Comparison of left ventricular hemodynamic forces measured by transthoracic echocardiography and cardiac magnetic resonance imaging in healthy adults

Abstract

Hemodynamic forces (HDF), which reflect the forces exchanged between blood and cardiac tissues, can be derived from cardiac magnetic resonance (CMR) or transthoracic echocardiography (TTE). Although normal values are reported for each imaging technique, no study has compared HDF values within the same cohort so far. We aimed to compare left ventricular (LV) HDF parameters obtained from CMR and TTE in healthy subjects. Twenty volunteers underwent both cine-CMR and 2D-TTE (within 7 days) at the Heart Center University Medical Center in Astana, Kazakhstan. Images were analyzed offline using dedicated software to extract standard volumetric, functional, strain, and HDF parameters: longitudinal (A-B) and transverse (L-S) HDF, L-S/A-B HDF ratio, and HDF vector angle. Statistical comparisons were performed with significance set at p < 0.05; Bland–Altman plots assessed agreement. TTE significantly underestimated LV volumes, ejection fraction, and global longitudinal strain compared to CMR. Similarly, HDF values were lower with TTE for both longitudinal and transverse forces (A-B HDF: 12.4 ± 3.4 vs. 26.1 ± 6.6; L-S HDF: 2.6 ± 1.2 vs. 5.2 ± 1.4; both p < 0.001). Bland–Altman analysis confirmed systematic underestimation of HDF by TTE. These findings suggest that TTE and CMR cannot be used interchangeably for HDF assessment, particularly in serial studies.

Introduction

The mechanical pumping performance of the left ventricle is a complex phenomenon that cannot be fully resolved by ejection fraction and Doppler indices of diastolic function. Strain imaging techniques measure myocardial deformation throughout the cardiac cycle and are also suitable for assessing subtle changes in myocardial contractility. However, deformation measurements are affected by loading conditions, ventricular geometry and conduction delays, which can limit their pathophysiological implications¹. Eventually, the primary driving force of blood flow is the development of intraventricular pressure gradients (IVPG). Therefore, analysis of LV hemodynamic forces (HDF), that correspond to the volume integral of the total value of IVPG, can resolve the quest of search for objective parameters of cardiac mechanics, where an abnormal deviation of IVPG from the physiological apex-to-base sequence represents unequivocal evidence of suboptimal myocardial function.

Currently, HDF can be assessed by analysing routine cardiac magnetic resonance (CMR) or transthoracic echocardiography (TTE) images using advanced software based on mathematical model (Fig. 1)^2,3. In theory, the parameters of the HDF should be similar, independent from the imaging technique. However, no study had yet compared HDF parameters in the same group of patients using both techniques. Accordingly, we sought to compare LV HDF parameters derived from two different imaging modalities (CMR and TTE) in a group of normal subjects.

Fig. 1.

Steps required for HDF analysis. Images are acquired using routine 2-D TTE or non-contrast CMR. Three apical views are selected and the endocardial border is traced. Tracking of the endocardial motion provides strain assessment. Aortic and mitral valve areas are calculated from the relative valve diameters. Finally, a mathematical model based on the first principle of fluid dynamics and mass conservation principles allows for HDF curve assessment and display. AV: aortic valve; MV: mitral valve.

Adapted from Salustri, A. et al. Left ventricular hemodynamic forces: gaining insight into left ventricular function. Explor. Cardiol. 3, 101,257 (2025), licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/)³

RESULTS

Demographic and physical parameters of the study population are reported in Table 1. The study cohort consisted of 13 males (65%) and 7 females (35%). The median age was 31 years (IQR: 27–34). There were no significant differences in HR and blood pressure values at the time of the two studies.

Table 1.

General characteristics of the study group.

Parameters	CMR	TTE
Age, yrs	31 (27; 34)
Weight, kg	74 (67; 80)
Height, cm	173 (167; 177)
BSA, m2	1.87 ± 0.21
HR, bpm	71 (67; 75)	70 (64; 75)
SBP, mmHg	117 (113; 122)	115 (110; 121)
DBP, mmHg	77 (73; 80)	76 (72; 79)
Gender, M/F (%)	13/7 (65)

BSA, Body Surface Area; DBP, diastolic blood pressure; HR, heart rate; SBP, systolic blood pressure. Data are expressed as median (IQR) or mean ± SD, as indicated.

Standard LV morphologic and functional parameters are reported in Table 2. Compared to TTE, CMR consistently showed larger LV volumes, particularly for end-diastolic volume (LVEDV: 121 mL [110; 129] vs. 87 mL [76; 94], p < 0.001; LVESV: 40 mL [36; 44] vs. 34 mL [31; 39], p = 0.028). Consequently, left ventricular ejection fraction (LVEF) was slightly higher with CMR than TTE (67% [64; 70] vs. 60% [56; 64], p < 0.001).

Table 2.

Standard morphologic and functional parameters.

Parameters	CMR	TTE	p value
LVEF, %	67 (64; 70)	60 (56; 64)	<0.001
LVEDV, mL	121 (110; 129)	87 (76; 94)	<0.001
LVEDVi, mL/m2	64.1 (43.3; 51.1)	46.8(43.3; 51.1)	<0.001
LVESV, mL	40 (36; 44)	34 (31; 39)	=0.028
LVESVi, mL/m2	20.9 (19.2; 22.8)	17.8 (16.7; 20.9)	=0.025
SV, mL	81 (69; 87)	53 (43; 59)	<0.001
SVi, mL/m2	42 ± 6.8	28.0 ± 5.5	<0.001
AAD, mm	20.2 ± 2.5	19.8 ± 1.7	=0.442
MAD, mm	33.3 ± 3.0	31.3 ± 3.5	=0.031

AAD: Aortic Annulus Diameter; LVESV: Left Ventricular End-Systolic Volume; LVESVi: Left Ventricular End-Systolic Volume indexed; LVEDV: Left Ventricular End-Diastolic Volume; LVEDVi: Left Ventricular End-Diastolic Volume indexed; MAD: Mitral Annulus Diameter; SV: Stroke Volume; SVi: Stroke Volume indexed. Data are expressed as median (IQR) or mean ± SD, as indicated.

Regarding myocardial strain parameters, global circumferential strain (GCS) was similar with CMR and TTE (−31.8 ± 4.5% vs. −31.3 ± 5.1%, p = 0.743). However, global longitudinal strain (GLS) was significantly lower with TTE compared to CMR (−24.5 ± 3.6% vs. −21.3 ± 3.9%, p = 0.012).

The HDF analysis results are summarized in Table 3 and an example of the HDF analysis of CMR and TTE images in a representative case is shown in Fig. 2. Both longitudinal and transverse HDF were significantly higher at CMR than at TTE. As a result, the L-S/A-B ratio and the impulse angle did not show significant differences between the two imaging modalities. The average values of LV volumes, dV/dt, and longitudinal HDF curves of the 20 subjects, both for CMR and TTE, are displayed in Fig. 3.

Table 3.

Hemodynamic forces parameters.

Parameters	CMR	TTE	p value
A-B, %	26.1± 6.6	12.4 ± 3.4	<0.001
L-S, %	5.2 ± 1.4	2.60 ± 1.2	<0.001
L-S/A-B ratio	20.7 ± 5.7	21.43 ± 9.3	=0.715
Impulse angle, degree	73.1 ± 2.9	71.90 ± 6.5	=0.351

A-B: Apex-to-Base HDF; L-S: Lateral-to-Septal HDF; L-S/A-B ratio: Ratio of Lateral-to-Septal to Apex-to-Base HDF; Impulse angle: Angle of the net HDF vector. Data are expressed as mean ± SD.

Fig. 2.

A representative case of HDF analysis derived from CMR (left) and TTE (right) images in the same subject. In each panel, the curves at the top represent longitudinal (red) and transverse (blue) HDF. The curves at the bottom represent LV volume (red) and dV/dt (blue). The polar histogram represents the direction (angle) of the HDF. In this subject, A-B HDF was 27.3% by CMR and 17.1% by TTE. L-S HDF was 5.4% and 2.5%, respectively

Fig. 3.

Averaged (± SD) values of LV volumes (panel a), dV/dt (panel b) and longitudinal HDF (panel c) curves derived from CMR (in blue) and TTE (in red) in the 20 normal volunteers. The higher LV volumes and the different slope of the volume curve at CMR compared to TTE result in different (wider) dV/dt curves and ultimately in higher values of longitudinal HDF

The Bland-Altman analysis comparing A-B HDF measurements from CMR and TTE (Fig. 4) showed a mean difference of 13.64% (95% CI: [10.78; 16.49]), indicating that TTE consistently underestimates A-B HDF compared to CMR (p < 0.001). The limits of agreement ranged from 0.85 to 26.43%, representing the interval within which 90% of the pairwise measurements between the two methods are expected to lie. The slope of the regression line of the measurement differences against their averages was 0.92 (p = 0.005), suggesting a significant proportional bias. This finding indicates that the level of disagreement between CMR and TTE increases progressively with higher A-B HDF values.

Fig. 4.

Bland-Altman plot for agreement between CMR and TTE for A-B HDF measurements

The Bland-Altman plot of L-S HDF (Fig. 5) measurements demonstrated a mean difference of 2.66% (95% CI: [1.94; 3.37]), reflecting a systematically lower L-S HDF value with TTE compared to CMR (p < 0.001). The limits of agreement ranged from − 0.54 to 5.85%, defining the range within which 95% of the differences between two methods are expected to fall. The regression line of the measurement differences against their average revealed no evidence of proportional bias (slope = 0.36, p = 0.325), suggesting that the difference between two methods remained consistent across the measurement range.

Fig. 5.

Bland-Altman plot for agreement between CMR and TTE for L-S HDF measurements

DISCUSSION

Fluid dynamics provide an alternative viewpoint to cardiac mechanics. By quantifying the pressure gradients driving blood flow, HDF assessment provides insights into ventricular mechanics beyond traditional metrics like ejection fraction and strain. Thus, HDF can be integrated to volumetric and deformation assessments to provide a further level of knowledge of cardiac mechanics and, possibly, indications of therapeutic outcome⁴.

Until recently, evaluation of HDF was possible only using the blood velocity recorded in the LV pool with 4D flow magnetic resonance or echocardiographic particle image velocitometry, hampering its widespread use. Recently, a mathematical model based on the mass conservation (velocity with zero divergence)² has made HDF analysis feasible from an integral over the LV boundary instead of internal volume, which allows for the assessment of the HDF from routine 2-D TTE or non-contrast cine-CMR. Both imaging modalities have been used for HDF analysis in several clinical conditions^5–22 and the results suggest that HDF could be a useful novel tool in cardiology, which can detect impaired cardiac physiology and provide an opportunity for medical intervention at an early stage.

Data on normal values of HDF are available for cine-CMR^8,17,23,24 and, to a lesser extent, for TTE^25,26. Ideally, normal values of HDF should be similar, independent from the imaging technique. However, although a direct comparison between the two imaging modalities cannot be derived from these studies, pooling data show statistically significant higher longitudinal HDF values for cine-CMR compared to TTE (19.34% [95% CI: 17.13–21.55%] vs. 14.93% [95% CI: 14.51–15.35%]; p < 0.001) (Fig. 6). Considerable heterogeneity was observed among the CMR studies (I² = 96.38%), reflecting substantial variability among those studies. Based on the results of these studies, a comparison of CMR and TTE for HDF analysis in the same group of (normal) subjects would be highly advisable.

Fig. 6.

Forest plot of subgroup meta-analyses of studies reporting longitudinal HDF assessed by CMR and TTE. Each square represents an individual study’s mean effect size, with the size of the square reflecting the relative weight of the study in the meta-analysis. Horizontal lines indicate 95% CI. Diamonds are the pooled mean effect estimates and their corresponding 95% CI for each method.

The present study is, to the best of our knowledge, the first head-to-head comparison of TTE and CMR for HDF assessment. First, we found that TTE significantly underestimates LVEDV and LVESV by 28% and 15%, respectively, which results in a relative 8% lower value of LVEF. These findings can be related to the different space resolution of the two methods and the risk of LV foreshortening with TTE, and are in line with previous studies where underestimation of LV volumes by up to 50% in comparison with CMR has been reported for TTE^27–32. Moreover, LV volumes we have found with TTE are consistent with the normal reference values provided by the British Society of Echocardiography and the American Society of Echocardiography^33,34. Secondly, in terms of myocardial deformation, no significant differences of GCS were found between the two imaging modalities, while GLS was significantly lower with TTE compared to CMR. These differences can be explained by the lower spatial resolution of TTE which can affect the accuracy of strain calculations³⁵. Lastly, the results of our study indicate that HDF are significantly different between CMR and TTE, with a systematic underestimation by TTE. The average values of the curves in all the subjects (Fig. 3) show a different slope of the LV volume curves during the systolic and early diastolic phases, which results in different dV/dt curves, with more negative values in systole and more positive values in early diastole. As a consequence, HDF curves show higher values with CMR compared to TTE during the same phases of the cardiac cycle. Bland-Altman charts indicate a systematic underestimation of both longitudinal and transverse HDF, with a proportional underestimation of longitudinal HDF, when TTE was compared to CMR. The agreement between the two methods was moderate, which may have relevant clinical implications. Thus, each center should have their own reference values and the same technique/equipment should be used in follow-up studies.

An additional factor that can explain the difference in HDF is the disproportionate degree of underestimation of LV volumes and valve areas between the two methods. Since mean velocity across mitral and aortic valves is evaluated by the LV volume rate divided by the valve area, the size of the valve should be reduced in scale in smaller ventricles to keep HDF constant. However, in our study aortic valve area was similar when assessed by the two methods, and we found only an 11% underestimation of mitral valve area compared to a 28% underestimation of LVEDV by TTE. This is not surprising, since linear measurements (i.e., valve diameters from which areas are derived) are much less affected by all the factors that hamper endocardial borders delineation and tracking, resulting in more consistent values of valve areas with the two imaging modalities.

HDF were analysed in a small group of young normal volunteers. Therefore, this study should be interpreted as a preliminary investigation due to the limited sample size, which may amplify the impact of inter-individual variability and reduce the generalizability of the findings. HDF values are also related to age and gender. Clearly, studies on larger sample sizes including different age groups and eventually different ethnicity are needed to confirm our findings. The results obtained with TTE are related to the ultrasound equipment we have used, therefore different HDF values could be obtained using more advanced echocardiographic systems with higher space resolution.

In conclusion, normal values of HDF are different according to the method used for image acquisition, with a consistent underestimation of TTE compared to CMR. Thus, the two imaging modalities cannot be mutually exchanged, in particular for follow-up studies of patients.

METHODS

Study population

Twenty healthy volunteers (age > 18 years) of Kazakh ethnic group without cardiovascular disorders and normal resting ECG were recruited among hospital employees, residents, and researchers’ acquaintances, and underwent a cine-CMR and 2D-TTE (maximum 7 days apart) at the Heart Center, University Medical Center in Astana, Kazakhstan. Exclusion criteria included a history of cardiovascular disease, diabetes, claustrophobia, other contraindications to CMR, or a body weight > 100 kg, which exceeded the technical limits of the CMR system.

Demographic information, physical parameters (height, weight, heart rate, and blood pressure), and clinical assessments were systematically obtained by proficient and certified personnel. Body surface area (BSA) was computed utilising the DuBois formula (0.20247 × height (m) 0.725 × weight (kg) 0.425). Informed consent was obtained from each subject participating in the studies.

Image acquisition

CMR acquisition

Non-contrast cine CMR were performed using a Siemens Magnetom Avanto 1.5 Tesla machine. The CMR scanning protocol included standard cardiac sequences recommended by the European Association of Cardiovascular Imaging (EACVI)³⁶. Details of the acquisition process have been already described elsewhere²³.

TTE acquisition

Two-dimensional TTE was performed using a Philips Affiniti 70 ultrasound system equipped with an X5-1 array transducer (1–5 MHz). The echocardiographic protocol was focused on standard apical views (2-, 3- and 4-Ch views) according to the EACVI recommendations³⁴. A parasternal long-axis view was obtained for measurement of the aortic valve diameter. All image acquisitions were optimized for temporal and spatial resolution to ensure high-quality data for post-processing.

Image analysis

All TTE and CMR images were anonymized prior to analysis using unique three-digit identification codes assigned to each participant. Investigators were blinded to all clinical information during image analysis to ensure objective and unbiased assessment. Image files were stored and retrieved from the institutional PACS in a de-identified format, and off-line analyses were performed using coded datasets to maintain blinding throughout the evaluation process.

Two observers (D.J. and A.Z.) with experience in cardiac imaging reviewed and analysed both cine-CMR and TTE datasets. For CMR, HDF were calculated using the three standard long-axis cine views and one short-axis view. TTE analysis was based on high-quality standard apical views. Images were analysed using a mathematical model integrated into a dedicated software (Q strain version 1.3.0.79, Medis, Leiden, the Netherlands). The calculation procedure is described in detail in Pedrizzetti et al.². Briefly, the total HDF vector, F(t), exchanged between blood and tissues at every instant during the cardiac cycle, is evaluated by the balance of momentum inside the LV volume:

1

where v(x, t) is the fluid velocity vector field measured at fixed points x at time t, S(t) is the closed surface bounding the volume, and n is the outward unit normal vector. This formula corresponds to the summation of values given by tissue position, velocity, and acceleration over the endocardial border. Therefore, it requires the knowledge of the velocity only over the endocardial boundary and across the valves, and does not require measuring the blood velocity inside the ventricular cavity. The time profile of the longitudinal component (Fz(t), apex-to-base) and of the main transversal component (Fx(t), lateral-septal) are thus obtained, and they are normalized by the instantaneous LV volume and the blood specific weight, to have a dimensionless number not directly affected by the chamber size. The global amplitude of the individual HDF components is described in terms of the root mean square (RMS) value of the corresponding time profile during the cardiac cycle.

HDF parameters

For the purpose of the study, the following HDF parameters were identified with both techniques:

HDF along the entire cardiac cycle:

1. Longitudinal (A-B) HDF - Apex-to-Base force (%).

2. Transverse (L-S) HDF - Lateral-to-Septal force (%).

3. L-S/A-B HDF ratio (representing the entity of the deviation of the HDF from the longitudinal direction).

4. Angle of the net HDF vector (degree).

In our laboratory, inter- and intra-observer variability for CMR-derived HDF is low (A-B HDF: ICC 0.85 [0.67–0.93] and 0.91 [0.78–0.93]; p < 0.001 for both; L-S HDF: ICC 0.86 [0.69–0.94] and 0.93 [0.83–0.97]; p < 0.001 for both)³⁷. For TTE-derived HDF, inter- and intra-observer variability is also low (A-B HDF: ICC 0.72 [0.08–0.92]; p = 0.022 and 0.92 [0.72; 0.98]; p < 0.001; L-S HDF: ICC 0.90 [0.65–0.97]; p < 0.01, and 0.92 [0.70; 0.98]; p < 0.001).

Statistical analysis

The distribution of continuous variables was assessed using the Shapiro–Wilk test. Depending on normality, either the paired t-test or Wilcoxon signed-rank test was applied to compare measurements obtained from CMR and TTE. Bland–Altman plots were constructed to assess agreement between CMR and TTE measurements for A-B HDF and L-S HDF. Statistical analyses were performed using Stata 18.0 (StataCorp, College Station, TX), with a significance level set at p < 0.05.

Author contributions

AZ: Design of the methodology, data analysis, writing the manuscript; GT: Data analysis, data interpretation; ZK: Data curation, formal analysis, statistics;; BT: Provision of patients, investigation, literature search; DJ: Data analysis; MB: Provision of patients, instrumentation, computing resources; NK: Data collection and analysis; TD: Provision of patients, instrumentation, computing resources; AS: Conceptualization, design of methodology, supervision, writing-review & editing. All authors had full access to the underlying data. All authors read and approved the final manuscript.

Funding

This study was funded by a grant of the Ministry of Science and Higher Education of the Republic of Kazakhstan, № AP23490021.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.

Ethics Approval

The Institutional Research Ethics Committee of the Nazarbayev University approved this study (Approval number: Nr. 901/13052024), conducted following the Declaration of Helsinki. Compliance with ethical guidelines was ensured by obtaining written informed consent from all participants before enrollment. Data confidentiality was guaranteed and each subject was assigned an ID number. Hard copy data collected or used during this study were stored in a cabinet in the locked office of the study personnel, and all electronic data were stored on password-protected computers of the study personnel in their locked offices and in the locked personnel building. Only essential research personnel had access to any of these patient files.