Quantitative analysis of intraventricular flow-energetics and vortex in ischaemic hearts

(1)

Quantitative analysis of intraventricular flow-energetics and vortex in ischaemic hearts

Bee Ting Chan

^a

, Hak Koon Yeoh

^b

, Yih Miin Liew

^a

, Socrates Dokos

^f

, Amr Al Abed

^f

, Kok Han Chee

^c

, Yang F. Abdul Aziz

^e

,

Ganiga Srinivasaiah Sridhar

^c

, Karuthan Chinna

^d

and Einly Lim

^a

Objective This study investigated the intraventricular flow dynamics in ischaemic heart disease patients.

Patients and methods Fourteen patients with normal ejection fraction and 16 patients with reduced ejection fraction were compared with 20 healthy individuals. Phase- contrast MRI was used to assess intraventricular flow variables and speckle-tracking echocardiography to assess myocardial strain and left ventricular (LV) dyssynchrony.

Infarct size was acquired using delayed-enhancement MRI.

Results The results obtained showed no significant differences in intraventricular flow variables between the healthy group and the patients with normal ejection fraction group, whereas considerable reductions in kinetic energy (KE) fluctuation index,E′(P<0.001) and vortex KE (P=0.003) were found in the patients with reduced ejection fraction group. In multivariate analysis, only vortex KE and infarct size were significantly related to LV ejection fraction (P<0.001); furthermore, vortex KE was correlated

negatively with energy dissipation, energy dissipation index (r=−0.44,P=0.021).

Conclusion This study highlights that flow energetic indices have limited applicability as early predictors of LV progressive dysfunction, whereas vortex KE could be an alternative to LV performance. Coron Artery Dis29:316–324 Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.

Coronary Artery Disease2018,29:316–324

Keywords: dissipation, flow, fluctuation, infarct, kinetic energy, vortex

aDepartment of Biomedical Engineering,^bDepartment of Chemical Engineering,

cDepartment of Medicine,^dDepartment of Social and Preventive Medicine,

eDepartment of Biomedical Imaging, University Malaya Research Imaging Centre, Faculty of Medicine, University of Malaya, Kuala Lumpur, Malaysia and^fGraduate School of Biomedical Engineering, University of New South Wales, Sydney, New South Wales, Australia

Correspondence to Einly Lim, PhD, Department of Biomedical Engineering, Faculty of Engineering, Universiti Malaya, 50603 Kuala Lumpur, Malaysia Tel: + 60 379 677 612; fax: + 60 379 674 579; e-mail: [email protected] Received13 September 2017Revised6 November 2017

Accepted17 November 2017

Introduction

Intraventricular vortex [1–3] and flow-energetic indices, such as energy dissipation index (DI) [4,5] and kinetic energy (KE) fluctuation [6,7], are potential early indica- tors for maladaptive cardiac function before a significant structural change occurs. It has been hypothesized that high fluctuations in vorticity and KE observed in patients with small infarcts and preserved ejection fraction [8] are a compensatory mechanism in maintaining left ventricular ejection fraction (LVEF) that, nevertheless, would favour adverse remodelling at a later stage. Most studies suggest that KE fluctuations lead to maladaptive flow dynamics, and thus inefficient left ventricular (LV) function. However, a recent image-based computational fluid dynamics study, which reported a substantial amount of turbulent KE in healthy LVs [9], suggests that LV flow turbulence is beneficial in preventing blood aggregation [10] while triggering mechano-sensitive feedback between blood flow and the myocardial wall [11]. Considering these contradictory findings, the question of whether the fluctuations in flow-energetic indices are indicative of a negative compensatory

mechanism or purely a measurable consequence of dynamic velocity conditions in the LV remains incon- clusive. Despite various hypotheses linking abnormal vortex dynamics and contraction inhomogeneity to flow- energetic indices in the LV, only a few studies have investigated their correlations to understand the mechanism leading to deterioration in LVEF function.

Even though contraction inhomogeneity was suggested to be the cause of high KE fluctuation and DI observed in postinfarct patients in a recent study, no assessment was performed in this respect [8]. This study quantita- tively investigated the correlation between intraventricular flow-energetic indices and vortex parameters with LV wall motion [reflected by global longitudinal strain (GLS)] and LV dyssynchrony [indicated by time-to-peak (TPS)-SD ] in ischaemic heart disease (IHD) patients with varying degrees of LVEFs. In contrast to the previous study that investigated flow-energetic variables in postinfarct patients using echocardiography particle imaging velocimetry (echo-PIV) [8], this study utilized two-dimensional (2D) PC-MRI to measure LV flow as it is more accurate for clustered high or low velocities [12].

(2)

In addition, the correlations among these important variables and their relationship with LV pumping efficiency (indicated by LVEF) were also examined.

Patients and methods

Study participants

The study participants included 20 healthy individuals and 30 IHD patients with ST elevation myocardial infarction (Table 1). The IHD patients were further classified into two groups on the basis of their LVEF: IHD patients with normal ejection fraction (PNEF) had LVEF of at least 50%, whereas patients with reduced ejection fraction (PREF) had LVEF less than 50%. The exclusion criteria for the patients were unstable angina, atrial fibril- lation, tachycardia (>100 bpm at rest) and severe valvular regurgitation or stenosis. The healthy individuals had no history of cardiovascular disease and had normal cardiac function, as confirmed by echocardiographic examination.

Individuals with contraindications to MRI, including claustrophobia and ferrous implants, were excluded.

Written informed consent was obtained from all participants before participation in this study. The research was approved by the University of Malaya Medical Centre Medical Ethics Committee (ref.: 98975).

Image acquisition Echocardiography

Transthoracic 2D echocardiographic examination was performed on all participants using an iE33 ultrasound machine (Phillips Medical Systems, Andover, Massachusetts, USA)

with an S5-1 Sector Array transducer. The myocardial contraction and LV dyssynchrony were assessed from three apical long-axis views in 2D Speckle-Tracking Echocardiography. GLS was acquired from the average segmental peak systolic strain in a 17-segment model. The LV dyssynchrony was evaluated using TPS-SD. To minimize operator error, the procedures were performed by three experienced sonographers and the average values were obtained for GLS and TPS-SD readings.

MRI

Cine MRI was performed using a 1.5-T MRI scanner (Signa HDxt; GE Healthcare, Milaukee, Wisconsin, USA) with the participants in a supine position. An eight- channel cardiac coil was placed on each patient’s chest.

The end systolic volume, end diastolic volume and LVEF were acquired from motion-corrected three- dimensional geometrical models reconstructed on the basis of the short axis and the long axis of the cine MRI of the LV [13]. The cine MRI acquisition protocol was the same as that reported in a previous study [13].

PC-MRI was performed using a three-chamber acquisition protocol with a multi-breath-hold 2D Fast Cine Phase- Contrast sequence, which enables shorter scan times through prospective gating and segmented k-space. The images were reconstructed retrospectively throughout the entire cardiac cycle. Velocity encoding was 150 cm/s, with a repetition time of 6.2–6.5 ms, resulting in 20 reconstructed phases. The echo time was 3.2–3.5 ms, the flip angle was 25° and the acquisition matrix was 256×256 pixels, in

Table 1 Participants’demographics

Healthy participants (n=20) PNEF (n=14) PREF (n=16) Group comparison (Pvalues)

Sex (male : female) 11 : 9 11 : 3 15 : 1 –

Age 50±8 54±11 56±7 0.071

BSA (m²) 1.8±0.2 1.8±0.1 1.7±0.2 0.492

SBP (mmHg) 119±7 137±26 126±16 0.094

DBP (mmHg) 75±7 81±10 72±11 0.124

Heart rate (bpm) 70±10 71±9 72±15 0.731

LVEF (%)^ƚ,¶ 70±9 65±11 36±9 <0.001

LVESV (ml)^ƚ,¶ 25±15 24±12 83±46 <0.001

LVEDV (ml)^¶ 81±42 68±29 125±60 0.019

SV (ml) 56±28 44±21 42±19 0.307

GLS (%)^*,ƚ −16±2 −13±3 −10±2 0.000

TPS-SD (ms)* 71±14 102±39 90±20 0.012

Number of infarct segments^ƚ,¶,* – 4±3 8±2 <0.001

Infarct size (%)^ƚ,¶,* – 8±8 21±11 <0.001

NYHA class – 1.6±0.5 1.9±0.6 0.400

Dyslipidaemia (%) – 77 44 0.132

Hypertension (%) – 77 56 0.351

Smoking (%) – 69 40 0.201

Diabetes mellitus (%) – 62 69 0.746

Infarct-related coronary artery (%)

Left anterior descending – 64 94 0.179

Left circumflex – 14 38 0.294

Right coronary artery – 36 50 0.525

BSA, body surface area; DBP, diastolic blood pressure; GLS, global longitudinal myocardial strain; LVEDV, left ventricular end diastolic volume; LVEF, left ventricular ejection fraction; LVESV, left ventricular end systolic volume; PNEF, patients with normal ejection fraction; PREF, patients with reduced ejection fraction; SBP, systolic blood pressure; SV, stroke volume; TPS, time-to-peak systolic strain.

ƚSignificant difference between healthy versus PREF.

¶Significant difference between PNEF versus PREF.

*Significant difference between healthy versus PNEF.

(3)

which the in-plane pixel resolution was 1.37 mm×1.37 mm and slice thickness was 8 mm.

Delayed-enhancement MRI was performed using a 20°flip angle and a 256×256 pixels acquisition matrix, in which the in-plane pixel resolution was 1.37 mm×1.37 mm and the slice thickness was 8 mm. Images were acquired 8 min after the injection of gadolinium as the contrast agent and close to end-systole. Visualization and quantification of infarct size were performed manually by a qualified radiologist.

The infarct size was defined as the percentage ratio of scar mass to total LV mass.

MRI analyses

Artefact compensation was performed to mask phase errors and random noise [14]. The LV endocardial con- tours in the three-chamber view from all 20 cardiac phases were segmented manually from the magnitude PC-MRI using the Segment software package (v1.9 R3216; Medviso AB, Lund, Scania, Sweden) [15].

The vorticity (ω) was derived from the 2D velocity field in Eq. (1):

oðx;y;tÞ ¼qvx

qy qvy

qx (1)

and stream function,ψ, was estimated from the vorticity using Poisson’s equation:

r²c¼ o (2) with homogeneous boundary conditions applied to the borders of the image and the LV wall.

The flow pattern was then obtained from the phase- averaged or the steady-streaming stream function presented in Agatiet al.[8] and Honget al.[16]. The vortical region was defined as the flow region with steady-streaming stream function values greater than 60% of the maximum.

The vortex flow parameters, including area, circulation, vortex sphericity, Reynolds number (Re) and KE, were calculated. KE was measured in mJ/m, whereas all other vortex measurements were dimensionless. The area was defined as the vortical region normalized by the LV end- diastolic area; the circulation was the vorticity integral in the vortical region normalized by the absolute vorticity integral within the LV [8]; the vortex sphericity was the vortex width-to-length ratio [16]; the vortex Re was the vortex circulation normalized by the kinematic viscosity of blood;

and the vortex KE was the energy contained in the vortical region normalized by the vortex radius [1].

DI,E′andW′were calculated as dimensionless indices [8].

DI described the KE loss because of frictional forces within the blood flow and was calculated using Eq. (3) [8]:

DI¼ TR

T

R

LV

D dxdydt R

T

R

LV

E dxdydt (3)

In the DI formulation,Eis the KE, which is defined by Eðx;y;tÞ ¼r

2v²_xþv²_y

(4) andDis the DI rate by friction [17], which is calculated as follows:

Dðx;y;tÞ ¼rm 2 qv_x qx 2

þ2 qvy

qy 2

(

þ qvx

qy þqvy

qx

2

2 qvx

qxþqvy

qy 2)

(5) where ρ is the blood density (1050 kg/m³), μ is the kinematic viscosity (3.3×10⁻⁶m²/s) [8], T is the heart- beat duration and vx and vy are the in-plane velocity components.

E′ and W′ represent the degree of fluctuations and are calculated using Eqs (6) and (7) [8]:

E⁰¼

r 2

R

T

R

LV

vxvx0

ð Þ²þ vyvy0

2

h i

dxdydt R

T

R

LV

E dxdydt (6)

W⁰¼ R

T

R

LV

oo0

ð Þ²dxdydt R

T

R

LV

o²dxdydt (7)

whereωis the vorticity, andvx0,vy0andω0are the steady- streaming (phase-averaged) velocity and vorticity. The integral was computed as the sum of values, whereas the differential was computed as the subtraction of values in space. In the following, ‘flow variables’ refer to the aforementioned intraventricular flow-energetic indices as well as vortex flow parameters. The algorithms for calculat- ing flow variables were programmed using MATLAB (vR2012a, Mathworks, Natick, Massachusetts, USA).

Statistical analysis

Statistical analysis was carried out using the statistical package for the social sciences (SPSS v22.0; SPSS Inc., Chicago, Illinois, USA). All continuous data were reported as mean±SD and categorical variables were expres- sed in percentages. The three groups were compared using the Kruskal–Wallis test and post-hoc analysis was used to determine significant differences between groups.

For each participant group, the comparison ofE′inside and outside of the vortical region was performed using the Mann–Whitney U-test. The relationships among flow-energetic indices, myocardial movement and vortex parameters were analysed using the Pearson correlation test. The determinants of LVEF were assessed using univariate and multivariate linear regression analyses.

(4)

Pvalues less than 0.05 were considered to be statistically significant.

Results

The intraventricular flow patterns shown by a healthy individual, a PNEF and a PREF are shown in Fig. 1.

During diastole, blood filled the LV through the mitral valve, and subsequently rolled up, forming two counter- rotating vortex rings near the LV base as a result of the interaction between the E-wave and flow residing in the LV. Unlike the relatively symmetrical vortex pairs shown in the healthy individual and the PNEF, a pair of asymmetrical vortex rings was observed in the PREF.

During systole, a substantial amount of flow was ejected through the aortic valve in both the healthy and PNEF participants, with the appearance of a clockwise vortex adjacent to the left ventricular outflow tract (LVOT).

Conversely, a large systolic vortex remained in the mid LV of the PREF, who was unable to eject an adequate flow volume.

The characteristics of the healthy participants, PNEF and PREF groups are presented in Table 1. Although ischaemia and LV dyssynchrony were observed in the

PNEF group, they had similar LV volumes and LVEF to those of healthy participants. In addition, considerable LV dilatation (enlarged end systolic volume and end diastolic volume) and myocardial injury (large infarct size) were noted in the PREF group.

Figure 2 shows box plots of the flow-energetic indices (DI,E′, andW′) and vortex parameters (area, circulation, vortex sphericity index, Re and KE) among the healthy participants, PNEF and PREF groups. Overall, the three groups showed significant differences in E′ (P<0.001), vortex area (P=0.003), vortex Re (P=0.045) and vortex KE (P=0.003). The PREF group had significantly lower E′ (0.58±0.07) and vortex KE (290±116 mJ/m), but a larger vortex area (0.14±0.04) than the healthy participants (E′: 0.68±0.07; vortex KE: 445±144 mJ/m; vortex area: 0.10±0.03) and the PNEF (E′: 0.68±0.07;

vortex KE: 485±181 mJ/m; vortex area: 0.09±0.02).

The vortex Re in the PREF group (1917±323) was significantly higher than that of the healthy participants (1575±381), but not the PNEF group (1718±269).

In the IHD patients, the determinants of LVEF were investigated among myocardial properties and flow

Fig. 1

Intraventricular flow patterns of a healthy subject, PNEF and PREF patients during diastole and systole. The flow directions were indicated by the arrows while the vortex appearance was indicated by the circular regions. AOV, aortic valve; PNEF, patients with normal ejection fraction; PREF, patients with reduced ejection fraction; MV, mitral valve; RV, right ventricle.

(5)

variables (Table 2). The measurements of GLS, infarct size, KE fluctuation index (E′), vortex area and vortex KE were correlated significantly with LVEF in univariate analysis. However, in multivariate analysis, only infarct size and vortex KE were associated independently with LVEF and could explain 71% of the variation in LVEF.

As the altered myocardial properties in IHD patients were previously suggested to be associated with intraventricular flow turbulence [8], the correlations of flow fluctuation (characterized by E′) with myocardial con- tractility (indicated by GLS), infarct size and inhomoge- neous contraction (indicated by TPS-SD) were, therefore, examined. Our results showed that E′ is

correlated significantly with GLS (r=−0.45, P=0.016) as well as infarct size (r=−0.61,P<0.001), as shown in Fig. 3a and b. However, no correlation was observed betweenE′and TPS-SD (Fig. 3d).

In addition, the relationship between DI and E′, as reported in a previous echo-PIV study [8], was further investigated using PC-MRI, in which we observed no association between the two parameters (Fig. 3e). The proposed beneficial roles of intracardiac vortex flow in minimizing DI, and helping LV ejection [18], is shown by the significant negative correlation between DI and vortex KE (r=−0.44,P=0.021) in Fig. 3c.

As shown in Fig. 4a, the healthy participants and the PNEF group had highE′concentrated inside the vortical region near the LVOT. The regions with concentrated highE′coincided with high velocity, as shown in Fig. 4b.

As shown in Fig. 4c, theE′inside the vortical region was significantly higher than that outside for the healthy participants (0.77±0.08 vs. 0.65±0.07,P<0.001) and for PNEF (0.75±0.13 vs. 0.65±0.06,P=0.012). In contrast, for PREF, high E′ was observed along the shear layer outside the vortex. There was no significant difference in E′inside and outside the vortical regions (0.62±0.11 vs.

0.56±0.07,P>0.05).

Discussion

Currently, 2D PC-MRI is typically used to assess cardiovascular flow patterns [19–22]. Our intraventricular flow patterns results (Fig. 1) are congruent with previously reported findings [2,19,22]. However, this is the first study to investigate the intraventricular flow-energetic indices in healthy participants and IHD patients using PC-MRI. Among the key findings are absence of a

Fig. 2

Comparison of flow-energetic indices (a–c) and vortex flow parameters (d–h) among the three participant groups. The box plots indicate the first to third quartiles and the midline in between indicates the median; the whiskers represent the maximum and minimum values. Significant differences between the groups are indicated by **P<0.01 and *P<0.05. DI, energy dissipation index;E′, KE fluctuation index; PNEF, patients with normal ejection fraction; PREF, patients with reduced ejection fraction; Re, Reynolds number;W′, vorticity fluctuation index.

Table 2 Univariate and multivariate linear regression analyses for independent correlates of left ventricular ejection fraction in ischaemic heart disease patients (patients with normal ejection fraction and patients with reduced ejection fraction)

Univariate Multivariate

r P-value B P-value R²

GLS (%) −0.538 0.003 0.713

TPS-SD (ms) 0.097 0.624

Infarct size (%) −0.717 <0.001 −0.576 <0.001

DI −0.023 0.904

E′ 0.606 <0.001

W′ 0.361 0.050

Vortex area −0.553 0.002

Vortex circulation −0.118 0.535

Vortex SI 0.145 0.444

Vortex Re −0.306 0.121

Vortex KE (mJ/m) 0.678 <0.001 0.434 0.002 B, standardized regression coefficient; DI, energy dissipation index;E′, KE fluctuation index; GLS, global longitudinal strain; KE, kinetic energy; r, Pearson’s correlation coefficient; Re, Reynolds number; TPS, time-to-peak systolic strain;

W′, vorticity fluctuation index.

Bold values means statistically significant (P<0.05).

(6)

significant difference in flow-energetic indices between the healthy participants and the PNEF group, and a marked reduction inE′in the PREF group. Our findings are in stark contrast to those of the echo-PIV study, in which significantly higher flow-energetic indices (DI,E′, andW′) were found in the PNEF group compared with the healthy participants [8]. The discrepancy could be because of the different imaging modalities used. In particular, the PC-MRI adopted in our study is known to be more accurate for high LV velocities (both the healthy and PNEF groups have high intraventricular flow velocities) than echo-PIV [12,23]. Although PC-MRI has a relatively lower temporal resolution than echo-PIV, this does not critically affect our key findings as the key variables involved in our study were computed on each frame individually and phase-averaged results were subsequently acquired. A follow-up study with identical participants using both PC-MRI and echo-PIV would be useful to cross-check this hypothesis.

In addition to magnitudes, the spatial distribution ofE′

was also quite similar in both the healthy and PNEF groups. We observed strong acceleration and deceleration of blood near the LVOT, which disturbed the retained blood volume, thus leading to a vigorous flow in the LV.

Subsequently, highE′was found near the LVOT as well as inside the vortex core region. This phenomenon was absent in the PREF group, most likely because of damaged heart muscles at critical locations, thus

impairing the effectiveness of the contraction process. In the healthy volunteers and the PNEF group, the E′

inside the vortex core was significantly higher than that outside, in contrast to the PREF cohort. These results together suggest that the magnitude and spatial distribution of flow-energetic indices can identify PREF patients, but are not sufficient to distinguish the PNEF group from healthy participants.

Among the IHD patients, we found no direct correlation between the levels of flow fluctuation (characterized by E′) with DI. In the univariate and multivariate analyses, only infarct size and vortex KE were associated independently with LVEF. Following exclusion of the PREF group, the results were similar, with the exception that only vortex KE remained correlated significantly with LVEF.

The above findings have two key implications: first, we theorize that the high levels of flow fluctuations (reflected by DI,E′andW′) observed in both the healthy and PNEF participants were predominantly measurable consequences of dynamic velocity conditions in the LV, instead of reflections of negative compensatory mechanisms (undesired turbulence due to LV extra effort) to help in LV ejection. This is further supported by a recent simulation study, known to have a much finer spatial resolution and accuracy than existing imaging modalities, which also reported considerable flow fluctuation in healthy LV [9]. Considering the sensitivities of

Fig. 3

Correlations betweenE′and myocardial properties (a, b and d) as well as DI and flow variables (c, e) in IHD patients. Significant correlations (P<0.05) are indicated by filled data points. DI, energy dissipation index;E′, KE fluctuation index; GLS, global longitudinal strain; IHD, ischaemic heart disease.

(7)

flow-energetic indices to varying measurement methods with different spatial resolutions and accuracies, more detailed studies using computer simulations and participant follow-up are required to ascertain the role of flow fluctuation in the progression of LV dysfunction.

Interestingly, although the E′ in IHD patients was not correlated with LVEF, E′ was found to be correlated significantly with GLS and infarct size. From this aspect, our results are in agreement with those reported by Agati et al.[8] As considerable flow irregularity (high E′) was

Fig. 4

Local distributions of (a) phase-averagedE′and (b) velocity in the LVOT plane of a healthy, PNEF and PREF participant. The profile of a single patient represents his/her respective groups. The vortex is indicated by the white circular regions in (a). The comparison ofE′inside and outside of the vortical regions is shown in (c) with significant levels at ***P<0.001 and *P<0.05.E′, KE fluctuation index; LVOT, left ventricular outflow tract; PNEF, patients with normal ejection fraction; PREF, patients with reduced ejection fraction; TPS, time-to-peak systolic strain.

(8)

observed in the PNEF, but not the PREF group, they hypothesized thatE′is associated with the homogeneity in myocardial contraction as well as the magnitude of the flow velocity. However, for IHD patients, using TPS-SD as an indicator for myocardial dyssynchrony, we found that this is not the case. Instead of contraction inhomogeneity, the myocardial contraction strength itself (indicated by GLS) affectsE′. This may explain why we did not observe an increase inE′in the PNEF group compared with the healthy participants, even though the former showed a significant degree of contraction inhomogeneity. Our results bring into question the applicability of E′ in determining progressive LV failure in IHD patients at an early stage.

Second, we hypothesize that vortex KE may be useful to maintain LV function. This supports the statement made in a previous study that the appearance of a vortex in normal LV helps to minimize energy loss through DI and promotes LV ejection [24]. We theorize that a high LVEF is achieved by an energetic vortex (with high vortex KE) that induces a low-pressure region in the LV, and subsequently draws intraventricular blood flow towards the LVOT. Among the IHD patients (PNEF and PREF), an increase in vortex KE was associated with a significant reduction in DI. This is a direct consequence of two effects in the definition of DI (Eq. 3).

First, the denominator (KE in LV, which includes KE inside and outside of the vortex) increases with increas- ing velocities in the vortex. Second, the stronger flow in the bulk reduces velocity gradients as fluctuation approaches, thus reducing the numerator (frictional DI rate in LV) of Eq. 3. In other words, although the KE in LV increases, the fraction that is dissipated as heat, as indicated by DI, decreases.

The main limitations of our study are the small number of participants and female patients [25]. A larger cohort is required to further verify the preliminary results. Another limitation is that the analysis only involved the 2D in- plane flow field in an LVOT plane, which resembled the intraventricular gross flow feature. The entire LV flow feature may be attainable from the newly developed 4D flow imaging [26]. The PC-MRI technique has a relatively low temporal resolution (limited by the number of phases per R–R interval). However, this does not critically affect the result because the intraventricular flow was computed on the basis of individual cardiac phases.

Although the Phillips speckle-tracking strain used in this study is not the industry standard, a good concordance between the Philips and the GE strain analysis has been shown recently [27].

Conclusion

This is the first reported work to systematically collate and correlate all commonly used flow variables with performance indices of the LVEF function in IHD patients. It narrowed the pool of candidate indices to

E′and vortex KE, which correlated with GLS and LVEF, respectively. Vortex KE, in particular, could be inspected as an alternative to LV performance. The quantitative data also enabled a few hypotheses to be supported or refuted;

for example, interpreting the relationship betweenE′and LVEF. Furthermore, the results of this study suggest that the sensitivity of the imaging technique for velocities, for example, MRI versus echo-PIV, may play a part in determining the suitability of a particular flow variable, and hence, warrants follow-up studies. The inability to distinguish the PNEF from healthy participants, however, calls for more discerning indices, possibly beyond using flow variables.

Acknowledgements

The authors thank the technologists in the MRI and Echocardiography laboratories at University of Malaya Medical Centre for their assistance. They also thank Dr Ng Choung Min, Dr Henrik Haraldsson (UCSF, San Francisco) and Professor Gianni Pedrizzetti (University of Trieste, Italy) for informative communication.

This work was supported by a University of Malaya Research Grant (RP028A/B/C-14HTM).

Conflicts of interest

There are no conflicts of interest.

References

1 Bermejo J, Benito Y, Alhama M, Yotti R, Martinez-Legazpi P, Del Villar CP, et al. Intraventricular vortex properties in nonischemic dilated

cardiomyopathy.Am J Physiol Heart Circ Physiol2014;306:H718–H729.

2 Kilner PJ, Yang GZ, Wilkes AJ, Mohiaddin RH, Firmin DN, Yacoub MH.

Asymmetric redirection of flow through the heart.Nature2000;

404:759–761.

3 Sengupta PP, Khandheria BK, Korinek J, Jahangir A, Yoshifuku S, Milosevic I, et al. Left ventricular isovolumic flow sequence during sinus and paced rhythms–new insights from use of high-resolution Doppler and ultrasonic digital particle imaging velocimetry.J Am Coll Cardiol2007;49:899–908.

4 Elbaz MS, van der Geest RJ, Calkoen EE, de Roos A, Lelieveldt BP, Roest AA,et al. Assessment of viscous energy loss and the association with three‐dimensional vortex ring formation in left ventricular inflow: In vivo evaluation using four‐dimensional flow MRI.Magn Reson Med2017;

77:794–805.

5 Zhong Y, Liu Y, Wu T, Song H, Chen Z, Zhu W,et al. Assessment of left ventricular dissipative energy loss by vector flow mapping in patients with end-stage renal disease.J Ultrasound Med2016;35:965–973.

6 Dyverfeldt P, Kvitting JP, Sigfridsson A, Engvall J, Bolger AF, Ebbers T.

Assessment of fluctuating velocities in disturbed cardiovascular blood flow:

in vivo feasibility of generalized phase-contrast MRI.J Magn Reson Imaging 2008;28:655–663.

7 Zajac J, Eriksson J, Dyverfeldt P, Bolger AF, Ebbers T, Carlhall CJ. Turbulent kinetic energy in normal and myopathic left ventricles.J Magn Reson Imaging 2015;41:1021–1029.

8 Agati L, Cimino S, Tonti G, Cicogna F, Petronilli V, De Luca L,et al.

Quantitative analysis of intraventricular blood flow dynamics by echocardiographic particle image velocimetry in patients with acute myocardial infarction at different stages of left ventricular dysfunction.Eur Heart J Cardiovasc Imaging2014;15:1203–1212.

9 Chnafa C, Mendez S, Nicoud F. Image-based simulations show important flow fluctuations in a normal left ventricle: what could be the implications?

Ann Biomed Eng2016;44:3346–3358.

10 Einav S, Bluestein D. Dynamics of blood flow and platelet transport in pathological vessels.Ann N Y Acad Sci2004;1015:351–366.

11 Pasipoularides A. Mechanotransduction mechanisms for intraventricular diastolic vortex forces and myocardial deformations: part 1.J Cardiovasc Transl Res2015;8:76–87.

(9)

12 Hong GR, Kim M, Pedrizzetti G, Vannan MA. Current clinical application of intracardiac flow analysis using echocardiography.J Cardiovasc Ultrasound 2013;21:155–162.

13 Liew Y, McLaughlin R, Chan B, Aziz YA, Chee K, Ung N,et al. Motion corrected LV quantification based on 3D modelling for improved functional assessment in cardiac MRI.Phys Med Biol2015;60:2715.

14 Walker PG, Cranney GB, Scheidegger MB, Waseleski G, Pohost GM, Yoganathan AP. Semiautomated method for noise reduction and background phase error correction in MR phase velocity data.J Magn Reson Imaging1993;3:521–530.

15 Heiberg E, Sjögren J, Ugander M, Carlsson M, Engblom H, Arheden H.

Design and validation of segment-freely available software for cardiovascular image analysis.BMC Med Imaging2010;10:1.

16 Hong GR, Pedrizzetti G, Tonti G, Li P, Wei Z, Kim JK,et al. Characterization and quantification of vortex flow in the human left ventricle by contrast echocardiography using vector particle image velocimetry.JACC Cardiovasc Imaging2008;1:705–717.

17 Bird RB, Stewart WE, Lightfoot EN.Transport phenomena. New York, NY:

JohnWiley & Sons; 2002.

18 Pedrizzetti G, Sengupta PP. Vortex imaging: new information gain from tracking cardiac energy loss.Eur Heart J Cardiovasc Imaging2015;

16:719–720.

19 Charonko JJ, Kumar R, Stewart K, Little WC, Vlachos PP. Vortices formed on the mitral valve tips aid normal left ventricular filling.Ann Biomed Eng2013;

41:1049–1061.

20 Sundareswaran KS, Haggerty CM, De Zélicourt D, Dasi LP, Pekkan K, Frakes DH,et al. Visualization of flow structures in Fontan patients using 3-dimensional phase contrast magnetic resonance imaging.J Thorac Cardiovasc Surg2012;143:1108–1116.

21 Wong KK, Kelso RM, Worthley SG, Sanders P, Mazumdar J, Abbott D.

Cardiac flow analysis applied to phase contrast magnetic resonance imaging of the heart.Ann Biomed Eng2009;37:1495–1515.

22 Kim WY, Walker PG, Pedersen EM, Poulsen JK, Oyre S, Houlind K,et al. Left ventricular blood flow patterns in normal subjects: a quantitative analysis by three-dimensional magnetic resonance velocity mapping.J Am Coll Cardiol 1995;26:224–238.

23 Sengupta PP, Pedrizzetti G, Kilner PJ, Kheradvar A, Ebbers T, Tonti G,et al.

Emerging trends in CV flow visualization.JACC Cardiovasc Imaging2012;

5:305–316.

24 Pedrizzetti G, Domenichini F. Nature optimizes the swirling flow in the human left ventricle.Phys Rev Lett2005;95:108101.

25 Foll D, Taeger S, Bode C, Jung B, Markl M. Age, gender, blood pressure, and ventricular geometry influence normal 3D blood flow characteristics in the left heart.Eur Heart J Cardiovasc Imaging2013;14:366–373.

26 Callaghan FM, Kozor R, Sherrah AG, Vallely M, Celermajer D, Figtree GA, et al. Use of multi‐velocity encoding 4D flow MRI to improve quantification of flow patterns in the aorta.J Magn Reson Imaging2016;43:352–363.

27 Shiino K, Yamada A, Ischenko M, Khandheria BK, Hudaverdi M, Speranza V, et al. Intervendor consistency and reproducibility of left ventricular 2D global and regional strain with two different high-end ultrasound systems.Eur Heart J Cardiovasc Imaging2017;18:707–716.