 3 The International Journal of Cardiovascular Imaging (2018) 34:905–920 907 [14]. Cardiac output (CO) was computed from the 4D flow amount of kinetic energy lost to that available over cardiac data as LV outflow × Heart rate (HR). Beginning and end- cycle. In this study, EL was computed over diastole as index ing of diastolic phases [early diastolic filling (E-filling) EL /KE , with KE being the total diastole average diastole average diastole and late diastolic filling (A-filling)] were derived from the average KE during total diastole. In order to compare our mitral valve flow-time curves that resulted from retrospec- results with a previous study reporting EL values normalized tive valve tracking. Segmentation of the LV cavity in the by stroke volume (SV) [1], we also report EL values in the 4D flow MRI acquisition, that is required for the energy current study as normalized by SV (reported as norm_ , E-avg EL ∙ ∙ ∙ analysis, was obtained by transforming the available norm_ , norm_ , norm_ and norm_ E-peak A−avg A−peak EL EL EL time-varying segmentation of multi-slice cine short-axis EL ). To be consistent with the previous study SV total diastole acquisition to the 4D flow MRI data. To correct for patient was derived from cine SSFP short-axis slices. motion related misalignment between the two acquisi- tions, automated image-based 3D rigid registration was Integral vorticity magnitude over LV diastole performed using the phase with optimal depiction of the (vorticity_LV) LV cavity in both scans with the Elastix image registration toolbox [17]. Kinetic energy, viscous energy loss and vor- The formulae that were used to calculate the integral vor- ticity analysis of segmented LV volumes was done by one ticity magnitude are shown in “Appendix 1”. In short, fol- investigator (MSME) using in-house developed software. lowing previously published work [8, 9], for each acquired time-phase, voxel-wise vorticity magnitude (1/s) was first KE analysis over LV diastole computed over the segmented LV volume. Then, the instan- taneous integral vorticity magnitude was computed as the The amount of KE during diastolic filling was computed cumulative sum of voxel-wise vorticity and multiplied following previously published methods [1]. KE for each by voxel volume to give the integral in (milliliter × 1/sec- voxel within the LV was computed as 1/2 m v , with (m) ond) i.e. (mL/s). Note that the computed vorticity integral as the mass representing the voxel volume multiplied by parameter is a scalar quantity and therefore does not take the density of blood (1.025 g/mL) and (v) as the 3-direc- the vorticity direction into account. We will refer to this tional velocity from 4D flow MRI. For each acquired time- vorticity integral over the LV as vorticity_LV throughout phase, volumetric KE was then computed by integrating (by the text, to differentiate it from voxel-wise vorticity. In order cumulative sum) the computed KE over the segmented 3D to quantify the integral vorticity_LV over diastolic filling, LV volume. In order to quantify KE during diastolic filling, the time-average and peak vorticity_LV during E-filling the time-averaged KE during diastolic phases (KE and (vorticity_LV, vorticity_LV , respectively) and E-avg E-avg E-peak KE ) and peak KE (KE and KE ), in Joule, A-filling (vorticity_LV, voriticity_LV , respec- A−avg E-peak A−peak A−avg A−peak were assessed. tively) were computed. Viscous EL analysis over LV diastole Scan–rescan analysis Following recently published methods [1], we have computed For the scan–rescan analysis, all data was blinded by one EL from 4D flow MRI using the dissipation terms from the observer (VPK) and presented in a random order to the Navier–Stokes energy equations, assuming blood as a New- observer (MSME) that performed the energy and vorticity tonian fluid. The formulae that were used to calculate EL analysis. Scan–rescan analysis was performed to test the are summarized in “Appendix 1”. The time-averaged viscous reproducibility of (1) KE over E-filling and A-filling; (2) energy loss rate [ EL, in Watt (W)] during E-filling ( ) EL over E-filling, A-filling and total diastolic filling and (3) E-avg EL ∙ ∙ ∙ ∙ vorticity_LV over E-filling and A-filling. and A-filling ( ) and peaks ( and ) A−avg E-peak A−peak EL EL EL EL were assessed. We have used the previously reported correla- Statistical analysis tion between KE and [1] as a measure of in-scan consist- EL ency. Furthermore, EL over the total diastole (EL ) total diastole Data analysis was performed using SPSS Statistics software in Joule (J) was computed. Given that the amount of viscous (v. 23.0 IBM SPSS, Chicago, IL). Variables were tested for energy lost is proportional to the amount of kinetic energy, normal distribution using the Shapiro–Wilk test. Continu- we computed a dimensionless energy loss parameter, EL , index ous data was expressed as mean ± standard deviation (SD) that reflects the amount of viscous energy loss indexed for with minimum and maximum values or as median [inter- the average kinetic energy over diastole. EL is a dimen- index quartile range] where suitable. Mean differences were deter - sionless index that was used in an earlier echo particle image mined for inter-scan comparison and significance was tested velocimetry study by Agati et al. [18] to indicate the relative by a paired samples t test or, in case of non-normality, the 1 3 908 The International Journal of Cardiovascular Imaging (2018) 34:905–920 Wilcoxon signed-rank test. Differences were computed as: correlation (scan 1: r = 0.92, P < 0.001; scan 2: ρ = 0.96, measurement in scan 2—measurement in scan 1. The coef- P < 0.001). Also, correlation between KE and A−avg ficient of variation (CV) was calculated with the root mean was strong–excellent in both scans (scan 1: ρ = 0.87, A−avg EL square method [19]. Correlation between the in-scan and P < 0.001; scan 2: r = 0.99, P < 0.001). inter-scan measurements done in repeated scans was tested by the Pearson correlation coefficient (r), or the Spearman Scan–rescan analysis of kinetic energy and viscous correlation coefficient (ρ) in case of non-normality of the energy loss rate over early diastolic filling data. The approach described by Bland and Altman [20] was used to study systematic differences between measurements Detailed results of the scan–rescan tests of KE and EL obtained from the two scans. Agreement between these meas- assessment over early diastolic filling are shown in Tables  2 urements was assessed by determining the intra-class correla- and 3 and Fig.  3. Scan–rescan assessment showed poor tion (ICC) coefficient. Correlation and agreement were clas- results for early diastolic filling, as shown in Table  2. Of sified as follows: r/ρ and ICC > 0.95: excellent, 0.95–0.85: note, Fig. 3 shows that for KE and assessment over early EL strong, 0.85–0.70: good, 0.70–0.5: moderate, < 0.5: poor. A diastolic filling two subjects showed more distinct differ - P value < 0.05 was considered statistically significant. ences. Throughout the text we will refer to these two sub- jects as “Subject 1” and “Subject 2”. Detailed scan–rescan information of these two subjects is provided in “Appen- Results dix  2”. In Fig.  3a, Subject 1 and 2 are indicated as dark (KE ) and light red triangles (KE ) and in Fig. 3b, as E-avg E-peak Volunteer characteristics are shown in Table 1. Heart rate and ∙ ∙ dark ( ) and light red triangles ( ). E-avg E-peak cardiac output were not significantly different between the two EL EL Possibly some factors related to the 4D flow MRI acqui- scans (HR: 60.8 ± 7.8 vs. 59.9 ± 6.9  bpm, P = 0.52 and CO: sition or physiological factors have resulted in these marked 5.5 ± 0.9 vs. 5.6 ± 1.3  L/min, P = 0.75). 4D flow MRI data differences, therefore we performed an evaluation for a sub- acquisition was successful in all volunteers. Figure 1a–c shows cohort without Subject 1 and 2. Scan–rescan correlations were cross-sectional mapping of the volumetric measurements of much stronger and variation was less for this sub-cohort, as KE, EL and vorticity inside the LV at peak E-filling in a stand- shown in Table  3. KE and assessment showed E-avg E-avg ard 4-chambers view. An example of the temporal evolution of EL non-significant differences between the two scans, strong KE and over LV diastole is shown in one subject in Fig. 1d. EL ∙ ICCs (KE : 0.95, P < 0.001 and : 0.91, P = 0.03) and E-avg E-avg EL ∙ ∙ CVs ≤ 11% (KE : 10% and : 11%). norm_ Intra-scan comparison of energetics (KE vs. EL) E-avg E-avg E-avg EL EL showed a strong ICC (0.90, P = 0.001) and a CV of 12%. Scan–rescan assessment of KE and showed non- Figure 2 shows the in-scan comparison of KE versus EL . E-peak E-peak EL significant differences. The ICC of KE was good (0.82, Comparison of KE to showed strong–excellent E-peak E-avg E-avg EL Table 1 Baseline characteristics Scan-independent characteristics N 12 Male (%) 6/12 (50%) Age (years) 27 ± 3 Height (cm) 175 ± 7 Weight (kg) 69 ± 12 BSA (m ) 1.8 ± 0.2 BMI (kg/m ) 22 ± 3 Characteristics per scan Scan 1 Scan 2 P value HR (bpm) 60.8 ± 7.8 59.9 ± 6.9 0.52 End-diastolic volume (mL) 143.8 (132.4–183.8) 158.6 ± 30.3 0.64 End-systolic volume (mL) 59.0 ± 12.8 56.2 ± 12.9 0.27 Stroke volume (mL) 100.8 ± 21.8 102.4 ± 21.0 0.44 Ejection fraction (%) 62.8 ± 3.6 63.1 (61.6–65.4) 0.21 CO (L/min) 5.5 ± 0.9 5.6 ± 1.3 0.75 bpm beat per minute, BMI Body Mass Index, BSA body surface area, CO cardiac output, HR heart rate, LV left ventricular 1 3 The International Journal of Cardiovascular Imaging (2018) 34:905–920 909 Fig. 1 Maps of left ventricular kinetic energy (KE), viscous energy ventricular viscous energy loss rate at peak early diastolic filling, c loss rate ( ) and vorticity over the LV of a healthy female subject left ventricular voxel-wise vorticity at peak early diastolic filling, d EL (age 20  years) in a standard 4-chambers MRI cross-sectional view, temporal evolution of volumetric kinetic energy, viscous energy loss a left ventricular kinetic energy at peak early diastolic filling, b left rate and vorticity over LV diastole now within the same range as the other values [Fig. 3c, indi- P = 0.01) with a CV of 15%. Similarly the ICC of E-peak EL cated as dark (KE ) or light red triangles (KE )]. was good (0.76, P = 0.03) with a CV of 17%. Furthermore, A−avg A−peak ∙ ∙ The same was observed when assessing and norm_ showed a strong ICC (0.86, P = 0.005) and a A−avg A−peak EL EL E-peak EL [Fig. 3d, indicated as dark ( ) or light red triangles CV of 16%. A−avg EL ( )]. However, to be consistent we repeated the A−peak EL evaluation in the sub-cohort without Subject 1 and Subject Scan–rescan analysis of kinetic energy and viscous 2. Scan–rescan correlations and variation were similar for energy loss rate over late diastolic filling the sub-cohort without Subject 1 and Subject 2, as shown in Table 3. Reproducibility of KE and assessment Detailed results of the scan–rescan tests of KE and assess- A−avg A−avg EL EL showed non-significant differences, good ICCs (KE : ment over late diastolic filling are shown in Tables  2 and 3 A−avg 0.77, P = 0.02 and : 0.75, P = 0.03) and CVs up and Fig. 3. Scan–rescan assessment showed good results for A−avg EL ∙ ∙ to 24% (KE : 23% and : 24%). norm_ late diastolic filling, as shown in Table  2. Subject 1 and 2 are A−avg A−avg A−avg EL EL 1 3 910 The International Journal of Cardiovascular Imaging (2018) 34:905–920 norm_EL ) and EL are shown in Tables 2 and total diastole index 3 and Fig. 3. Scan–rescan assessment showed poor results for total diastole, but strong results for EL , as shown in index Table 2. Scan–rescan correlations were much stronger and variation was less for the sub-cohort without Subject 1 and Subject 2 (Table 3). Scan–rescan assessment of EL total diastole showed a non-significant difference between the scans, a strong ICC of 0.91 (P = 0.001) and a CV of 11%. norm_ EL showed a non-significant difference between total diastole the scans, a good ICC (0.81, P = 0.01) and a CV of 12%. Figure 3f shows the Bland–Altman plot of EL . For all index subjects, scan–rescan assessment of EL showed excel- index lent reproducibility with a small non-significant difference between the scans, a strong ICC of 0.94 (P < 0.001) and a CV of 9%. When evaluating the sub-cohort without Subject 1 and 2 the results remained similar (ICC: 0.95, P < 0.001 and CV: 8%). Scan–rescan analysis of volumetric vorticity_LV over early diastolic filling Detailed results of the scan–rescan tests of vorticity_LV assess- ment over early diastolic filling are shown in Tables  4 and 5 and Fig. 4. Scan–rescan assessment showed moderate results for early diastolic filling as shown in Table  4. The Bland–Alt- man plots of the assessment of vorticity_LV and vorticity_ E-avg LV showed higher differences between scan and rescan E-peak measurements for Subject 1 and 2 (Fig. 4a, indicated as dark (vorticity_LV ) or light red triangles (vorticity_LV )). E-avg E-peak Scan–rescan correlations were much stronger and variation was less for the sub-cohort without Subject 1 and 2 (Table 5). Fig. 2 Scatter plots of kinetic energy (KE) versus viscous energy ∙ ∙ Scan–rescan assessment of v orticity_LV and vorticity_ E-peak loss rate ( ) during early diastolic filling and KE versus during EL EL LV showed a non-significant difference between the late diastolic filling. a Scatter plot depicting the correlation between E-avg KE and measured in scan 1 (grey) and scan 2 (black), scans, good-strong ICCs (vorticity_LV : 0.83, P = 0.01 E-avg ELE-avg E-peak in all 12 subjects, b scatter plot depicting the correlation between and vorticity_LV : 0.95, P < 0.001) and CVs up to 11% E-avg KE and measured in scan 1 (grey) and scan 2 (black), A−avg ELA−avg (vorticity_LV : 11% and vorticity_LV : 7%). E-peak E-avg in all 12 subjects Scan–rescan analysis of volumetric vorticity_LV showed a non-significant difference between the scans, a over late diastolic filling good ICC (0.70, P = 0.048) and a CV of 22%. Scan–rescan assessment of KE and showed non-signifi- A−peak A−peak EL Detailed results of the scan–rescan tests of vorticity_LV cant differences, good ICCs (KE : 0.83, P = 0.01 and A−peak assessment over late diastolic filling are shown in Tables  4 : 0.79, P = 0.02) and CVs up to 25% (KE : A−peak A−peak EL and 5 and Fig.  4. Scan–rescan assessment showed good 23% and : 25%). Lastly, scan–rescan assessment of A−peak EL results for late diastolic filling as shown in Table  4. Subject norm_ showed a non-significant difference between A−peak EL 1 and 2 are now within the same range as the other values the scans, a moderate ICC (0.64, P = 0.08) and a CV of 24%. (Fig. 4b, indicated as dark (vorticity_LV ) and light red A−avg triangles (vorticity_LV )). Scan–rescan variation was A−peak Scan–rescan analysis of kinetic energy and viscous similar for the sub-cohort without Subject 1 and 2 (Table 5). energy loss over total diastole Scan–rescan assessment of v orticity_LV and vorticity_ A−peak LV showed non-significant differences between the A−avg Detailed results of the scan–rescan tests of KE and scans, good-strong ICCs (vorticity_LV : 0.91, P = 0.001 A−peak assessment over total diastole (EL and and vorticity_LV : 0.89, P = 0.002) and CVs of 11%. total diastole A−avg EL 1 3 The International Journal of Cardiovascular Imaging (2018) 34:905–920 911 1 3 Table 2 Scan–rescan comparison of energy variables for the cohort with all subjects (N = 12) Scan 1 Scan 2 Difference (scan 2–scan 1) Pearson correlation Intraclass correlation coef- Coefficient of coefficient ficient variation (%) Mean ± SD (min–max) or Mean ± SD (min–max) or Mean ± SD or (min–max) ICC (95% CI) median [IQR] (min–max) median [IQR] (min–max) median [IQR] (min–max) Early filling  KE (mJ) 4.9 ± 1.9 (2.6–9.6) 4.8 ± 1.4 (2.0–7.6) 0.1 ± 2.3 (− 4.8 to 5.0) 0.06 0.12 (− 2.78 to 0.76) 28 E-peak  KE (mJ) 1.9 ± 0.7 (0.8–3.5) 1.9 [1.3–2.3] (1.1–4.0) − 0.01 [− 0.1 to 0.3] (− 2.2 0.52 0.31 (− 1.85 to 0.81) 26 E-avg to 2.1) 1.5 ± 0.4 (0.8–2.1) 1.5 ± 0.6 (0.7–2.9) 0.04 ± 0.7 (− 1.4 to 1.7) 0.02 0.04 (− 3.25 to 0.74) 31   (mW) EL E-peak 3 15.2 ± 5.1 (9.4–24.9) 14.9 ± 5.1 (6.0–21.8) − 0.2 ± 6.2 (− 12.6 to 12.3) 0.27 0.45 (− 1.17 to 0.85) 30  norm_ (W/m ) EL E-peak 0.7 ± 0.2 (0.3–1.1) 0.7 ± 0.3 (0.4–1.4) − 0.01 [− 0.1 to 0.3] (− 0.7 0.11 0.21 (− 2.36 to 0.78) 28   (mW) ELE-avg to 0.8) 7.0 ± 2.3 (4.2–11.4) 6.6 ± 2.1 (3.2–10.4) − 0.4 ± 2.8 (− 6.4 to 5.8) 0.20 0.35 (− 1.55 to 0.82) 28  norm_ (W/m ) EL E-avg Late filling c c b b  KE (mJ) 0.9 [0.7–2.0] (0.7–2.5) 1.0 [0.9–1.6] (0.6–2.6) 0.1 [− 0.2 to 0.3] (− 1.2 to 0.62 0.87 (0.55–0.96) 23 A−peak 0.5) c b b  KE (mJ) 0.6 [0.5–1.2] (0.4–1.4) 0.8 ± 0.4 (0.4–1.6) − 0.03 ± 0.3 (− 0.7 to 0.4) 0.68 0.84 (0.42–0.95) 23 A−avg b b 0.5 ± 0.3 (0.3–1.1) 0.5 ± 0.2 (0.3–1.0) − 0.007 ± 0.2 (–0.5 to 0.3) 0.67 0.81 (0.33–0.95) 24   (mW) ELA−peak ∙ b 5.3 ± 1.8 (2.5–8.3) 5.1 ± 1.5 (3.3–7.9) − 0.2 ± 1.7 (− 3.2 to 2.2) 0.50 0.68 (− 0.19 to 0.91) 22  norm_ (W/m ) ELA−peak ∙ c b 0.3 [0.2–0.6] (0.2–0.7) 0.4 ± 0.2 (0.2–0.8) − 0.02 ± 0.1 (− 0.4 to 0.2) 0.50 0.80 (0.29–0.94) 23   (mW) EL A−avg b b 3.7 ± 1.3 (2.1–6.6) 3.4 ± 1.1 (2.2–5.2) − 0.3 ± 1.1 (− 2.8 to 1.5) 0.58 0.73 (0.07–0.92) 21  norm_ (W/m ) EL A−avg Total diastole c b  EL (mJ) 0.5 ± 0.2 (0.3–1.0) 0.5 [0.4–0.6] (0.3–1.3) − 0.006 [− 0.07 to 0.06] 0.81 0.58 (− 0.60 to 0.88) 22 total diastole (− 0.4 to 0.7) 3 c  norm_EL (J/m ) 5.3 ± 1.5 (3.4–8.1) 4.8 [4.2–5.3] (3.1–9.8) − 0.1 ± 2.2 (− 3.9 to 5.3) 0.30 0.17 (− 2.52 to 0.77) 23 total diastole a a  EL 0.3 ± 0.07 (0.2–0.4) 0.3 ± 0.06 (0.2–0.4) − 0.001 ± 0.03 (− 0.04 to 0.89 0.94 (0.80–0.98) 9 index 0.06) KE kinetic energy, E early diastolic filling, A late diastolic filling, EL viscous energy loss P < 0.001 P < 0.05 P < 0.05 for non-normality (Shapiro–Wilk test) 912 The International Journal of Cardiovascular Imaging (2018) 34:905–920 1 3 Table 3 Scan–rescan comparison of energy variables for the sub-cohort without Subject 1 and 2 (N = 10) Scan 1 Scan 2 Difference (scan 2–scan 1) Pearson correlation Intraclass correlation coef- Coefficient of coefficient ficient variation (%) Mean ± SD (min–max) or Mean ± SD (min–max) or Mean ± SD or (min–max) ICC (95% CI) median [IQR] (min–max) median [IQR] (min–max) median [IQR] (min–max) Early filling c b  KE (mJ) 5.0 [4.0–5.3] (2.0–5.4) 4.6 ± 1.1 (2.6–6.4) 0.1 ± 0.9 (− 0.9 to 2.0) 0.62 0.82 (0.27–0.96) 15 E-peak a a  KE (mJ) 1.7 ± 0.5 (0.8–2.3) 1.8 ± 0.5 (1.1–2.4) 0.05 ± 0.2 (− 0.2 to 0.5) 0.91 0.95 (0.82–0.99) 10 E-avg 1.5 ± 0.4 (0.8–2.1) 1.5 ± 0.4 (0.7–2.2) 0.01 ± 0.4 (− 0.5 to 0.9) 0.58 0.76 (− 0.07 to 0.94) 17   (mW) ELE-peak ∙ b b 15.4 ± 5.2 (9.6–24.9) 15.2 ± 4.2 (8.9–21.8) − 0.3 ± 3.5 (− 4.7 to 6.9) 0.75 0.86 (0.40–0.97) 16  norm_ (W/m ) EL E-peak ∙ b b 0.7 ± 0.2 (0.3–0.9) 0.6 ± 0.2 (0.4–0.9) − 0.02 ± 0.1 (− 0.2 to 0.2) 0.83 0.91 (0.64–0.98) 11   (mW) EL E-avg b b 7.0 ± 2.3 (4.2–11.4) 6.6 ± 1.6 (4.2–9.0) − 0.4 ± 1.2 (− 2.4 to 1.6) 0.87 0.90 (0.60–0.97) 12  norm_ (W/m ) EL E-avg Late filling c b  KE (mJ) 0.9 [0.7–1.7] (0.7–2.5) 1.2 ± 0.5 (0.6–2.4) 0.2 [− 0.1 to 0.3] (− 1.2 to 0.59 0.83 (0.25–0.96) 23 A−peak 0.5) c b b  KE (mJ) 0.5 [0.5–1.2] (0.4–1.4) 0.7 ± 0.3 (0.4–1.3) − 0.02 ± 0.3 (− 0.7 to 0.4) 0.65 0.77 (0.02–0.94) 23 A−avg ∙ c b 0.4 [0.3–0.7] (0.3–1.1) 0.5 ± 0.2 (0.3–0.9) − 0.03 ± 0.2 (− 0.5 to 0.3) 0.50 0.79 (0.11–0.95) 25   (mW) ELA−peak 5.3 ± 2.0 (2.5–8.3) 5.0 ± 1.3 (3.3–7.0) − 0.3 ± 1.7 (− 3.2 to 2.2) 0.49 0.64 (− 0.56 to 0.91) 24  norm_ (W/m ) EL A−peak c b 0.3 [0.2–0.6] (0.2–0.8) 0.3 ± 0.1 (0.2–0.6) − 0.03 ± 0.1 (− 0.4 to 0.2) 0.42 0.75 (− 0.05 to 0.94) 24   (mW) EL A−avg 3.6 ± 1.4 (2.1–6.6) 3.4 ± 1.0 (2.3–5.2) − 0.3 ± 1.2 (− 2.8 to 1.5) 0.57 0.70 (− 0.21 to 0.93) 22  norm_ (W/m ) EL A−avg Total diastole b b  EL (mJ) 0.5 ± 0.2 (0.3–1.0) 0.5 ± 0.1 (0.3–0.7) − 0.01 [− 0.05 to 0.03] (− 0.3 0.88 0.91 (0.64–0.98) 11 total diastole to 0.08) 3 b b  norm_EL (J/m ) 5.1 ± 1.3 (3.4–7.0) 4.8 ± 0.9 (3.1–6.4) − 0.3 ± 0.9 (− 2.0 to 1.2) 0.73 0.81 (0.27–0.95) 12 total diastole a a  EL 0.3 ± 0.06 (0.2–0.4) 0.3 ± 0.07 (0.2–0.4) − 0.01 ± 0.03 (− 0.04 to 0.06) 0.89 0.95 (0.79–0.99) 8 index KE kinetic energy, E early diastolic filling, A late diastolic filling, EL viscous energy loss *Subjects with measurements outside the 95% confidence interval of the respective Bland–Altman plots P < 0.001 P < 0.05 P < 0.05 for non-normality (Shapiro–Wilk test) The International Journal of Cardiovascular Imaging (2018) 34:905–920 913 filling rate and the area-under-the-curve are sensitive to tem- Discussion poral and spatial resolution. These observations are important to take into consideration when evaluating flow energetics and In the current study, kinetic energy, viscous energy loss and vorticity in a research or clinical setting. vorticity inside the left ventricle during diastole are derived from 4D flow MRI and scan–rescan reproducibility of these Kinetic energy (KE) over LV diastole parameters is tested. Scan–rescan reproducibility is essential for the clinical application of a parameter since it reflects Multiple 4D flow MRI studies showed that patients with LV the reliability of a measurement and feasibility of repeated dysfunction present altered flow patterns through the LV with measurement evaluations. The main findings of this study impaired preservation of inflow KE to the end of diastole and are: (1) internal consistency between kinetic energy and vis- altered KE-time curves (the amount of KE inside the LV cous energy loss is strong-excellent in both scans during during each time step over the total cardiac cycle) [4, 6, 7, early and late diastolic filling; (2) In the majority (10 out of 21], even in patients with normal to mild LV remodeling and 12) of subjects, reproducibility of peak and average kinetic normal to mildly depressed LV systolic function [7]. These energy and viscous energy loss during early, late and total KE changes in the LV could be a valuable diagnostic marker diastolic filling shows non-significant differences but with a to evaluate diastolic function and might be useful for early good to excellent agreement (by means of ICC) and CVs up detection of deteriorating ventricular function [1, 4–7, 21, to 25%; (3) In the studies parameters, time-averaged meas- 22], which could reduce patient morbidity and mortality [23]. urements over E- and A-filling show stronger reproducibility However, there is a lack of studies validating LV KE derived than peak measurements. (4) For all subjects, EL shows index from 4D flow MRI in a scan–rescan setting. Therefore, reli- good reproducibility with a small non-significant difference ability and reproducibility of KE measurements from 4D flow between the scans, strong agreement and a CV of 9%; (5) MRI in a repeated scan setting remains largely unknown. Assessment of volumetric vorticity over the left ventricle Kanski et al. [24] compared mean KE and peak KE between during early and late diastolic filling shows non-significant two scans (with and without respiratory gating) with the aim differences, good–excellent ICCs and CVs up to 11%. to evaluate the impact of respiratory gating on KE measure- In the Bland–Altman plots for E–filling parameters, meas- ments. They found a strong correlation between the KE meas- urements obtained in two subjects (Subject 1 and 2 in Appen- urements in both scans. The current study differs from the dix 2) showed distinct higher differences between scan and study by Kanski et al. [24] in that we used the same protocol rescan measurements than the measurements of the majority for both scans. In the current study, moderate–strong corre- of the cohort. These high differences had an impact on the lation was found between KE measurements in both scans. scan–rescan reproducibility for KE and measures. This EL Absolute values of KE and KE reported in the cur- impact seemed to be largest during early diastolic filling, where E-peak A−peak rent study are in agreement with previous studies [1, 25, 26]. reproducibility was much higher for the sub-cohort without In the current study we showed the reproducibil- Subject 1 and 2. However, during late diastolic filling, repro- ity of the KE measurements assessed with 4D flow MRI ducibility was similar. There is no obvious explanation for the with good–strong agreement (by means of ICCs), how- measurements that showed distinct higher die ff rences between ever substantial CVs up to 23% were also found. Based scan and rescan measurement, but this could be related to the on the Bland–Altman analysis the variability of K E technical restrictions of the 4D flow MRI acquisition or to phys- E-avg and KE was less than the variability of KE and iological differences between scans, or a combination of both. A−avg E-peak KE as shown by the smaller limits of agreement. This It is to be expected that physiological differences were small A−peak might be expected, given that the definition of average is as all subjects are healthy volunteers that were scanned twice computed over multiple time points and therefore evens under the same circumstances with only a short break between out the variations more than a single-time measure such as the scans (± 10 min) and heart rate was not significantly differ - the peak, especially when the definition of the peak also is ent between both scans. Still, subtle physiological differences affected by the temporal resolution of the 4D flow MRI data. could result in poor scan–rescan reproducibility for these few cases. Table 6 in “Appendix 2” shows the HR, LV outflow and Viscous EL over LV diastole CO of all subjects. Subject 1 and 2 present the highest CO dif- ferences, however there are other subjects with CO differences The assessment of intracardiac EL could provide crucial within the same range. Table 7 in “Appendix 2” shows subject- details on the function of the heart apart from the standard specific scan–rescan information from the mitral valve flow MRI parameters and could be used to further unravel the curves. Subject 1 and 2 have the highest difference in area under influence of complex surgery for congenital heart defects [1 ]. the curve of the E-filling and peak filling rate of the E-filling. To our knowledge, this is the first study assessing This could indicate that differences may be related to technical scan–rescan reproducibility of in vivo LV EL over diastole restrictions of the acquisition, as the E-filling duration, E peak 1 3 914 The International Journal of Cardiovascular Imaging (2018) 34:905–920 1 3 The International Journal of Cardiovascular Imaging (2018) 34:905–920 915 ◂Fig. 3 Bland–Altman plots of kinetic energy (KE) and viscous energy their absolute changes as in other tested parameters. As such, ∙ ∙ loss rate ( ) at peak early diastolic filling (KE and ) EL EL E-peak E-peak ∙ the reported reproducibility of EL could also be consid- index and peak late diastolic filling (KE and ), EL EL A−peak A−peak total diastole ered as another reflection of a good internal consistency in and EL . a Bland–Altman plot depicting the agreement between index this study. Although a similar EL parameter was reported KE (black) and KE (grey) in scan 1 and scan 2. The subjects index E-avg E-peak with distinct higher differences between scan and rescan measure- in a previous study and was shown to be significantly altered ments (Subject 1 and 2) are depicted as red triangles (KE dark E-avg in patients with acute myocardial infarction [18], EL was index red; KE light red). b Bland–Altman plot depicting the agreement E-peak ∙ ∙ computed in that study from 2D echo particle image veloci- between (black) and (grey) in scan 1 and scan 2. Sub- EL EL E-avg E-peak metry and over the complete cardiac cycle as compared to this ject 1 and 2 are depicted as red triangles (KE dark red; KE E-avg E-peak light red). c Bland–Altman plot depicting the agreement between study’s volumetric measurement from 4D flow MRI and over KE (black) and KE (grey) in scan 1 and scan 2. Subject 1 A−avg A−peak diastole only. Therefore, it is not possible to perform a direct and 2 are depicted as red triangles (KE dark red; KE light A−avg A−peak comparison with the published results of that study. red). d Bland–Altman plot depicting the agreement between EL A−avg (black) and (grey) in scan 1 and scan 2. Subject 1 and 2 EL A−peak ∙ ∙ Vorticity inside the LV during diastolic filling are depicted as red triangles ( dark red; light red). EL EL A−avg A−peak e Bland–Altman plot depicting the agreement between EL total diastole in scan 1 and scan 2. Subject 1 and 2 are depicted as red triangles. f LV vortex quantification parameters, such as vorticity, Bland–Altman plot depicting the agreement between EL in scan 1 index could be useful in the assessment of LV and RV diastolic and scan 2. Subject 1 and 2 are depicted as red triangles (dys-)function [1, 8, 9, 11, 12]. In recent studies, vorticity was shown to be a marker of diastolic dysfunction, both assessed by 4D flow MRI. Results of norm_ in the in the LV [9] and the RV [8] of patients with pulmonary E-peak EL current study correspond well with that of a different healthy hypertension. controls cohort of a recent 4D flow MRI study by Elbaz et al. To our knowledge, no previous study is available on [1] Also norm_ values are consistent. Furthermore, assessing scan–rescan reproducibility of in vivo vorticity_LV E-avg EL ∙ ∙ both results of norm_ , as well as norm_ are over diastole from 4D flow MRI. Fenster et al. [8 ] assessed A−peak A−avg EL EL similar to previously reported numbers [1]. vorticity inside the RV using the integral of vorticity mag- Both EL and norm_EL are slightly nitude over the volumes and found results in the same order total diastole total diastole higher in our study than in the study by Elbaz et al. [1]. of magnitude as the results in the current study. In a recent However, results remain in the same range. These differ- paper by Schafer et al. [9] LV vorticity in healthy subjects ences in results could be explained in part by differences was assessed. However, it is not clear whether this was com- in heart rate between the volunteers of this study and puted over the whole LV volume and therefore we cannot those of the previous study. This is because the total vis- compare results to our measurements. The current study cous energy loss over diastole is computed over the time showed scan–rescan reproducibility of integral vorticity_LV period between the first and the last phase of the diastole during E-filling as well as during A-filling with good–excel- (Eq. 3) in “Appendix 1”) and heart rate mainly affects the lent ICCs and CVs up to 11%. Based on the Bland–Altman duration of diastole. Still, the EL-time curve over total analyses, similar to the KE and EL results, the variability of diastole in the current study (Fig. 1d) agrees well with the average of the vorticity_LV and vorticity_LV E-avg A−avg reported in vivo [1] and in vitro [2] EL time curves. was less than the variability of the peaks v orticity_LV E-peak In this study we demonstrated the reproducibility of the and vorticity_LV. Vorticity_LV shows less vari- A−peak E-avg EL parameters with moderate–strong ICCs and substan- ability than vorticity_LV but vorticity_LV shows A−avg E-peak tial CVs of up to 25%. Similar to KE, the Bland–Altman slightly more variability than vorticity_LV . A−peak ∙ ∙ analysis shows that the variability of and E-avg A−avg EL EL ∙ ∙ was less than the variability of and Technical considerations E-peak A−peak EL EL The amount of energy lost over diastole relative to the aver- age kinetic energy as measured by means of EL shows The post-processing steps that are followed for obtaining index, good reproducibility for all subjects with an ICC of 0.94 these flow energetics and vorticity parameters involve man- (P < 0.001) and an CV of 9%. The Bland–Altman plot shows ual segmentation, registration and valve tracking. Manual that all subjects are within the same range of differences. This segmentation on cine short axis images can be performed suggests that among tested parameters EL is the least sensi- with excellent reproducibility [27]. In this study, manual index tive to subtle physiological variations or discrepancies affected segmentation was performed by one observer with over 2 by technical limitations of the 4D flow MRI data in the healthy years of experience in MRI and verified by another observer subjects, which might have affected the lesser reproducibility with over 15 years of experience, with validated software of KE and EL in some subjects. This observation could be [28], which warrants high accuracy. Next, to correct for attributed to the fact that EL is a dimensionless parameter patient motion related misalignment and minimize errors index concerned with the relative changes in EL to KE and not with between the cine short axis and the 4D flow acquisitions, 1 3 916 The International Journal of Cardiovascular Imaging (2018) 34:905–920 1 3 Table 4 Scan–rescan comparison of vortex flow variables for the cohort with all subjects (N = 12) Scan 1 Scan 2 Difference (scan 2–scan 1) Pearson correla- Intraclass correla- Coefficient of tion coefficient tion coefficient variation (%) Mean ± SD (min–max) or Mean ± SD (min–max) or Mean ± SD (min–max) or median [IQR] ICC (95% CI) median [IQR] (min–max) median [IQR] (min–max) (min–max) Early filling  Vorticity_LV (mL/s) 4671 ± 992 (2950–6370) 4551 ± 1043 (2930–6200) − 120 ± 1075 (− 239 to 1710) 0.44 0.63 (− 0.37 to 0.90) 15 E-peak  Vorticity_LV (mL/s) 2862 ± 636 (1720–3840) 2788 ± 671 (1810–3930) − 74 ± 608 (− 1510 to 1060) 0.57 0.74 (0.06–0.93) 13 E-avg Late filling b a  Vorticity_LV (mL/s) 2715 [2502–4142] (2176– 3193 ± 945 (2174–4914) − 10 ± 547 (− 1270 to 840) 0.75 0.92 (0.73–0.98) 11 A−peak 5463) b a  Vorticity_LV (mL/s) 2660 ± 842 (1740–4180) 2599 ± 766 (1730–4050) − 61 ± 484 (− 1270 to 740) 0.82 0.91 (0.68–0.97) 11 A−avg E early diastolic filling, A late diastolic filling *Subjects with measurements outside the 95% confidence interval of the respective Bland–Altman plots P < 0.001 P < 0.05 P < 0.05 for non-normality (Shapiro–Wilk test) Table 5 Scan–rescan comparison of vortex flow variables for the sub-cohort without Subject 1 and 2 (N = 10) Scan 1 Scan 2 Difference (scan 2–scan 1) Pearson correlation Intraclass correlation Coefficient of coefficient coefficient variation (%) Mean ± SD (min–max) or Mean ± SD (min–max) or Mean ± SD (min–max) or ICC (95% CI) median [IQR] (min–max) median [IQR] (min–max) median [IQR] (min–max) Early filling b b  Vorticity_LV (mL/s) 4546 ± 1008 (2950–6370) 4520 ± 1004 (2930–6200) − 27 ± 796 (− 940 to 1710) 0.69 0.83 (0.27–0.96) 11 E-peak a a  Vorticity_LV (mL/s) 2764 ± 614 (1720–3560) 2720 ± 613 (1810–3610) − 44 ± 277 (− 380 to 480) 0.90 0.95 (0.80–0.99) 7 E-avg Late filling b b  Vorticity_LV (mL/s) 2563 [2468–3628] (2176– 3012 ± 852 (2174–4586) − 69 ± 575 (− 1270 to 840) 0.64 0.91 (0.64–0.98) 11 A−peak 5463) b b  Vorticity_LV (mL/s) 2269 [1889–3251] (1740– 2465 ± 679 (1730–3730) − 75 ± 511 (− 1270 to 740) 0.82 0.89 (0.53–0.97) 11 A−avg 4180) E early diastolic filling, A late diastolic filling *Subjects with measurements outside the 95% confidence interval of the respective Bland–Altman plots P < 0.001 P < 0.05 P < 0.05 for non-normality (Shapiro–Wilk test) The International Journal of Cardiovascular Imaging (2018) 34:905–920 917 Limitations A limitation of this study is the small number of subjects. Furthermore, no patients were scanned in this study as this is difficult to assess in clinical research. In most of our patient studies, 4D flow MRI is part of a clinical CMR evaluation of about 75–90 min, sometimes involving the use of contrast for late gadolinium enhancement and sometimes involving a Dobutamine rest/stress protocol. Scan–rescan evaluation in such cases would imply repeating some of these evaluations as well. Furthermore, repeating 4D flow MRI for scan–rescan purposes would imply an additional scan-time of approxi- mately 30 min (because of replanning), which makes the total scan-time too long to keep the patient in the same physiologi- cal state. Another limitation is that the influence of a different scanning protocol or scanner was not assessed. However, it is important to note that in the current study the aim was to assess scan–rescan reproducibility by using the exact same protocol and the same scanner machine twice. The use of dif- ferent scanners or scanning protocols could result in altered reproducibility, which should be evaluated in future studies. In conclusion, left ventricular kinetic energy and viscous energy loss quantification from 4D flow MRI in healthy vol- unteers shows strong–excellent in-scan consistency. Scan–res- can assessment of left ventricular kinetic energy, viscous energy loss and vorticity shows overall good agreement in the majority of the scanned subjects. Nevertheless, in two out of twelve subjects, considerable variation between the scans was found. Agreement of A-filling measurements is better than E-filling between scans in the studied parameters. Further - more, time-averaged measurements over early and late filling show better reproducibility compared to peak measurements. Strong reproducibility for all cases is found in the dimension- less index, EL , that measures the ratio of the amount of index viscous energy lost relative to the average kinetic energy over diastole. EL seems to be less influenced by technical and/ Fig. 4 Bland–Altman plots of volumetric vorticity over peak early index filling (vorticity_LV and vorticity_LV ) and late filling or slight physiological differences between scans and may E-avg E-peak (vorticity_LV and vorticity_LV ). a Bland–Altman plot A−avg A−peak therefore be a useful parameter of energetics for future studies. depicting the agreement between v orticity_LV (black) and E-avg vorticity_LV (grey) in scan 1 and scan 2. Subject 1 and 2 are E-peak Acknowledgements The authors thank professor Dr. S. le Cessie for depicted as red triangles (v orticity_LV dark red; v orticity_ E-avg statistical advice. LV light red). b Bland–Altman plot depicting the agreement E-peak vorticity_LV (black) and v orticity_LV (grey) in scan 1 A−avg A−peak Funding V.P. Kamphuis and R.L.F. van der Palen are financially sup- and scan 2. Subject 1 and 2 are depicted as red triangles (vorticity_ ported by grants from the Dutch Heart Foundation (Grant Numbers LV dark red; vorticity_LV light red) A−avg A−peak 2013T091 and 2014T087, respectively). J.J.M. Westenberg is finan - cially supported by a Grant of ZonMw (Project Number 104003001). automated image-based 3D rigid registration was performed Compliance with ethical standards using the validated Elastix image registration toolbox [17]. Another post-processing step requires retrospective valve Conflict of interest The authors declare that they have no conflict of tracking and mitral flow velocity mapping for assessing the interest. beginning and ending of diastole. The reproducibility and observer variability of this semi-automated method was Open Access This article is distributed under the terms of the Crea- shown to be excellent [14, 15]. tive Commons Attribution 4.0 International License (http://crea- tivecommons.org/licenses/by/4.0/), which permits unrestricted use, 1 3 918 The International Journal of Cardiovascular Imaging (2018) 34:905–920 distribution, and reproduction in any medium, provided you give appro- As a result, the total viscous energy loss ( EL ) in joules over priate credit to the original author(s) and the source, provide a link to time period T starting at phase t and ending at t can be start end the Creative Commons license, and indicate if changes were made. computed as: end EL = EL p [Joule(J)] (3) T d d Appendix 1 d=t start Viscous energy loss computation from 4D flow MRI with p the time step (temporal resolution) of the acquired 4D flow MRI. Given the acquired velocity field v, the rate of viscous energy loss ( ) in watt (W) and the total energy loss ( EL ) in total EL Vorticity computation from 4D flow MRI joule (J) over a given period of time T can be computed from 4D flow MRI using the viscous dissipation function  in the If u,v, w denote the three velocity field components acquired Newtonian Navier–Stokes energy equations: from 4D flow MRI over the principal velocity directions 3 3 2 x, y, z, respectively, the vorticity (  ) at voxel i of an acquired i,t 1 i 2 = + − (∇ ⋅ v) , time phase t v ij 2 x x 3 j i i=1 j=1 w v u w v u i,t i,t i,t i,t i,t i,t = − , − , − 1∕s = 1, if i = j i,t ij −2 (1) y z z x x y [s ] i,t i,t i,t i,t i,t i,t = 0, if i ≠ j ij (4) Then, Vorticity_LV denoting the integral sum of vorticity represents the rate of viscous energy dissipation per unit volume. i, j correspond to the principal velocity direc- over the segmented LV volume at an acquired time phase t in liter (mL) per second (s) can be computed as tions x, y, z. ∇ ⋅ v denotes the divergence of the velocity field. Therefore, the rate of viscous energy loss ( ) in Watt M EL at an acquired time phase t can be computed as: Vorticity_LV =  L mL∕s (5) t i,t i,t i=1 With   as the magnitude of the vorticity vector, M as i,t EL =μ  L [Watt(W)] (2) t v i the total number of voxels in the segmented LV volume and i=1 L as the voxel volume. i,t assuming the blood as a Newtonian fluid, the dynamic vis- cosity is μ = 0.004 Pa  s, N as the total number of voxels in the given domain of interest (e.g. LV), L as the voxel Appendix 2 volume. See Tables 6 and 7. Table 6 Subject specific scan–rescan assessment Subject HR 1 (bpm) HR 2 Difference LV outflow LV outflow Difference LV CO 1 (L/min) CO 2 (L/min) Difference (bpm) HR (bpm) 1 (mL) 2 (mL) outflow (mL) CO (%) 1 67.1 69.6 2.5 104.1 124.8 20.7 7.0 8.7 24.4 2 53.8 51.6 − 2.2 112.9 94.5 − 18.4 6.1 4.9 − 19.7 3 58.0 51.8 − 6.2 110.0 126.4 16.4 6.4 6.6 2.7 4 61.5 59.1 − 2.4 81.5 72.4 − 9.1 5.0 4.3 − 14.7 5 71.9 66.2 − 5.7 68.4 66.0 − 2.4 4.9 4.4 − 11.2 6 50.5 52.9 2.4 100.7 108.6 7.9 5.1 5.7 12.9 7 63.7 60.4 − 3.3 104.0 107.8 3.8 6.6 6.5 − 1.7 8 57.7 53.1 − 4.6 85.4 86.4 1.0 4.9 4.6 − 6.9 9 71.1 72.1 1.0 75.3 84.4 9.1 5.4 6.1 13.7 10 68.0 61.3 − 6.7 85.0 81.8 − 3.2 5.8 5.0 − 13.3 11 48.5 58.3 9.8 90.3 88.7 − 1.5 4.4 5.2 18.2 12 58.0 62.0 4.0 71.1 76.3 5.2 4.1 4.7 14.7 Differences were calculated as: value scan 2 − value scan 1 A late diastolic filling, AUC area under the cure, CO cardiac output, E early diastolic filling, HR heart rate Subjects with marked high differences for the assessment of early filling parameters 1 3 The International Journal of Cardiovascular Imaging (2018) 34:905–920 919 1 3 Table 7 Subject specific scan–rescan assessment: information from the MV flow curves Subject E AUC E AUC Differ - E PFR 1 E PFR 2 Differ - E dur 1 E dur 2 Differ - A AUC A AUC Differ - A A Differ - A dur 1 A dur 2 Differ - 1 (mL) 2 (mL) ence E (mL/s) (mL/s) ence (ms) (ms) ence 1 (mL) 2 (mL) ence A  PFR 1 PFR 2 ence (ms) (ms) ence A AUC E PFR E dur AUC (mL/s) (mL/s) A PFR dur (ms) (mL) (mL/s) (ms) (mL) (mL/s) 1 78.6 94.8 16.2 522.8 738.8 216.0 332 292 − 40 28.1 29.3 1.2 342.1 357.7 15.6 166 175 9 2 84.3 71.3 − 13.0 567.4 388.4 − 179.0 323 413 90 19.1 14.1 − 5.0 243.5 159.6 − 83.9 144 188 44 3 85.7 96.4 10.7 744.5 734.7 − 9.8 309 270 − 39 32.8 28.0 − 4.8 409.5 347.0 − 62.5 172 194 22 4 68.5 61.4 − 7.1 551.4 537.9 − 13.5 316 263 − 53 15.3 12.4 − 2.9 161.8 158.0 − 3.8 190 131 − 59 5 52.6 46.8 − 5.8 343.6 362.2 18.6 348 297 − 51 16.7 15.0 − 1.7 187.0 176.3 − 10.7 174 208 34 6 83.9 84.6 0.7 500.2 545.2 45.0 378 328 − 50 13.6 22.0 8.4 137.4 231.3 93.9 189 197 8 7 79.3 87.5 8.2 683.6 647.7 − 35.9 314 351 37 32.3 29.6 − 2.7 409.2 374.7 − 34.5 157 160 3 8 72.2 71.6 − 0.6 592.5 543.6 − 48.9 384 350 − 34 17.5 17.5 0.0 221.1 217.1 − 4.0 175 176 1 9 59.4 66.8 7.4 506.1 563.6 57.5 272 300 28 18.3 19.6 1.3 243.7 245.0 1.3 136 137 1 10 58.7 58.9 0.2 367.5 376.7 9.2 322 373 51 17.6 14.4 − 3.2 185.3 173.6 − 11.7 162 170 8 11 64.5 60.9 − 3.6 398.4 363.9 − 34.5 349 343 − 6 17.5 14.4 − 3.1 181.8 172.3 − 9.5 209 205 − 4 12 52.8 47.4 − 5.4 380.3 314.4 − 65.9 330 313 − 17 8.5 12.9 4.4 114.5 164.0 49.5 165 187 22 Differences were calculated as: value scan 2 − value scan 1 A late diastolic filling, AUC area under the cure, CO cardiac output, dur duration, E early diastolic filling, HR heart rate, PFR peak filling rate Subjects with marked high differences for the assessment of early filling parameters 920 The International Journal of Cardiovascular Imaging (2018) 34:905–920 15. Westenberg JJ, Roes SD, Ajmone Marsan N, Binnendijk NM, References Doornbos J, Bax JJ, Reiber JH, de Roos A, van der Geest RJ (2008) Mitral valve and tricuspid valve blood flow: accurate 1. Elbaz MS, van der Geest RJ, Calkoen EE, de Roos A, Lelieveldt quantification with 3D velocity-encoded MR imaging with retro - BP, Roest AA, Westenberg JJ (2016) Assessment of viscous spective valve tracking. Radiology 249(3):792–800. https://doi. energy loss and the association with three-dimensional vortex ring org/10.1148/radiol.2492080146 formation in left ventricular inflow: in vivo evaluation using four- 16. Garg P, Westenberg JJM, van den Boogaard PJ, Swoboda PP, Aziz dimensional flow MRI. Magn Reson Med. https://doi.org/10.1002/ R, Foley JRJ, Fent GJ, Tyl FGJ, Coratella L, ElBaz MSM, van mrm.26129 der Geest RJ, Higgins DM, Greenwood JP, Plein S (2017) Com- 2. Pedrizzetti G, Domenichini F (2005) Nature optimizes the swirl- parison of fast acquisition strategies in whole-heart four-dimen- ing flow in the human left ventricle. Phys Rev Lett 95(10):108101. sional flow cardiac MR: two-center, 1.5 T, phantom and in vivo https://doi.org/10.1103/PhysRevLett.95.108101 validation study. J Magn Reson Imaging. https://doi.org/10.1002/ 3. Kamphuis VP, Westenberg JJM, van der Palen RLF, Blom NA, jmri.25746 de Roos A, van der Geest R, Elbaz MSM, Roest AAW (2017) 17. Klein S, Staring M, Murphy K, Viergever MA, Pluim JP (2010) Unravelling cardiovascular disease using four dimensional flow elastix: a toolbox for intensity-based medical image registration. cardiovascular magnetic resonance. Int J Cardiovasc Imaging IEEE Trans Med Imaging 29(1):196–205. https://doi.org/10.1109/ 33(7):1069–1081. https://doi.org/10.1007/s10554-016-1031-9 TMI.2009.2035616 4. Eriksson J, Bolger AF, Ebbers T, Carlhall CJ (2013) Four-dimen- 18. Agati L, Cimino S, Tonti G, Cicogna F, Petronilli V, De Luca L, sional blood flow-specific markers of LV dysfunction in dilated Iacoboni C, Pedrizzetti G (2014) Quantitative analysis of intraven- cardiomyopathy. Eur Heart J Cardiov Imaging 14(5):417–424. tricular blood flow dynamics by echocardiographic particle image https://doi.org/10.1093/ehjci/jes159 velocimetry in patients with acute myocardial infarction at differ - 5. Fredriksson AG, Svalbring E, Eriksson J, Dyverfeldt P, Alehagen ent stages of left ventricular dysfunction. Eur Heart J Cardiovasc U, Engvall J, Ebbers T, Carlhall CJ (2016) 4D flow MRI can Imaging 15(11):1203–1212. https://doi.org/10.1093/ehjci/jeu106 detect subtle right ventricular dysfunction in primary left ven- 19. Hyslop NP, White WH (2009) Estimating precision using dupli- tricular disease. J Magn Reson Imaging 43(3):558–565. https:// cate measurements. J Air Waste Manag Assoc 59(9):1032–1039 doi.org/10.1002/jmri.25015 20. Bland JM, Altman DG (1986) Statistical methods for assessing 6. Kanski M, Arvidsson PM, Toger J, Borgquist R, Heiberg E, Carls- agreement between two methods of clinical measurement. Lancet son M, Arheden H (2015) Left ventricular fluid kinetic energy 1(8476):307–310 time curves in heart failure from cardiovascular magnetic reso- 21. Bolger AF, Heiberg E, Karlsson M, Wigstrom L, Engvall J, Sig- nance 4D flow data. J Cardiovasc Magn Reson 17:111. https://doi. fridsson A, Ebbers T, Kvitting JP, Carlhall CJ, Wranne B (2007) org/10.1186/s12968-015-0211-4 Transit of blood flow through the human left ventricle mapped 7. Svalbring E, Fredriksson A, Eriksson J, Dyverfeldt P, Ebbers by cardiovascular magnetic resonance. J Cardiovasc Magn Reson T, Bolger AF, Engvall J, Carlhall CJ (2016) Altered diastolic 9(5):741–747. https://doi.org/10.1080/10976640701544530 flow patterns and kinetic energy in subtle left ventricular remod- 22. Mohiaddin RH (1995) Flow patterns in the dilated ischemic left eling and dysfunction detected by 4D flow MRI. PLoS ONE ventricle studied by MR imaging with velocity vector mapping. J 11(8):e0161391. https://doi.org/10.1371/journal.pone.0161391 Magn Reson Imaging 5(5):493–498 8. Fenster BE, Browning J, Schroeder JD, Schafer M, Podgorski CA, 23. Francis GS (2001) Pathophysiology of chronic heart failure. Am Smyser J, Silveira LJ, Buckner JK, Hertzberg JR (2015) Vorticity J Med 110(Suppl 7A):37S–46S is a marker of right ventricular diastolic dysfunction. Am J Physiol 24. Kanski M, Toger J, Steding-Ehrenborg K, Xanthis C, Bloch KM, Heart Circ Physiol 309(6):H1087–H1093. https://doi.org/10.1152/ Heiberg E, Carlsson M, Arheden H (2015) Whole-heart four- ajpheart.00278.2015 dimensional flow can be acquired with preserved quality with- 9. Schafer M, Browning J, Schroeder JD, Shandas R, Kheyfets out respiratory gating, facilitating clinical use: a head-to-head VO, Buckner JK, Hunter KS, Hertzberg JR, Fenster BE (2016) comparison. BMC Med Imaging 15:20. https://doi.org/10.1186/ Vorticity is a marker of diastolic ventricular interdependency s12880-015-0061-4 in pulmonary hypertension. Pulm Circ 6(1):46–54. https://doi. 25. Carlsson M, Heiberg E, Toger J, Arheden H (2012) Quantification org/10.1086/685052 of left and right ventricular kinetic energy using four-dimensional 10. Kheradvar AP, G (2012) Vortex formation in the heart. In: Vortex intracardiac magnetic resonance imaging flow measurements. formation in the cardiovascular system. Springer, London, p 19 Am J Physiol Heart Circ Physiol 302(4):H893–H900. https://doi. 11. Kheradvar A, Assadi R, Falahatpisheh A, Sengupta PP (2012) org/10.1152/ajpheart.00942.2011 Assessment of transmitral vortex formation in patients with dias- 26. Hussaini SF, Rutkowski DR, Roldan-Alzate A, Francois CJ (2016) tolic dysfunction. J Am Soc Echocardiogr 25(2):220–227. https:// Left and right ventricular kinetic energy using time-resolved ver- doi.org/10.1016/j.echo.2011.10.003 sus time-average ventricular volumes. J Magn Reson Imaging. 12. Stewart KC, Charonko JC, Niebel CL, Little WC, Vlachos PP https://doi.org/10.1002/jmri.25416 (2012) Left ventricular vortex formation is unaffected by diastolic 27. Grothues F, Smith GC, Moon JC, Bellenger NG, Collins P, Klein impairment. Am J Physiol Heart Circ Physiol 303(10):H1255– HU, Pennell DJ (2002) Comparison of interstudy reproducibility H1262. https://doi.org/10.1152/ajpheart.00093.2012 of cardiovascular magnetic resonance with two-dimensional echo- 13. Kamphuis VP, Roest AAW, Westenberg JJM, Elbaz MSM (2017) cardiography in normal subjects and in patients with heart failure Biventricular vortex ring formation corresponds to regions of or left ventricular hypertrophy. Am J Cardiol 90(1):29–34 highest intraventricular viscous energy loss in a Fontan patient: 28. van der Geest RJ, Reiber JH (1999) Quantification in cardiac MRI. analysis by 4D Flow MRI. Int J Cardiovasc Imaging. https://doi. J Magn Reson Imaging 10(5):602–608 org/10.1007/s10554-017-1250-8 14. Kamphuis VP, van der Palen RLF, de Koning PJH, Elbaz MSM, van der Geest RJ, de Roos A, Roest AAW, Westenberg JJM (2017) In-scan and scan–rescan assessment of LV in- and out-flow vol- umes by 4D flow MRI versus 2D planimetry. J Magn Reson Imag- ing. https://doi.org/10.1002/jmri.25792 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The International Journal of Cardiovascular Imaging Springer Journals

Scan–rescan reproducibility of diastolic left ventricular kinetic energy, viscous energy loss and vorticity assessment using 4D flow MRI: analysis in healthy subjects

Loading next page...
 
/lp/springer-journals/scan-rescan-reproducibility-of-diastolic-left-ventricular-kinetic-BHRY0kn6rG

References (34)

Publisher
Springer Journals
Copyright
Copyright © 2017 by The Author(s)
Subject
Medicine & Public Health; Cardiology; Imaging / Radiology; Cardiac Imaging
ISSN
1569-5794
eISSN
1573-0743
DOI
10.1007/s10554-017-1291-z
Publisher site
See Article on Publisher Site

Abstract

The aim of the current study was to assess the scan–rescan reproducibility of left ventricular (LV) kinetic energy (KE), viscous energy loss (EL) and vorticity during diastole from four-dimensional flow magnetic resonance imaging (4D flow MRI) in healthy subjects. Twelve volunteers (age 27 ± 3 years) underwent whole-heart 4D flow MRI twice in one session. In-scan consistency was evaluated by correlation between KE and EL. EL was computed to measure the amount of EL index relative to KE over diastole. Scan–rescan analysis was performed to test reproducibility of volumetric measurements of KE, EL, EL and vorticity in the LV over early (E) and late (A) diastolic filling. In-scan consistency between KE and EL was index strong-excellent (E-filling scan1: r = 0.92, P < 0.001; scan2: ρ = 0.96, P < 0.001 and A-filling scan1: ρ = 0.87, P < 0.001; scan2: r = 0.99, P < 0.001). For the majority of subjects (10 out of 12), KE and EL measures showed good to strong reproducibility. However, with a wide range of agreement [intraclass correlation (ICC): 0.64–0.95] and coefficients of variation (CV) ≤ 25%. EL showed strong reproducibility for all 12 subjects with a strong ICC (0.94, P < 0.001) and a CV of 9%. Scan–rescan index reproducibility of volumetric vorticity showed good–excellent ICCs (0.83–0.95) with CVs ≤ 11%. In conclusion, the current study shows strong–excellent in-scan consistency and overall good agreement between scans for 4D flow MRI assessment of left ventricular kinetic energy, energy loss and vorticity over diastole. However, substantial differences between the scans were also found in some parameters in two out of twelve subjects. Strong reproducibility was found in the dimensionless EL , which measures the amount of viscous energy loss relative to the average kinetic energy over diastole. index Keywords 4D flow MRI · Magnetic resonance imaging · Kinetic energy · Viscous energy loss · Vorticity Introduction imaging and analysis techniques are needed to evaluate volu- metric changes in such complex hemodynamic parameters. Congenital and acquired heart diseases affect the efficacy Four-dimensional flow magnetic resonance imaging (4D of intracardiac flow patterns and energy distribution [1 , flow MRI) allows for comprehensive non-invasive assess- 2]. Given the three-dimensional (3D) time-varying nature ment of 3D time-varying blood flow properties in the heart of these flow patterns and energetics, specialized in vivo and great vessels in all three velocity encoding directions and spatial dimensions over the cardiac cycle [3]. 4D flow MRI has recently emerged as a novel tool * Vivian P. Kamphuis for in vivo quantification of intracardiac f low energetics, v.p.kamphuis@lumc.nl associated energy losses and vortical flow patterns by means of kinetic energy (KE) [4–7], viscous energy loss Division of Pediatric Cardiology, Department of Pediatrics, Leiden University Medical Center, Leiden, The Netherlands (EL) [1] and vorticity [8, 9]. KE is the energy contained in the flow of the bloodstream due to motion and EL is Netherlands Heart Institute, Utrecht, The Netherlands 3 the KE that is irreversibly lost due to viscosity-induced Department of Radiology, Leiden University Medical Center, frictional forces within the blood flow [1]. In acquired Leiden, The Netherlands 4 heart disease, remodelling occurs which can lead to alter- Division of Pediatric Cardiology, Department of Pediatrics, ations in intracardiac hemodynamics [4]. Alteration in Academic Medical Center, Amsterdam, The Netherlands Vol.:(0123456789) 1 3 906 The International Journal of Cardiovascular Imaging (2018) 34:905–920 intraventricular KE derived from 4D f low MRI has been Cardiovascular magnetic resonance acquisition used to assess left ventricular (LV) and right ventricular and data preparation (RV) (dys)function in patients with different stages of heart failure (HF) [4–7]. This is also the case in various Whole-heart 4D flow MRI was obtained on a 3 T scan- congenital heart diseases. Even after correction, patients ner (Ingenia, Philips Medical Systems, The Netherlands) may develop systolic and/or diastolic dysfunction, leading with maximal amplitude of 45 mT/m for each axis, slew to changes in intracardiac flow energetics [1]. In addition, rate of 200  T/m/s and a combination of FlexCoverage intracardiac anatomy may not be restored. In corrected Posterior coil in the table top with a dStream Torso coil, atrioventricular septal defect patients, elevated EL was providing up to 32 coil elements for signal reception. The associated with altered 3D vortex ring formation in the orientation of the acquisition of 4D flow data was identi- LV filling pattern [1]. Vorticity, the curl of velocity, is a cal to the 4-chamber orientation (usually double-oblique fundamental quantity in fluid mechanics that describes axial or coronal). Velocity-encoding of 150 cm/s in all the local spinning rate of fluid particles and can charac- three directions was used in a standard four-point encod- terize vortex flow [10]. Quantitative vortex parameters, ing scheme, spatial resolution 3.0 × 3.0 × 3.0  mm , field- such as vorticity, have been used to assess diastolic (dys) of-view 400 mm, flip angle 10°, echo time (TE) 3.7 ms, function in several patient groups [8, 9, 11, 12]. Further- repetition time (TR) 10  ms, true temporal resolution more, in patients with complex congenital intracardiac 40 ms, SENSitivity Encoding (SENSE) factor 2 in ante- deformations such as after the Fontan operation, flow rior–posterior direction and Echo Planar Imaging (EPI) collision with remaining septal structures and stagnation readout with a factor 5. Free breathing was allowed and of flow through a ventricular septal defect may result in no respiratory motion compensation was performed. Ret- altered EL and vortex formation [13]. rospective gating was used with 30 phases reconstructed Recently, good reproducibility of inf low- and outf low- to represent one cardiac cycle. Expected scan-time for assessment from 4D flow MRI was shown [14]. However, the 4D flow MRI acquisition for a patient with a heart there is a lack of studies validating the reproducibility of rate of 60 bpm and 39 slices would be 9 min and 11 s. intracardiac energy and quantitative vorticity parameters This 4D flow MRI sequence with EPI readout has been from 4D f low MRI in a scan–rescan setting. Scan–rescan validated in vivo and in vitro [15] and compared to other reproducibility is important for clinical applicability as 4D flow MRI sequences [16]. Concomitant gradient cor - it expresses the reliability in repeated quantitation, for rection was performed from standard available scanner example during serial follow-up or in case of a rest–stress software. Cine two-dimensional (2D) left 2-chamber, protocol. Therefore, the aim of this study was to assess 4-chamber, coronal and sagittal aorta views and a cine the scan–rescan reproducibility of 4D flow MRI measure- multi-2D short-axis stack of slices were acquired, using ments of kinetic energy, viscous energy loss and vorticity steady-state free-precession (SSFP) sequences with TE/TR within the LV during diastolic filling in healthy subjects. 1.5/3.0, 350 mm field-of-view, 45° flip angle, acquisition resolution 1.9 × 2.0 × 8.0  mm . Retrospective gating was used with 30 phases reconstructed to represent one cardiac cycle. Expected scan-time for the cine multi-2D short-axis Materials and methods acquisition for a patient with a heart rate of 60 bpm and 12 slices would be 1 min. Free breathing was allowed without Study population using motion suppression, three signal averages were taken to minimize effects of breathing motion, which makes the The study protocol was approved by the local Medical expected scan-time 3 min. Image analysis was performed Ethical Committee of the Leiden University Medical by one observer (VPK) with over 2 years of experience Center and informed consent was obtained from all par- in MRI and verified by one observer (JJMW) with over ticipants. Twelve healthy volunteers with no history of 15 years of experience in MRI. The endocardial border cardiac disease were included. All subjects underwent an was manually traced in all slices and phases in the multi- MRI scan including whole-heart 4D flow MRI between slice 2D cine short-axis images and ventricular volume July 2015 and April 2017. The same scanning protocol was calculated at the end-diastolic and end-systolic phases was performed twice in the same session with 10-min using in-house developed MASS software. Papillary mus- breaks between the scans and repositioning and replan- cles were disregarded and assumed to be included in the ning for every volunteer. Ten of these volunteers were ventricular volume. LV in- and outflow was assessed using included in a recent study [14]. That study did not report the 4D flow MRI data with retrospective valve tracking assessment of KE, EL or vorticity. of the mitral and aortic valve, as shown in a recent study 1 3 The International Journal of Cardiovascular Imaging (2018) 34:905–920 907 [14]. Cardiac output (CO) was computed from the 4D flow amount of kinetic energy lost to that available over cardiac data as LV outflow × Heart rate (HR). Beginning and end- cycle. In this study, EL was computed over diastole as index ing of diastolic phases [early diastolic filling (E-filling) EL /KE , with KE being the total diastole average diastole average diastole and late diastolic filling (A-filling)] were derived from the average KE during total diastole. In order to compare our mitral valve flow-time curves that resulted from retrospec- results with a previous study reporting EL values normalized tive valve tracking. Segmentation of the LV cavity in the by stroke volume (SV) [1], we also report EL values in the 4D flow MRI acquisition, that is required for the energy current study as normalized by SV (reported as norm_ , E-avg EL ∙ ∙ ∙ analysis, was obtained by transforming the available norm_ , norm_ , norm_ and norm_ E-peak A−avg A−peak EL EL EL time-varying segmentation of multi-slice cine short-axis EL ). To be consistent with the previous study SV total diastole acquisition to the 4D flow MRI data. To correct for patient was derived from cine SSFP short-axis slices. motion related misalignment between the two acquisi- tions, automated image-based 3D rigid registration was Integral vorticity magnitude over LV diastole performed using the phase with optimal depiction of the (vorticity_LV) LV cavity in both scans with the Elastix image registration toolbox [17]. Kinetic energy, viscous energy loss and vor- The formulae that were used to calculate the integral vor- ticity analysis of segmented LV volumes was done by one ticity magnitude are shown in “Appendix 1”. In short, fol- investigator (MSME) using in-house developed software. lowing previously published work [8, 9], for each acquired time-phase, voxel-wise vorticity magnitude (1/s) was first KE analysis over LV diastole computed over the segmented LV volume. Then, the instan- taneous integral vorticity magnitude was computed as the The amount of KE during diastolic filling was computed cumulative sum of voxel-wise vorticity and multiplied following previously published methods [1]. KE for each by voxel volume to give the integral in (milliliter × 1/sec- voxel within the LV was computed as 1/2 m v , with (m) ond) i.e. (mL/s). Note that the computed vorticity integral as the mass representing the voxel volume multiplied by parameter is a scalar quantity and therefore does not take the density of blood (1.025 g/mL) and (v) as the 3-direc- the vorticity direction into account. We will refer to this tional velocity from 4D flow MRI. For each acquired time- vorticity integral over the LV as vorticity_LV throughout phase, volumetric KE was then computed by integrating (by the text, to differentiate it from voxel-wise vorticity. In order cumulative sum) the computed KE over the segmented 3D to quantify the integral vorticity_LV over diastolic filling, LV volume. In order to quantify KE during diastolic filling, the time-average and peak vorticity_LV during E-filling the time-averaged KE during diastolic phases (KE and (vorticity_LV, vorticity_LV , respectively) and E-avg E-avg E-peak KE ) and peak KE (KE and KE ), in Joule, A-filling (vorticity_LV, voriticity_LV , respec- A−avg E-peak A−peak A−avg A−peak were assessed. tively) were computed. Viscous EL analysis over LV diastole Scan–rescan analysis Following recently published methods [1], we have computed For the scan–rescan analysis, all data was blinded by one EL from 4D flow MRI using the dissipation terms from the observer (VPK) and presented in a random order to the Navier–Stokes energy equations, assuming blood as a New- observer (MSME) that performed the energy and vorticity tonian fluid. The formulae that were used to calculate EL analysis. Scan–rescan analysis was performed to test the are summarized in “Appendix 1”. The time-averaged viscous reproducibility of (1) KE over E-filling and A-filling; (2) energy loss rate [ EL, in Watt (W)] during E-filling ( ) EL over E-filling, A-filling and total diastolic filling and (3) E-avg EL ∙ ∙ ∙ ∙ vorticity_LV over E-filling and A-filling. and A-filling ( ) and peaks ( and ) A−avg E-peak A−peak EL EL EL EL were assessed. We have used the previously reported correla- Statistical analysis tion between KE and [1] as a measure of in-scan consist- EL ency. Furthermore, EL over the total diastole (EL ) total diastole Data analysis was performed using SPSS Statistics software in Joule (J) was computed. Given that the amount of viscous (v. 23.0 IBM SPSS, Chicago, IL). Variables were tested for energy lost is proportional to the amount of kinetic energy, normal distribution using the Shapiro–Wilk test. Continu- we computed a dimensionless energy loss parameter, EL , index ous data was expressed as mean ± standard deviation (SD) that reflects the amount of viscous energy loss indexed for with minimum and maximum values or as median [inter- the average kinetic energy over diastole. EL is a dimen- index quartile range] where suitable. Mean differences were deter - sionless index that was used in an earlier echo particle image mined for inter-scan comparison and significance was tested velocimetry study by Agati et al. [18] to indicate the relative by a paired samples t test or, in case of non-normality, the 1 3 908 The International Journal of Cardiovascular Imaging (2018) 34:905–920 Wilcoxon signed-rank test. Differences were computed as: correlation (scan 1: r = 0.92, P < 0.001; scan 2: ρ = 0.96, measurement in scan 2—measurement in scan 1. The coef- P < 0.001). Also, correlation between KE and A−avg ficient of variation (CV) was calculated with the root mean was strong–excellent in both scans (scan 1: ρ = 0.87, A−avg EL square method [19]. Correlation between the in-scan and P < 0.001; scan 2: r = 0.99, P < 0.001). inter-scan measurements done in repeated scans was tested by the Pearson correlation coefficient (r), or the Spearman Scan–rescan analysis of kinetic energy and viscous correlation coefficient (ρ) in case of non-normality of the energy loss rate over early diastolic filling data. The approach described by Bland and Altman [20] was used to study systematic differences between measurements Detailed results of the scan–rescan tests of KE and EL obtained from the two scans. Agreement between these meas- assessment over early diastolic filling are shown in Tables  2 urements was assessed by determining the intra-class correla- and 3 and Fig.  3. Scan–rescan assessment showed poor tion (ICC) coefficient. Correlation and agreement were clas- results for early diastolic filling, as shown in Table  2. Of sified as follows: r/ρ and ICC > 0.95: excellent, 0.95–0.85: note, Fig. 3 shows that for KE and assessment over early EL strong, 0.85–0.70: good, 0.70–0.5: moderate, < 0.5: poor. A diastolic filling two subjects showed more distinct differ - P value < 0.05 was considered statistically significant. ences. Throughout the text we will refer to these two sub- jects as “Subject 1” and “Subject 2”. Detailed scan–rescan information of these two subjects is provided in “Appen- Results dix  2”. In Fig.  3a, Subject 1 and 2 are indicated as dark (KE ) and light red triangles (KE ) and in Fig. 3b, as E-avg E-peak Volunteer characteristics are shown in Table 1. Heart rate and ∙ ∙ dark ( ) and light red triangles ( ). E-avg E-peak cardiac output were not significantly different between the two EL EL Possibly some factors related to the 4D flow MRI acqui- scans (HR: 60.8 ± 7.8 vs. 59.9 ± 6.9  bpm, P = 0.52 and CO: sition or physiological factors have resulted in these marked 5.5 ± 0.9 vs. 5.6 ± 1.3  L/min, P = 0.75). 4D flow MRI data differences, therefore we performed an evaluation for a sub- acquisition was successful in all volunteers. Figure 1a–c shows cohort without Subject 1 and 2. Scan–rescan correlations were cross-sectional mapping of the volumetric measurements of much stronger and variation was less for this sub-cohort, as KE, EL and vorticity inside the LV at peak E-filling in a stand- shown in Table  3. KE and assessment showed E-avg E-avg ard 4-chambers view. An example of the temporal evolution of EL non-significant differences between the two scans, strong KE and over LV diastole is shown in one subject in Fig. 1d. EL ∙ ICCs (KE : 0.95, P < 0.001 and : 0.91, P = 0.03) and E-avg E-avg EL ∙ ∙ CVs ≤ 11% (KE : 10% and : 11%). norm_ Intra-scan comparison of energetics (KE vs. EL) E-avg E-avg E-avg EL EL showed a strong ICC (0.90, P = 0.001) and a CV of 12%. Scan–rescan assessment of KE and showed non- Figure 2 shows the in-scan comparison of KE versus EL . E-peak E-peak EL significant differences. The ICC of KE was good (0.82, Comparison of KE to showed strong–excellent E-peak E-avg E-avg EL Table 1 Baseline characteristics Scan-independent characteristics N 12 Male (%) 6/12 (50%) Age (years) 27 ± 3 Height (cm) 175 ± 7 Weight (kg) 69 ± 12 BSA (m ) 1.8 ± 0.2 BMI (kg/m ) 22 ± 3 Characteristics per scan Scan 1 Scan 2 P value HR (bpm) 60.8 ± 7.8 59.9 ± 6.9 0.52 End-diastolic volume (mL) 143.8 (132.4–183.8) 158.6 ± 30.3 0.64 End-systolic volume (mL) 59.0 ± 12.8 56.2 ± 12.9 0.27 Stroke volume (mL) 100.8 ± 21.8 102.4 ± 21.0 0.44 Ejection fraction (%) 62.8 ± 3.6 63.1 (61.6–65.4) 0.21 CO (L/min) 5.5 ± 0.9 5.6 ± 1.3 0.75 bpm beat per minute, BMI Body Mass Index, BSA body surface area, CO cardiac output, HR heart rate, LV left ventricular 1 3 The International Journal of Cardiovascular Imaging (2018) 34:905–920 909 Fig. 1 Maps of left ventricular kinetic energy (KE), viscous energy ventricular viscous energy loss rate at peak early diastolic filling, c loss rate ( ) and vorticity over the LV of a healthy female subject left ventricular voxel-wise vorticity at peak early diastolic filling, d EL (age 20  years) in a standard 4-chambers MRI cross-sectional view, temporal evolution of volumetric kinetic energy, viscous energy loss a left ventricular kinetic energy at peak early diastolic filling, b left rate and vorticity over LV diastole now within the same range as the other values [Fig. 3c, indi- P = 0.01) with a CV of 15%. Similarly the ICC of E-peak EL cated as dark (KE ) or light red triangles (KE )]. was good (0.76, P = 0.03) with a CV of 17%. Furthermore, A−avg A−peak ∙ ∙ The same was observed when assessing and norm_ showed a strong ICC (0.86, P = 0.005) and a A−avg A−peak EL EL E-peak EL [Fig. 3d, indicated as dark ( ) or light red triangles CV of 16%. A−avg EL ( )]. However, to be consistent we repeated the A−peak EL evaluation in the sub-cohort without Subject 1 and Subject Scan–rescan analysis of kinetic energy and viscous 2. Scan–rescan correlations and variation were similar for energy loss rate over late diastolic filling the sub-cohort without Subject 1 and Subject 2, as shown in Table 3. Reproducibility of KE and assessment Detailed results of the scan–rescan tests of KE and assess- A−avg A−avg EL EL showed non-significant differences, good ICCs (KE : ment over late diastolic filling are shown in Tables  2 and 3 A−avg 0.77, P = 0.02 and : 0.75, P = 0.03) and CVs up and Fig. 3. Scan–rescan assessment showed good results for A−avg EL ∙ ∙ to 24% (KE : 23% and : 24%). norm_ late diastolic filling, as shown in Table  2. Subject 1 and 2 are A−avg A−avg A−avg EL EL 1 3 910 The International Journal of Cardiovascular Imaging (2018) 34:905–920 norm_EL ) and EL are shown in Tables 2 and total diastole index 3 and Fig. 3. Scan–rescan assessment showed poor results for total diastole, but strong results for EL , as shown in index Table 2. Scan–rescan correlations were much stronger and variation was less for the sub-cohort without Subject 1 and Subject 2 (Table 3). Scan–rescan assessment of EL total diastole showed a non-significant difference between the scans, a strong ICC of 0.91 (P = 0.001) and a CV of 11%. norm_ EL showed a non-significant difference between total diastole the scans, a good ICC (0.81, P = 0.01) and a CV of 12%. Figure 3f shows the Bland–Altman plot of EL . For all index subjects, scan–rescan assessment of EL showed excel- index lent reproducibility with a small non-significant difference between the scans, a strong ICC of 0.94 (P < 0.001) and a CV of 9%. When evaluating the sub-cohort without Subject 1 and 2 the results remained similar (ICC: 0.95, P < 0.001 and CV: 8%). Scan–rescan analysis of volumetric vorticity_LV over early diastolic filling Detailed results of the scan–rescan tests of vorticity_LV assess- ment over early diastolic filling are shown in Tables  4 and 5 and Fig. 4. Scan–rescan assessment showed moderate results for early diastolic filling as shown in Table  4. The Bland–Alt- man plots of the assessment of vorticity_LV and vorticity_ E-avg LV showed higher differences between scan and rescan E-peak measurements for Subject 1 and 2 (Fig. 4a, indicated as dark (vorticity_LV ) or light red triangles (vorticity_LV )). E-avg E-peak Scan–rescan correlations were much stronger and variation was less for the sub-cohort without Subject 1 and 2 (Table 5). Fig. 2 Scatter plots of kinetic energy (KE) versus viscous energy ∙ ∙ Scan–rescan assessment of v orticity_LV and vorticity_ E-peak loss rate ( ) during early diastolic filling and KE versus during EL EL LV showed a non-significant difference between the late diastolic filling. a Scatter plot depicting the correlation between E-avg KE and measured in scan 1 (grey) and scan 2 (black), scans, good-strong ICCs (vorticity_LV : 0.83, P = 0.01 E-avg ELE-avg E-peak in all 12 subjects, b scatter plot depicting the correlation between and vorticity_LV : 0.95, P < 0.001) and CVs up to 11% E-avg KE and measured in scan 1 (grey) and scan 2 (black), A−avg ELA−avg (vorticity_LV : 11% and vorticity_LV : 7%). E-peak E-avg in all 12 subjects Scan–rescan analysis of volumetric vorticity_LV showed a non-significant difference between the scans, a over late diastolic filling good ICC (0.70, P = 0.048) and a CV of 22%. Scan–rescan assessment of KE and showed non-signifi- A−peak A−peak EL Detailed results of the scan–rescan tests of vorticity_LV cant differences, good ICCs (KE : 0.83, P = 0.01 and A−peak assessment over late diastolic filling are shown in Tables  4 : 0.79, P = 0.02) and CVs up to 25% (KE : A−peak A−peak EL and 5 and Fig.  4. Scan–rescan assessment showed good 23% and : 25%). Lastly, scan–rescan assessment of A−peak EL results for late diastolic filling as shown in Table  4. Subject norm_ showed a non-significant difference between A−peak EL 1 and 2 are now within the same range as the other values the scans, a moderate ICC (0.64, P = 0.08) and a CV of 24%. (Fig. 4b, indicated as dark (vorticity_LV ) and light red A−avg triangles (vorticity_LV )). Scan–rescan variation was A−peak Scan–rescan analysis of kinetic energy and viscous similar for the sub-cohort without Subject 1 and 2 (Table 5). energy loss over total diastole Scan–rescan assessment of v orticity_LV and vorticity_ A−peak LV showed non-significant differences between the A−avg Detailed results of the scan–rescan tests of KE and scans, good-strong ICCs (vorticity_LV : 0.91, P = 0.001 A−peak assessment over total diastole (EL and and vorticity_LV : 0.89, P = 0.002) and CVs of 11%. total diastole A−avg EL 1 3 The International Journal of Cardiovascular Imaging (2018) 34:905–920 911 1 3 Table 2 Scan–rescan comparison of energy variables for the cohort with all subjects (N = 12) Scan 1 Scan 2 Difference (scan 2–scan 1) Pearson correlation Intraclass correlation coef- Coefficient of coefficient ficient variation (%) Mean ± SD (min–max) or Mean ± SD (min–max) or Mean ± SD or (min–max) ICC (95% CI) median [IQR] (min–max) median [IQR] (min–max) median [IQR] (min–max) Early filling  KE (mJ) 4.9 ± 1.9 (2.6–9.6) 4.8 ± 1.4 (2.0–7.6) 0.1 ± 2.3 (− 4.8 to 5.0) 0.06 0.12 (− 2.78 to 0.76) 28 E-peak  KE (mJ) 1.9 ± 0.7 (0.8–3.5) 1.9 [1.3–2.3] (1.1–4.0) − 0.01 [− 0.1 to 0.3] (− 2.2 0.52 0.31 (− 1.85 to 0.81) 26 E-avg to 2.1) 1.5 ± 0.4 (0.8–2.1) 1.5 ± 0.6 (0.7–2.9) 0.04 ± 0.7 (− 1.4 to 1.7) 0.02 0.04 (− 3.25 to 0.74) 31   (mW) EL E-peak 3 15.2 ± 5.1 (9.4–24.9) 14.9 ± 5.1 (6.0–21.8) − 0.2 ± 6.2 (− 12.6 to 12.3) 0.27 0.45 (− 1.17 to 0.85) 30  norm_ (W/m ) EL E-peak 0.7 ± 0.2 (0.3–1.1) 0.7 ± 0.3 (0.4–1.4) − 0.01 [− 0.1 to 0.3] (− 0.7 0.11 0.21 (− 2.36 to 0.78) 28   (mW) ELE-avg to 0.8) 7.0 ± 2.3 (4.2–11.4) 6.6 ± 2.1 (3.2–10.4) − 0.4 ± 2.8 (− 6.4 to 5.8) 0.20 0.35 (− 1.55 to 0.82) 28  norm_ (W/m ) EL E-avg Late filling c c b b  KE (mJ) 0.9 [0.7–2.0] (0.7–2.5) 1.0 [0.9–1.6] (0.6–2.6) 0.1 [− 0.2 to 0.3] (− 1.2 to 0.62 0.87 (0.55–0.96) 23 A−peak 0.5) c b b  KE (mJ) 0.6 [0.5–1.2] (0.4–1.4) 0.8 ± 0.4 (0.4–1.6) − 0.03 ± 0.3 (− 0.7 to 0.4) 0.68 0.84 (0.42–0.95) 23 A−avg b b 0.5 ± 0.3 (0.3–1.1) 0.5 ± 0.2 (0.3–1.0) − 0.007 ± 0.2 (–0.5 to 0.3) 0.67 0.81 (0.33–0.95) 24   (mW) ELA−peak ∙ b 5.3 ± 1.8 (2.5–8.3) 5.1 ± 1.5 (3.3–7.9) − 0.2 ± 1.7 (− 3.2 to 2.2) 0.50 0.68 (− 0.19 to 0.91) 22  norm_ (W/m ) ELA−peak ∙ c b 0.3 [0.2–0.6] (0.2–0.7) 0.4 ± 0.2 (0.2–0.8) − 0.02 ± 0.1 (− 0.4 to 0.2) 0.50 0.80 (0.29–0.94) 23   (mW) EL A−avg b b 3.7 ± 1.3 (2.1–6.6) 3.4 ± 1.1 (2.2–5.2) − 0.3 ± 1.1 (− 2.8 to 1.5) 0.58 0.73 (0.07–0.92) 21  norm_ (W/m ) EL A−avg Total diastole c b  EL (mJ) 0.5 ± 0.2 (0.3–1.0) 0.5 [0.4–0.6] (0.3–1.3) − 0.006 [− 0.07 to 0.06] 0.81 0.58 (− 0.60 to 0.88) 22 total diastole (− 0.4 to 0.7) 3 c  norm_EL (J/m ) 5.3 ± 1.5 (3.4–8.1) 4.8 [4.2–5.3] (3.1–9.8) − 0.1 ± 2.2 (− 3.9 to 5.3) 0.30 0.17 (− 2.52 to 0.77) 23 total diastole a a  EL 0.3 ± 0.07 (0.2–0.4) 0.3 ± 0.06 (0.2–0.4) − 0.001 ± 0.03 (− 0.04 to 0.89 0.94 (0.80–0.98) 9 index 0.06) KE kinetic energy, E early diastolic filling, A late diastolic filling, EL viscous energy loss P < 0.001 P < 0.05 P < 0.05 for non-normality (Shapiro–Wilk test) 912 The International Journal of Cardiovascular Imaging (2018) 34:905–920 1 3 Table 3 Scan–rescan comparison of energy variables for the sub-cohort without Subject 1 and 2 (N = 10) Scan 1 Scan 2 Difference (scan 2–scan 1) Pearson correlation Intraclass correlation coef- Coefficient of coefficient ficient variation (%) Mean ± SD (min–max) or Mean ± SD (min–max) or Mean ± SD or (min–max) ICC (95% CI) median [IQR] (min–max) median [IQR] (min–max) median [IQR] (min–max) Early filling c b  KE (mJ) 5.0 [4.0–5.3] (2.0–5.4) 4.6 ± 1.1 (2.6–6.4) 0.1 ± 0.9 (− 0.9 to 2.0) 0.62 0.82 (0.27–0.96) 15 E-peak a a  KE (mJ) 1.7 ± 0.5 (0.8–2.3) 1.8 ± 0.5 (1.1–2.4) 0.05 ± 0.2 (− 0.2 to 0.5) 0.91 0.95 (0.82–0.99) 10 E-avg 1.5 ± 0.4 (0.8–2.1) 1.5 ± 0.4 (0.7–2.2) 0.01 ± 0.4 (− 0.5 to 0.9) 0.58 0.76 (− 0.07 to 0.94) 17   (mW) ELE-peak ∙ b b 15.4 ± 5.2 (9.6–24.9) 15.2 ± 4.2 (8.9–21.8) − 0.3 ± 3.5 (− 4.7 to 6.9) 0.75 0.86 (0.40–0.97) 16  norm_ (W/m ) EL E-peak ∙ b b 0.7 ± 0.2 (0.3–0.9) 0.6 ± 0.2 (0.4–0.9) − 0.02 ± 0.1 (− 0.2 to 0.2) 0.83 0.91 (0.64–0.98) 11   (mW) EL E-avg b b 7.0 ± 2.3 (4.2–11.4) 6.6 ± 1.6 (4.2–9.0) − 0.4 ± 1.2 (− 2.4 to 1.6) 0.87 0.90 (0.60–0.97) 12  norm_ (W/m ) EL E-avg Late filling c b  KE (mJ) 0.9 [0.7–1.7] (0.7–2.5) 1.2 ± 0.5 (0.6–2.4) 0.2 [− 0.1 to 0.3] (− 1.2 to 0.59 0.83 (0.25–0.96) 23 A−peak 0.5) c b b  KE (mJ) 0.5 [0.5–1.2] (0.4–1.4) 0.7 ± 0.3 (0.4–1.3) − 0.02 ± 0.3 (− 0.7 to 0.4) 0.65 0.77 (0.02–0.94) 23 A−avg ∙ c b 0.4 [0.3–0.7] (0.3–1.1) 0.5 ± 0.2 (0.3–0.9) − 0.03 ± 0.2 (− 0.5 to 0.3) 0.50 0.79 (0.11–0.95) 25   (mW) ELA−peak 5.3 ± 2.0 (2.5–8.3) 5.0 ± 1.3 (3.3–7.0) − 0.3 ± 1.7 (− 3.2 to 2.2) 0.49 0.64 (− 0.56 to 0.91) 24  norm_ (W/m ) EL A−peak c b 0.3 [0.2–0.6] (0.2–0.8) 0.3 ± 0.1 (0.2–0.6) − 0.03 ± 0.1 (− 0.4 to 0.2) 0.42 0.75 (− 0.05 to 0.94) 24   (mW) EL A−avg 3.6 ± 1.4 (2.1–6.6) 3.4 ± 1.0 (2.3–5.2) − 0.3 ± 1.2 (− 2.8 to 1.5) 0.57 0.70 (− 0.21 to 0.93) 22  norm_ (W/m ) EL A−avg Total diastole b b  EL (mJ) 0.5 ± 0.2 (0.3–1.0) 0.5 ± 0.1 (0.3–0.7) − 0.01 [− 0.05 to 0.03] (− 0.3 0.88 0.91 (0.64–0.98) 11 total diastole to 0.08) 3 b b  norm_EL (J/m ) 5.1 ± 1.3 (3.4–7.0) 4.8 ± 0.9 (3.1–6.4) − 0.3 ± 0.9 (− 2.0 to 1.2) 0.73 0.81 (0.27–0.95) 12 total diastole a a  EL 0.3 ± 0.06 (0.2–0.4) 0.3 ± 0.07 (0.2–0.4) − 0.01 ± 0.03 (− 0.04 to 0.06) 0.89 0.95 (0.79–0.99) 8 index KE kinetic energy, E early diastolic filling, A late diastolic filling, EL viscous energy loss *Subjects with measurements outside the 95% confidence interval of the respective Bland–Altman plots P < 0.001 P < 0.05 P < 0.05 for non-normality (Shapiro–Wilk test) The International Journal of Cardiovascular Imaging (2018) 34:905–920 913 filling rate and the area-under-the-curve are sensitive to tem- Discussion poral and spatial resolution. These observations are important to take into consideration when evaluating flow energetics and In the current study, kinetic energy, viscous energy loss and vorticity in a research or clinical setting. vorticity inside the left ventricle during diastole are derived from 4D flow MRI and scan–rescan reproducibility of these Kinetic energy (KE) over LV diastole parameters is tested. Scan–rescan reproducibility is essential for the clinical application of a parameter since it reflects Multiple 4D flow MRI studies showed that patients with LV the reliability of a measurement and feasibility of repeated dysfunction present altered flow patterns through the LV with measurement evaluations. The main findings of this study impaired preservation of inflow KE to the end of diastole and are: (1) internal consistency between kinetic energy and vis- altered KE-time curves (the amount of KE inside the LV cous energy loss is strong-excellent in both scans during during each time step over the total cardiac cycle) [4, 6, 7, early and late diastolic filling; (2) In the majority (10 out of 21], even in patients with normal to mild LV remodeling and 12) of subjects, reproducibility of peak and average kinetic normal to mildly depressed LV systolic function [7]. These energy and viscous energy loss during early, late and total KE changes in the LV could be a valuable diagnostic marker diastolic filling shows non-significant differences but with a to evaluate diastolic function and might be useful for early good to excellent agreement (by means of ICC) and CVs up detection of deteriorating ventricular function [1, 4–7, 21, to 25%; (3) In the studies parameters, time-averaged meas- 22], which could reduce patient morbidity and mortality [23]. urements over E- and A-filling show stronger reproducibility However, there is a lack of studies validating LV KE derived than peak measurements. (4) For all subjects, EL shows index from 4D flow MRI in a scan–rescan setting. Therefore, reli- good reproducibility with a small non-significant difference ability and reproducibility of KE measurements from 4D flow between the scans, strong agreement and a CV of 9%; (5) MRI in a repeated scan setting remains largely unknown. Assessment of volumetric vorticity over the left ventricle Kanski et al. [24] compared mean KE and peak KE between during early and late diastolic filling shows non-significant two scans (with and without respiratory gating) with the aim differences, good–excellent ICCs and CVs up to 11%. to evaluate the impact of respiratory gating on KE measure- In the Bland–Altman plots for E–filling parameters, meas- ments. They found a strong correlation between the KE meas- urements obtained in two subjects (Subject 1 and 2 in Appen- urements in both scans. The current study differs from the dix 2) showed distinct higher differences between scan and study by Kanski et al. [24] in that we used the same protocol rescan measurements than the measurements of the majority for both scans. In the current study, moderate–strong corre- of the cohort. These high differences had an impact on the lation was found between KE measurements in both scans. scan–rescan reproducibility for KE and measures. This EL Absolute values of KE and KE reported in the cur- impact seemed to be largest during early diastolic filling, where E-peak A−peak rent study are in agreement with previous studies [1, 25, 26]. reproducibility was much higher for the sub-cohort without In the current study we showed the reproducibil- Subject 1 and 2. However, during late diastolic filling, repro- ity of the KE measurements assessed with 4D flow MRI ducibility was similar. There is no obvious explanation for the with good–strong agreement (by means of ICCs), how- measurements that showed distinct higher die ff rences between ever substantial CVs up to 23% were also found. Based scan and rescan measurement, but this could be related to the on the Bland–Altman analysis the variability of K E technical restrictions of the 4D flow MRI acquisition or to phys- E-avg and KE was less than the variability of KE and iological differences between scans, or a combination of both. A−avg E-peak KE as shown by the smaller limits of agreement. This It is to be expected that physiological differences were small A−peak might be expected, given that the definition of average is as all subjects are healthy volunteers that were scanned twice computed over multiple time points and therefore evens under the same circumstances with only a short break between out the variations more than a single-time measure such as the scans (± 10 min) and heart rate was not significantly differ - the peak, especially when the definition of the peak also is ent between both scans. Still, subtle physiological differences affected by the temporal resolution of the 4D flow MRI data. could result in poor scan–rescan reproducibility for these few cases. Table 6 in “Appendix 2” shows the HR, LV outflow and Viscous EL over LV diastole CO of all subjects. Subject 1 and 2 present the highest CO dif- ferences, however there are other subjects with CO differences The assessment of intracardiac EL could provide crucial within the same range. Table 7 in “Appendix 2” shows subject- details on the function of the heart apart from the standard specific scan–rescan information from the mitral valve flow MRI parameters and could be used to further unravel the curves. Subject 1 and 2 have the highest difference in area under influence of complex surgery for congenital heart defects [1 ]. the curve of the E-filling and peak filling rate of the E-filling. To our knowledge, this is the first study assessing This could indicate that differences may be related to technical scan–rescan reproducibility of in vivo LV EL over diastole restrictions of the acquisition, as the E-filling duration, E peak 1 3 914 The International Journal of Cardiovascular Imaging (2018) 34:905–920 1 3 The International Journal of Cardiovascular Imaging (2018) 34:905–920 915 ◂Fig. 3 Bland–Altman plots of kinetic energy (KE) and viscous energy their absolute changes as in other tested parameters. As such, ∙ ∙ loss rate ( ) at peak early diastolic filling (KE and ) EL EL E-peak E-peak ∙ the reported reproducibility of EL could also be consid- index and peak late diastolic filling (KE and ), EL EL A−peak A−peak total diastole ered as another reflection of a good internal consistency in and EL . a Bland–Altman plot depicting the agreement between index this study. Although a similar EL parameter was reported KE (black) and KE (grey) in scan 1 and scan 2. The subjects index E-avg E-peak with distinct higher differences between scan and rescan measure- in a previous study and was shown to be significantly altered ments (Subject 1 and 2) are depicted as red triangles (KE dark E-avg in patients with acute myocardial infarction [18], EL was index red; KE light red). b Bland–Altman plot depicting the agreement E-peak ∙ ∙ computed in that study from 2D echo particle image veloci- between (black) and (grey) in scan 1 and scan 2. Sub- EL EL E-avg E-peak metry and over the complete cardiac cycle as compared to this ject 1 and 2 are depicted as red triangles (KE dark red; KE E-avg E-peak light red). c Bland–Altman plot depicting the agreement between study’s volumetric measurement from 4D flow MRI and over KE (black) and KE (grey) in scan 1 and scan 2. Subject 1 A−avg A−peak diastole only. Therefore, it is not possible to perform a direct and 2 are depicted as red triangles (KE dark red; KE light A−avg A−peak comparison with the published results of that study. red). d Bland–Altman plot depicting the agreement between EL A−avg (black) and (grey) in scan 1 and scan 2. Subject 1 and 2 EL A−peak ∙ ∙ Vorticity inside the LV during diastolic filling are depicted as red triangles ( dark red; light red). EL EL A−avg A−peak e Bland–Altman plot depicting the agreement between EL total diastole in scan 1 and scan 2. Subject 1 and 2 are depicted as red triangles. f LV vortex quantification parameters, such as vorticity, Bland–Altman plot depicting the agreement between EL in scan 1 index could be useful in the assessment of LV and RV diastolic and scan 2. Subject 1 and 2 are depicted as red triangles (dys-)function [1, 8, 9, 11, 12]. In recent studies, vorticity was shown to be a marker of diastolic dysfunction, both assessed by 4D flow MRI. Results of norm_ in the in the LV [9] and the RV [8] of patients with pulmonary E-peak EL current study correspond well with that of a different healthy hypertension. controls cohort of a recent 4D flow MRI study by Elbaz et al. To our knowledge, no previous study is available on [1] Also norm_ values are consistent. Furthermore, assessing scan–rescan reproducibility of in vivo vorticity_LV E-avg EL ∙ ∙ both results of norm_ , as well as norm_ are over diastole from 4D flow MRI. Fenster et al. [8 ] assessed A−peak A−avg EL EL similar to previously reported numbers [1]. vorticity inside the RV using the integral of vorticity mag- Both EL and norm_EL are slightly nitude over the volumes and found results in the same order total diastole total diastole higher in our study than in the study by Elbaz et al. [1]. of magnitude as the results in the current study. In a recent However, results remain in the same range. These differ- paper by Schafer et al. [9] LV vorticity in healthy subjects ences in results could be explained in part by differences was assessed. However, it is not clear whether this was com- in heart rate between the volunteers of this study and puted over the whole LV volume and therefore we cannot those of the previous study. This is because the total vis- compare results to our measurements. The current study cous energy loss over diastole is computed over the time showed scan–rescan reproducibility of integral vorticity_LV period between the first and the last phase of the diastole during E-filling as well as during A-filling with good–excel- (Eq. 3) in “Appendix 1”) and heart rate mainly affects the lent ICCs and CVs up to 11%. Based on the Bland–Altman duration of diastole. Still, the EL-time curve over total analyses, similar to the KE and EL results, the variability of diastole in the current study (Fig. 1d) agrees well with the average of the vorticity_LV and vorticity_LV E-avg A−avg reported in vivo [1] and in vitro [2] EL time curves. was less than the variability of the peaks v orticity_LV E-peak In this study we demonstrated the reproducibility of the and vorticity_LV. Vorticity_LV shows less vari- A−peak E-avg EL parameters with moderate–strong ICCs and substan- ability than vorticity_LV but vorticity_LV shows A−avg E-peak tial CVs of up to 25%. Similar to KE, the Bland–Altman slightly more variability than vorticity_LV . A−peak ∙ ∙ analysis shows that the variability of and E-avg A−avg EL EL ∙ ∙ was less than the variability of and Technical considerations E-peak A−peak EL EL The amount of energy lost over diastole relative to the aver- age kinetic energy as measured by means of EL shows The post-processing steps that are followed for obtaining index, good reproducibility for all subjects with an ICC of 0.94 these flow energetics and vorticity parameters involve man- (P < 0.001) and an CV of 9%. The Bland–Altman plot shows ual segmentation, registration and valve tracking. Manual that all subjects are within the same range of differences. This segmentation on cine short axis images can be performed suggests that among tested parameters EL is the least sensi- with excellent reproducibility [27]. In this study, manual index tive to subtle physiological variations or discrepancies affected segmentation was performed by one observer with over 2 by technical limitations of the 4D flow MRI data in the healthy years of experience in MRI and verified by another observer subjects, which might have affected the lesser reproducibility with over 15 years of experience, with validated software of KE and EL in some subjects. This observation could be [28], which warrants high accuracy. Next, to correct for attributed to the fact that EL is a dimensionless parameter patient motion related misalignment and minimize errors index concerned with the relative changes in EL to KE and not with between the cine short axis and the 4D flow acquisitions, 1 3 916 The International Journal of Cardiovascular Imaging (2018) 34:905–920 1 3 Table 4 Scan–rescan comparison of vortex flow variables for the cohort with all subjects (N = 12) Scan 1 Scan 2 Difference (scan 2–scan 1) Pearson correla- Intraclass correla- Coefficient of tion coefficient tion coefficient variation (%) Mean ± SD (min–max) or Mean ± SD (min–max) or Mean ± SD (min–max) or median [IQR] ICC (95% CI) median [IQR] (min–max) median [IQR] (min–max) (min–max) Early filling  Vorticity_LV (mL/s) 4671 ± 992 (2950–6370) 4551 ± 1043 (2930–6200) − 120 ± 1075 (− 239 to 1710) 0.44 0.63 (− 0.37 to 0.90) 15 E-peak  Vorticity_LV (mL/s) 2862 ± 636 (1720–3840) 2788 ± 671 (1810–3930) − 74 ± 608 (− 1510 to 1060) 0.57 0.74 (0.06–0.93) 13 E-avg Late filling b a  Vorticity_LV (mL/s) 2715 [2502–4142] (2176– 3193 ± 945 (2174–4914) − 10 ± 547 (− 1270 to 840) 0.75 0.92 (0.73–0.98) 11 A−peak 5463) b a  Vorticity_LV (mL/s) 2660 ± 842 (1740–4180) 2599 ± 766 (1730–4050) − 61 ± 484 (− 1270 to 740) 0.82 0.91 (0.68–0.97) 11 A−avg E early diastolic filling, A late diastolic filling *Subjects with measurements outside the 95% confidence interval of the respective Bland–Altman plots P < 0.001 P < 0.05 P < 0.05 for non-normality (Shapiro–Wilk test) Table 5 Scan–rescan comparison of vortex flow variables for the sub-cohort without Subject 1 and 2 (N = 10) Scan 1 Scan 2 Difference (scan 2–scan 1) Pearson correlation Intraclass correlation Coefficient of coefficient coefficient variation (%) Mean ± SD (min–max) or Mean ± SD (min–max) or Mean ± SD (min–max) or ICC (95% CI) median [IQR] (min–max) median [IQR] (min–max) median [IQR] (min–max) Early filling b b  Vorticity_LV (mL/s) 4546 ± 1008 (2950–6370) 4520 ± 1004 (2930–6200) − 27 ± 796 (− 940 to 1710) 0.69 0.83 (0.27–0.96) 11 E-peak a a  Vorticity_LV (mL/s) 2764 ± 614 (1720–3560) 2720 ± 613 (1810–3610) − 44 ± 277 (− 380 to 480) 0.90 0.95 (0.80–0.99) 7 E-avg Late filling b b  Vorticity_LV (mL/s) 2563 [2468–3628] (2176– 3012 ± 852 (2174–4586) − 69 ± 575 (− 1270 to 840) 0.64 0.91 (0.64–0.98) 11 A−peak 5463) b b  Vorticity_LV (mL/s) 2269 [1889–3251] (1740– 2465 ± 679 (1730–3730) − 75 ± 511 (− 1270 to 740) 0.82 0.89 (0.53–0.97) 11 A−avg 4180) E early diastolic filling, A late diastolic filling *Subjects with measurements outside the 95% confidence interval of the respective Bland–Altman plots P < 0.001 P < 0.05 P < 0.05 for non-normality (Shapiro–Wilk test) The International Journal of Cardiovascular Imaging (2018) 34:905–920 917 Limitations A limitation of this study is the small number of subjects. Furthermore, no patients were scanned in this study as this is difficult to assess in clinical research. In most of our patient studies, 4D flow MRI is part of a clinical CMR evaluation of about 75–90 min, sometimes involving the use of contrast for late gadolinium enhancement and sometimes involving a Dobutamine rest/stress protocol. Scan–rescan evaluation in such cases would imply repeating some of these evaluations as well. Furthermore, repeating 4D flow MRI for scan–rescan purposes would imply an additional scan-time of approxi- mately 30 min (because of replanning), which makes the total scan-time too long to keep the patient in the same physiologi- cal state. Another limitation is that the influence of a different scanning protocol or scanner was not assessed. However, it is important to note that in the current study the aim was to assess scan–rescan reproducibility by using the exact same protocol and the same scanner machine twice. The use of dif- ferent scanners or scanning protocols could result in altered reproducibility, which should be evaluated in future studies. In conclusion, left ventricular kinetic energy and viscous energy loss quantification from 4D flow MRI in healthy vol- unteers shows strong–excellent in-scan consistency. Scan–res- can assessment of left ventricular kinetic energy, viscous energy loss and vorticity shows overall good agreement in the majority of the scanned subjects. Nevertheless, in two out of twelve subjects, considerable variation between the scans was found. Agreement of A-filling measurements is better than E-filling between scans in the studied parameters. Further - more, time-averaged measurements over early and late filling show better reproducibility compared to peak measurements. Strong reproducibility for all cases is found in the dimension- less index, EL , that measures the ratio of the amount of index viscous energy lost relative to the average kinetic energy over diastole. EL seems to be less influenced by technical and/ Fig. 4 Bland–Altman plots of volumetric vorticity over peak early index filling (vorticity_LV and vorticity_LV ) and late filling or slight physiological differences between scans and may E-avg E-peak (vorticity_LV and vorticity_LV ). a Bland–Altman plot A−avg A−peak therefore be a useful parameter of energetics for future studies. depicting the agreement between v orticity_LV (black) and E-avg vorticity_LV (grey) in scan 1 and scan 2. Subject 1 and 2 are E-peak Acknowledgements The authors thank professor Dr. S. le Cessie for depicted as red triangles (v orticity_LV dark red; v orticity_ E-avg statistical advice. LV light red). b Bland–Altman plot depicting the agreement E-peak vorticity_LV (black) and v orticity_LV (grey) in scan 1 A−avg A−peak Funding V.P. Kamphuis and R.L.F. van der Palen are financially sup- and scan 2. Subject 1 and 2 are depicted as red triangles (vorticity_ ported by grants from the Dutch Heart Foundation (Grant Numbers LV dark red; vorticity_LV light red) A−avg A−peak 2013T091 and 2014T087, respectively). J.J.M. Westenberg is finan - cially supported by a Grant of ZonMw (Project Number 104003001). automated image-based 3D rigid registration was performed Compliance with ethical standards using the validated Elastix image registration toolbox [17]. Another post-processing step requires retrospective valve Conflict of interest The authors declare that they have no conflict of tracking and mitral flow velocity mapping for assessing the interest. beginning and ending of diastole. The reproducibility and observer variability of this semi-automated method was Open Access This article is distributed under the terms of the Crea- shown to be excellent [14, 15]. tive Commons Attribution 4.0 International License (http://crea- tivecommons.org/licenses/by/4.0/), which permits unrestricted use, 1 3 918 The International Journal of Cardiovascular Imaging (2018) 34:905–920 distribution, and reproduction in any medium, provided you give appro- As a result, the total viscous energy loss ( EL ) in joules over priate credit to the original author(s) and the source, provide a link to time period T starting at phase t and ending at t can be start end the Creative Commons license, and indicate if changes were made. computed as: end EL = EL p [Joule(J)] (3) T d d Appendix 1 d=t start Viscous energy loss computation from 4D flow MRI with p the time step (temporal resolution) of the acquired 4D flow MRI. Given the acquired velocity field v, the rate of viscous energy loss ( ) in watt (W) and the total energy loss ( EL ) in total EL Vorticity computation from 4D flow MRI joule (J) over a given period of time T can be computed from 4D flow MRI using the viscous dissipation function  in the If u,v, w denote the three velocity field components acquired Newtonian Navier–Stokes energy equations: from 4D flow MRI over the principal velocity directions 3 3 2 x, y, z, respectively, the vorticity (  ) at voxel i of an acquired i,t 1 i 2 = + − (∇ ⋅ v) , time phase t v ij 2 x x 3 j i i=1 j=1 w v u w v u i,t i,t i,t i,t i,t i,t = − , − , − 1∕s = 1, if i = j i,t ij −2 (1) y z z x x y [s ] i,t i,t i,t i,t i,t i,t = 0, if i ≠ j ij (4) Then, Vorticity_LV denoting the integral sum of vorticity represents the rate of viscous energy dissipation per unit volume. i, j correspond to the principal velocity direc- over the segmented LV volume at an acquired time phase t in liter (mL) per second (s) can be computed as tions x, y, z. ∇ ⋅ v denotes the divergence of the velocity field. Therefore, the rate of viscous energy loss ( ) in Watt M EL at an acquired time phase t can be computed as: Vorticity_LV =  L mL∕s (5) t i,t i,t i=1 With   as the magnitude of the vorticity vector, M as i,t EL =μ  L [Watt(W)] (2) t v i the total number of voxels in the segmented LV volume and i=1 L as the voxel volume. i,t assuming the blood as a Newtonian fluid, the dynamic vis- cosity is μ = 0.004 Pa  s, N as the total number of voxels in the given domain of interest (e.g. LV), L as the voxel Appendix 2 volume. See Tables 6 and 7. Table 6 Subject specific scan–rescan assessment Subject HR 1 (bpm) HR 2 Difference LV outflow LV outflow Difference LV CO 1 (L/min) CO 2 (L/min) Difference (bpm) HR (bpm) 1 (mL) 2 (mL) outflow (mL) CO (%) 1 67.1 69.6 2.5 104.1 124.8 20.7 7.0 8.7 24.4 2 53.8 51.6 − 2.2 112.9 94.5 − 18.4 6.1 4.9 − 19.7 3 58.0 51.8 − 6.2 110.0 126.4 16.4 6.4 6.6 2.7 4 61.5 59.1 − 2.4 81.5 72.4 − 9.1 5.0 4.3 − 14.7 5 71.9 66.2 − 5.7 68.4 66.0 − 2.4 4.9 4.4 − 11.2 6 50.5 52.9 2.4 100.7 108.6 7.9 5.1 5.7 12.9 7 63.7 60.4 − 3.3 104.0 107.8 3.8 6.6 6.5 − 1.7 8 57.7 53.1 − 4.6 85.4 86.4 1.0 4.9 4.6 − 6.9 9 71.1 72.1 1.0 75.3 84.4 9.1 5.4 6.1 13.7 10 68.0 61.3 − 6.7 85.0 81.8 − 3.2 5.8 5.0 − 13.3 11 48.5 58.3 9.8 90.3 88.7 − 1.5 4.4 5.2 18.2 12 58.0 62.0 4.0 71.1 76.3 5.2 4.1 4.7 14.7 Differences were calculated as: value scan 2 − value scan 1 A late diastolic filling, AUC area under the cure, CO cardiac output, E early diastolic filling, HR heart rate Subjects with marked high differences for the assessment of early filling parameters 1 3 The International Journal of Cardiovascular Imaging (2018) 34:905–920 919 1 3 Table 7 Subject specific scan–rescan assessment: information from the MV flow curves Subject E AUC E AUC Differ - E PFR 1 E PFR 2 Differ - E dur 1 E dur 2 Differ - A AUC A AUC Differ - A A Differ - A dur 1 A dur 2 Differ - 1 (mL) 2 (mL) ence E (mL/s) (mL/s) ence (ms) (ms) ence 1 (mL) 2 (mL) ence A  PFR 1 PFR 2 ence (ms) (ms) ence A AUC E PFR E dur AUC (mL/s) (mL/s) A PFR dur (ms) (mL) (mL/s) (ms) (mL) (mL/s) 1 78.6 94.8 16.2 522.8 738.8 216.0 332 292 − 40 28.1 29.3 1.2 342.1 357.7 15.6 166 175 9 2 84.3 71.3 − 13.0 567.4 388.4 − 179.0 323 413 90 19.1 14.1 − 5.0 243.5 159.6 − 83.9 144 188 44 3 85.7 96.4 10.7 744.5 734.7 − 9.8 309 270 − 39 32.8 28.0 − 4.8 409.5 347.0 − 62.5 172 194 22 4 68.5 61.4 − 7.1 551.4 537.9 − 13.5 316 263 − 53 15.3 12.4 − 2.9 161.8 158.0 − 3.8 190 131 − 59 5 52.6 46.8 − 5.8 343.6 362.2 18.6 348 297 − 51 16.7 15.0 − 1.7 187.0 176.3 − 10.7 174 208 34 6 83.9 84.6 0.7 500.2 545.2 45.0 378 328 − 50 13.6 22.0 8.4 137.4 231.3 93.9 189 197 8 7 79.3 87.5 8.2 683.6 647.7 − 35.9 314 351 37 32.3 29.6 − 2.7 409.2 374.7 − 34.5 157 160 3 8 72.2 71.6 − 0.6 592.5 543.6 − 48.9 384 350 − 34 17.5 17.5 0.0 221.1 217.1 − 4.0 175 176 1 9 59.4 66.8 7.4 506.1 563.6 57.5 272 300 28 18.3 19.6 1.3 243.7 245.0 1.3 136 137 1 10 58.7 58.9 0.2 367.5 376.7 9.2 322 373 51 17.6 14.4 − 3.2 185.3 173.6 − 11.7 162 170 8 11 64.5 60.9 − 3.6 398.4 363.9 − 34.5 349 343 − 6 17.5 14.4 − 3.1 181.8 172.3 − 9.5 209 205 − 4 12 52.8 47.4 − 5.4 380.3 314.4 − 65.9 330 313 − 17 8.5 12.9 4.4 114.5 164.0 49.5 165 187 22 Differences were calculated as: value scan 2 − value scan 1 A late diastolic filling, AUC area under the cure, CO cardiac output, dur duration, E early diastolic filling, HR heart rate, PFR peak filling rate Subjects with marked high differences for the assessment of early filling parameters 920 The International Journal of Cardiovascular Imaging (2018) 34:905–920 15. Westenberg JJ, Roes SD, Ajmone Marsan N, Binnendijk NM, References Doornbos J, Bax JJ, Reiber JH, de Roos A, van der Geest RJ (2008) Mitral valve and tricuspid valve blood flow: accurate 1. Elbaz MS, van der Geest RJ, Calkoen EE, de Roos A, Lelieveldt quantification with 3D velocity-encoded MR imaging with retro - BP, Roest AA, Westenberg JJ (2016) Assessment of viscous spective valve tracking. Radiology 249(3):792–800. https://doi. energy loss and the association with three-dimensional vortex ring org/10.1148/radiol.2492080146 formation in left ventricular inflow: in vivo evaluation using four- 16. Garg P, Westenberg JJM, van den Boogaard PJ, Swoboda PP, Aziz dimensional flow MRI. Magn Reson Med. https://doi.org/10.1002/ R, Foley JRJ, Fent GJ, Tyl FGJ, Coratella L, ElBaz MSM, van mrm.26129 der Geest RJ, Higgins DM, Greenwood JP, Plein S (2017) Com- 2. Pedrizzetti G, Domenichini F (2005) Nature optimizes the swirl- parison of fast acquisition strategies in whole-heart four-dimen- ing flow in the human left ventricle. Phys Rev Lett 95(10):108101. sional flow cardiac MR: two-center, 1.5 T, phantom and in vivo https://doi.org/10.1103/PhysRevLett.95.108101 validation study. J Magn Reson Imaging. https://doi.org/10.1002/ 3. Kamphuis VP, Westenberg JJM, van der Palen RLF, Blom NA, jmri.25746 de Roos A, van der Geest R, Elbaz MSM, Roest AAW (2017) 17. Klein S, Staring M, Murphy K, Viergever MA, Pluim JP (2010) Unravelling cardiovascular disease using four dimensional flow elastix: a toolbox for intensity-based medical image registration. cardiovascular magnetic resonance. Int J Cardiovasc Imaging IEEE Trans Med Imaging 29(1):196–205. https://doi.org/10.1109/ 33(7):1069–1081. https://doi.org/10.1007/s10554-016-1031-9 TMI.2009.2035616 4. Eriksson J, Bolger AF, Ebbers T, Carlhall CJ (2013) Four-dimen- 18. Agati L, Cimino S, Tonti G, Cicogna F, Petronilli V, De Luca L, sional blood flow-specific markers of LV dysfunction in dilated Iacoboni C, Pedrizzetti G (2014) Quantitative analysis of intraven- cardiomyopathy. Eur Heart J Cardiov Imaging 14(5):417–424. tricular blood flow dynamics by echocardiographic particle image https://doi.org/10.1093/ehjci/jes159 velocimetry in patients with acute myocardial infarction at differ - 5. Fredriksson AG, Svalbring E, Eriksson J, Dyverfeldt P, Alehagen ent stages of left ventricular dysfunction. Eur Heart J Cardiovasc U, Engvall J, Ebbers T, Carlhall CJ (2016) 4D flow MRI can Imaging 15(11):1203–1212. https://doi.org/10.1093/ehjci/jeu106 detect subtle right ventricular dysfunction in primary left ven- 19. Hyslop NP, White WH (2009) Estimating precision using dupli- tricular disease. J Magn Reson Imaging 43(3):558–565. https:// cate measurements. J Air Waste Manag Assoc 59(9):1032–1039 doi.org/10.1002/jmri.25015 20. Bland JM, Altman DG (1986) Statistical methods for assessing 6. Kanski M, Arvidsson PM, Toger J, Borgquist R, Heiberg E, Carls- agreement between two methods of clinical measurement. Lancet son M, Arheden H (2015) Left ventricular fluid kinetic energy 1(8476):307–310 time curves in heart failure from cardiovascular magnetic reso- 21. Bolger AF, Heiberg E, Karlsson M, Wigstrom L, Engvall J, Sig- nance 4D flow data. J Cardiovasc Magn Reson 17:111. https://doi. fridsson A, Ebbers T, Kvitting JP, Carlhall CJ, Wranne B (2007) org/10.1186/s12968-015-0211-4 Transit of blood flow through the human left ventricle mapped 7. Svalbring E, Fredriksson A, Eriksson J, Dyverfeldt P, Ebbers by cardiovascular magnetic resonance. J Cardiovasc Magn Reson T, Bolger AF, Engvall J, Carlhall CJ (2016) Altered diastolic 9(5):741–747. https://doi.org/10.1080/10976640701544530 flow patterns and kinetic energy in subtle left ventricular remod- 22. Mohiaddin RH (1995) Flow patterns in the dilated ischemic left eling and dysfunction detected by 4D flow MRI. PLoS ONE ventricle studied by MR imaging with velocity vector mapping. J 11(8):e0161391. https://doi.org/10.1371/journal.pone.0161391 Magn Reson Imaging 5(5):493–498 8. Fenster BE, Browning J, Schroeder JD, Schafer M, Podgorski CA, 23. Francis GS (2001) Pathophysiology of chronic heart failure. Am Smyser J, Silveira LJ, Buckner JK, Hertzberg JR (2015) Vorticity J Med 110(Suppl 7A):37S–46S is a marker of right ventricular diastolic dysfunction. Am J Physiol 24. Kanski M, Toger J, Steding-Ehrenborg K, Xanthis C, Bloch KM, Heart Circ Physiol 309(6):H1087–H1093. https://doi.org/10.1152/ Heiberg E, Carlsson M, Arheden H (2015) Whole-heart four- ajpheart.00278.2015 dimensional flow can be acquired with preserved quality with- 9. Schafer M, Browning J, Schroeder JD, Shandas R, Kheyfets out respiratory gating, facilitating clinical use: a head-to-head VO, Buckner JK, Hunter KS, Hertzberg JR, Fenster BE (2016) comparison. BMC Med Imaging 15:20. https://doi.org/10.1186/ Vorticity is a marker of diastolic ventricular interdependency s12880-015-0061-4 in pulmonary hypertension. Pulm Circ 6(1):46–54. https://doi. 25. Carlsson M, Heiberg E, Toger J, Arheden H (2012) Quantification org/10.1086/685052 of left and right ventricular kinetic energy using four-dimensional 10. Kheradvar AP, G (2012) Vortex formation in the heart. In: Vortex intracardiac magnetic resonance imaging flow measurements. formation in the cardiovascular system. Springer, London, p 19 Am J Physiol Heart Circ Physiol 302(4):H893–H900. https://doi. 11. Kheradvar A, Assadi R, Falahatpisheh A, Sengupta PP (2012) org/10.1152/ajpheart.00942.2011 Assessment of transmitral vortex formation in patients with dias- 26. Hussaini SF, Rutkowski DR, Roldan-Alzate A, Francois CJ (2016) tolic dysfunction. J Am Soc Echocardiogr 25(2):220–227. https:// Left and right ventricular kinetic energy using time-resolved ver- doi.org/10.1016/j.echo.2011.10.003 sus time-average ventricular volumes. J Magn Reson Imaging. 12. Stewart KC, Charonko JC, Niebel CL, Little WC, Vlachos PP https://doi.org/10.1002/jmri.25416 (2012) Left ventricular vortex formation is unaffected by diastolic 27. Grothues F, Smith GC, Moon JC, Bellenger NG, Collins P, Klein impairment. Am J Physiol Heart Circ Physiol 303(10):H1255– HU, Pennell DJ (2002) Comparison of interstudy reproducibility H1262. https://doi.org/10.1152/ajpheart.00093.2012 of cardiovascular magnetic resonance with two-dimensional echo- 13. Kamphuis VP, Roest AAW, Westenberg JJM, Elbaz MSM (2017) cardiography in normal subjects and in patients with heart failure Biventricular vortex ring formation corresponds to regions of or left ventricular hypertrophy. Am J Cardiol 90(1):29–34 highest intraventricular viscous energy loss in a Fontan patient: 28. van der Geest RJ, Reiber JH (1999) Quantification in cardiac MRI. analysis by 4D Flow MRI. Int J Cardiovasc Imaging. https://doi. J Magn Reson Imaging 10(5):602–608 org/10.1007/s10554-017-1250-8 14. Kamphuis VP, van der Palen RLF, de Koning PJH, Elbaz MSM, van der Geest RJ, de Roos A, Roest AAW, Westenberg JJM (2017) In-scan and scan–rescan assessment of LV in- and out-flow vol- umes by 4D flow MRI versus 2D planimetry. J Magn Reson Imag- ing. https://doi.org/10.1002/jmri.25792 1 3

Journal

The International Journal of Cardiovascular ImagingSpringer Journals

Published: Jan 5, 2018

There are no references for this article.