Ultrasound Obstet Gynecol 2011; 37: 150–157 Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/uog.8815
Temporal and spatial performance of vector velocity imaging in the human fetal heart H. MATSUI*, I. GERMANAKIS*, E. KULINSKAYA† and H. M. GARDINER*‡ *Department of Reproductive Biology, Division of Cancer, Faculty of Medicine, Imperial College at Queen Charlotte’s and Chelsea Hospital, London, UK; †Statistical Advisory Service, Imperial College, London, UK; ‡Department of Paediatric Cardiology, Royal Brompton and Harefield NHS Foundation Trust Hospital, London, UK
K E Y W O R D S: fetus; myocardial strain; speckle tracking; tissue Doppler velocities; vector velocity imaging
ABSTRACT Objectives To assess the spatial and temporal performance of fetal myocardial speckle tracking, using highframe-rate (HFR) storing and Lagrangian strain analysis. Methods Dummy electrocardiographic signaling permitted DICOM HFR in 124 normal fetuses and paired low-frame-rate (LFR) video storing at 25 Hz in 93 of them. Vector velocity imaging (VVI) tracking co-ordinates were used to compare time and spatial domain measures. We compared tracking success, Lagrangian strain, peak diastolic velocity and positive strain rate values in HFR vs. LFR video storing. Further comparisons within an HFR subset included Lagrangian vs. natural strain, VVI vs. M-mode annular displacement, and VVI vs. pulsed-wave tissue Doppler imaging (TDI) peak velocities. Results HFR (average 79.4 Hz) tracking was more successful than LFR (86 vs. 76%, P = 0.024). Lagrangian and natural HFR strain correlated highly (left ventricle (LV): r = 0.883, P < 0.001; right ventricle (RV): r = 0.792, P < 0.001) but natural strain gave 20% lower values, suggesting reduced reliability of measurement. Lagrangian HFR strain was similar in LV and RV and decreased with gestation (P = 0.015 and P < 0.001, respectively). LV Lagrangian LFR strain was significantly lower than the values for the RV (P < 0.001) and those using paired LV-HFR recordings (P = 0.007). Annular displacement methods correlated highly (LV = 1.046, r = 0.90, P < 0.001; RV = 1.170, r = 0.88, P < 0.001). Early diastolic waves were visible in 95% of TDI, but in only 26% of HFR and 0% of LFR recordings, and HFR-VVI velocities were significantly lower than those for TDI (P < 0.001). Conclusions Doppler estimation of velocities remains superior to VVI but image gating and use of original co-ordinates should improve offline VVI assessment of
fetal myocardial function. Copyright 2011 ISUOG. Published by John Wiley & Sons, Ltd.
INTRODUCTION Myocardial functional assessment has technical limitations when applied to the fetus, including the need for considerable expertise in data acquisition, reduced reproducibility by the requirement for correct alignment to the structures being interrogated and lack of a concurrent electrocardiogram (ECG) to time mechanical events accurately1 . Furthermore, most noninvasive methods are load dependent2 – 5 while optimal, relatively load-independent Doppler measures such as isovolumic acceleration have reduced reproducibility6 . Tissue Doppler imaging (TDI) technologies range from pulsedwave Doppler assessment of long axis myocardial velocities, to more complex color Doppler techniques1,7 – 10 . Tissue Color Doppler records velocities 10–20% lower than pulsed-wave Doppler in the adult heart11 and is associated with low reproducibility in the human fetus, because of large pixel size relative to small fetal myocardial tissue volumes12,13 . Non-Doppler methods based on speckle-tracking techniques, including vector velocity imaging (VVI), use frame by frame tracking of myocardial speckles in twodimensional (2D) echocardiographic images to provide velocity and displacement data and permit quantification of myocardial deformation or strain14,15 . Whereas 2D fetal echocardiography is based on unidimensional border detection (motion perpendicular to the echo beam and parallel to the beam in TDI), VVI allows assessment of function independent of fetal lie, thus shortening examination time and simplifying more sophisticated measures such as deformation and shear stress14 – 16 . Fetal speckle-tracking VVI software has been used to analyze
Correspondence to: Dr H. M. Gardiner, Faculty of Medicine, Institute of Reproductive and Developmental Biology, Imperial College, Du Cane Road, London, W12 0HS, UK (e-mail:
[email protected]) Accepted: 18 August 2010
Copyright 2011 ISUOG. Published by John Wiley & Sons, Ltd.
ORIGINAL PAPER
Fetal speckle tracking
151
images, stored at standard frame rates (25–30 Hz), referred to here as low frame rate (LFR), to measure point or natural strain17 (and is the system reported here) while other machines permit high-frame-rate (HFR) stored images for global 2D strain measurements using techniques such as automatic functional imaging18 . There is considerable variation in the reported VVI fetal point strain within each segment19 , and small size and low temporal resolution of imaging in the fetal heart, particularly in systems without true DICOM video loop storing, may be responsible. The aim of this study was to compare the performance of speckle-tracking image analysis using Syngo VVI v1 software (Siemens Healthcare, Erlangen, Germany) in the normal fetal heart in the second and third trimesters using standard and modified video loop storing. Specifically we compared measurements in the spatial and temporal domains following LFR and HFR and compared HFRVVI with Doppler and M-mode. Secondly we compared natural strain calculated automatically by the Syngo VVI software with Lagrangian strain calculated from the X,Y speckle-tracking co-ordinates and created gestational reference ranges for Lagrangian strain.
METHODS We prospectively recorded four-chamber views of the heart in fetuses with normal hearts in a cohort of 144 women with singleton pregnancies, of gestational age 14 + 1 to 39 + 1 weeks; Figure 1 is a flow chart of the cohort numbers used in the analysis. In all cases
gestational age had been determined by measurement of first-trimester crown–rump length. We excluded women with known systemic disease (such as diabetes), multiple pregnancy or those whose fetuses had extracardiac malformations, aneuploidy or growth restriction. All studies were performed by one examiner using Acuson Sequoia ultrasound systems (Siemens Healthcare) with a 6–2-multi-Hertz curvilinear probe. The four-chamber image was optimized to achieve high contrast between the myocardial walls and cavity and the clearest loop of each image was stored digitally, in DICOM format, in two ways sequentially: at the standard frame rate (25 Hz) for fetal video loop storing (LFR), where ECG triggering is not available, and using our modification to acquire the original (variable) HFR. To enable HFR storing, we used a commercial metronome (BOSS DB-60, Roland Corp, Japan), attached to the line input of the ultrasound system. The metronome’s dummy spike-signals, set to 60 beats per min, allowed storing at original frame rate (HFR) of two virtual cycles (2 s) DICOM video loops. HFR loops were obtained prospectively and in consecutive fetuses with normal cardiac anatomy and LFR loops were stored immediately after this in the same plane without the metronome signal. All cases with both LFR and HFR loops were used in paired studies of strain. Pulsed-wave TDI and M-mode long axis annular displacement information were recorded in a subset of fetuses for which fetal lie was appropriate and clinic time available. The institutional review board deemed that ethical approval was unnecessary because the evaluation was an integral part of routine clinical visits in which the
Eligible studies: both HFR and LFR storage for VVI speckle tracking (n = 144) Excluded: VVI tracking failure: both HFR and LFR (n = 4) HFR only (n = 16) LFR only (n = 31) Paired HFR and LFR-VVI studies (n = 93) (35 with paired M-mode and TDI data)
Temporal analysis
Spatial analysis
Natural vs. Lagrangian strain HFR storage (n = 24)
HFR vs. LFR storage Lagrangian strain (n = 93)
HFR-VVI vs. M-mode displacement (n = 35)
HFR vs. LFR-VVI Velocity (peak diastolic) Strain rate (peak positive) (n = 81)
HFR-VVI vs. TDI Peak diastolic and systolic velocities (n = 35)
Figure 1 Flow chart showing cohort numbers used in the comparisons made in the study. HFR, high frame rate; LFR, low frame rate; TDI, tissue Doppler imaging; VVI, vector velocity imaging.
Copyright 2011 ISUOG. Published by John Wiley & Sons, Ltd.
Ultrasound Obstet Gynecol 2011; 37: 150–157.
Matsui et al.
152
mother had consented to the examination. All scans were anonymized before analysis.
Offline VVI tracking The DICOM digital images were exported from the ultrasound system to optical disks and imported into Syngo Work Station using Syngo VVI version 1. The fetal heart rate was determined by the time interval between two consecutive end diastolic frames defined by mitral valve closure, and we selected a single heartbeat from each HFR and LFR acquisition for analysis. The endocardial border was tracked manually in end diastole and tracking curves were created automatically by the software in subsequent frames. We allowed 10 min for tracking and considered recordings taking longer unsuitable for analysis. We positioned a reference point at a point furthest from the annulus along the tracking line to assess the velocity of tracking points (Figure 2). In VVI analysis the software provides an uneven number of tracking points, assuming the median point is the apex of the heart17 , but this is unreliable and was checked before making the calculation. The X,Y tracking co-ordinates were available from a database accessible within the program and used in calculations of Lagrangian strain and annular displacement. We considered annular displacement and strain as variables dependent predominantly on the quality of 2D acquisition (spatial domain parameters) and myocardial velocities and strain rate to be dependent predominantly on the quality of temporal resolution (temporal domain parameters).
LFR vs. HFR analysis We compared evaluation of myocardial velocities and strain rate in paired HFR and LFR loops in 35 fetuses from the cohort. VVI does not measure peak velocity but
the mean of the velocities of two tracked points so we compared the effect of frame rate acquisition on the highest velocities (peak diastolic velocities) in the fetal cardiac cycle to ensure that any differences would be more easily recognized. We compared values of peak positive strain rate (PPSR) at the base of the heart by LFR and HFR to assess the effect of frame rate on the software calculation of PPSR. The proportion of cases with visible isovolumic acceleration peak and biphasic diastolic wave by VVI analysis was documented.
Longitudinal myocardial strain Strain describes the stretching and compression of cardiomyocytes relative to their original length and can be calculated in two ways; Lagrangian strain is used more commonly in magnetic resonance imaging and natural or point strain in adult echocardiography. Lagrangian strain is calculated from the entire length of the free wall from base to apex rather than taking the average of individual small or point strain values as in natural strain18,20 . The mathematics of both methods differ as shown in Appendix S1. Overall free wall lengths were calculated from the database of X,Y coordinates of the speckletracking points and the mean of three recordings taken to calculate Lagrangian strain (Figure 3). Natural strain was calculated automatically by the speckle-tracking software based on instantaneous velocity differences. The basal, apical and midwall regions along the free wall and septum of each ventricle produce six segmental values of strain per ventricle and the averaged value of all six is termed ‘global strain’19 . We performed a paired comparison of HFR and LFR free wall Lagrangian strain values and assessed reproducibility of HFR strain measurements, using a second investigator to trace the endocardial border of the same cardiac cycle and remeasure strain from 21 randomly selected cases. We compared Lagrangian and natural strain in a subgroup of fetuses and created gestational age reference ranges of Lagrangian strain for right ventricular (RV) and left ventricular (LV) walls from HFR of 124 single observations made between 14 + 1 and 39 + 1 weeks’ gestation (Figure 4).
Comparison of annular displacement of the atrioventricular ring
End systole
End diastole
Figure 2 Schematic diagram showing positioning of apical reference points for recording annular velocities. Each apical ) of the mitral reference point (⊗) is positioned along the line ( () and tricuspid () annular movements.
Copyright 2011 ISUOG. Published by John Wiley & Sons, Ltd.
The magnitude of displacement was compared using M-mode and HFR-VVI, in a subgroup of 35 fetuses (Figure 5). Maximal longitudinal displacement was recorded from the four-chamber view of the heart by placing the M-mode cursor at right angles to the mitral and tricuspid valve rings1 . The mean of the maximal systolic displacement from three good-quality sequential registrations was taken. The maximal displacement recorded by HFR-VVI was calculated from the X,Y tracking coordinates using the following formula: Dis.max (mm) = PD × {(Xend diastole − Xend systole )∧ 2 + (Yend diastole − Yend systole )∧ 2} ∧ 1/2,
Ultrasound Obstet Gynecol 2011; 37: 150–157.
Fetal speckle tracking
153
0
gate in the free walls at the base of the heart and the mean of three good-quality sequential recordings taken1 .
L1(0) 1
L2(0)
Statistical analysis
2
L3(0) 0
L1(t) L2(t) L3(t)
The data were analyzed using SPSS v16 (SPSS Inc., Chicago, IL, USA) and R (The R Foundation, Vienna, Austria)21 . Exploratory analysis and Shapiro–Wilks test were used to assess normality of the data. Wilcoxon signed ranks test and sign test were used for paired comparisons of temporal and spatial resolution. Spearman’s correlation coefficient was used to assess correlations between continuous measurements. Robust regression analysis (procedure lmrob from the ‘robust’ R package) was used to determine the relationship between myocardial strain and gestational age, and P < 0.05 was regarded as statistically significant. Bland–Altman analysis and intraclass correlation were used to assess interobserver reproducibility.
3 1
2 3
Ln−1(t) Ln(t) n n
Ln(0)
Ln−1(0)
Figure 3 Schematic diagram illustrating calculation of Lagrangian myocardial strain. In vector velocity imaging analysis the software provides tracking points (open circles 0– n) in each frame of the two-dimensional echocardiographic image. The length of the free wall is calculated from the summation of the small segments between consecutive tracking points (L1 (t) to Ln (t)). Lagrangian myocardial strain is calculated from the ratio of change of the free wall length. , tracking line in end diastole; , tracking line at time t.
where Dis.max is the maximal displacement, PD is the pixel dimension (mm/pixel) and X and Y are the coordinates of each speckle-tracking point.
Comparison of VVI and TDI at the atrioventricular ring We compared the basal peak systolic and diastolic myocardial velocities recorded sequentially using myocardial TDI and HFR-VVI in the same 35 fetuses. TDI velocities were recorded by placing a 2-mm pulsed Doppler (a)
RESULTS High and standard frame rate storing Ninety-three paired observations of speckle tracking acquired at LFR and HFR were analyzed (Table 1). Speckle tracking was more successful in HFR than LFR (86 vs. 76%; P = 0.024, chi-square test) and equally successful in both RV and LV using both methods of acquisition.
Frame rates and Lagrangian strain Table 2 summarizes the comparative results. Strain was similar in both ventricles using HFR storing, but mean LV strain was significantly lower than mean RV strain following LFR storage. HFR strain decreased significantly with gestational age (RV, P < 0.001; LV, P = 0.015) while Lagrangian strain stored using LFR decreased only in LV (P = 0.040). We report results from HFR storage. Interobserver variability of Lagrangian strain was (b) 0.3 Right ventricular strain
Left ventricular strain
0.3
0.2
0.1 10
20
30
40
Gestational age (weeks)
0.2
0.1 10
20
30
40
Gestational age (weeks)
Figure 4 Graphs of reference ranges for Lagrangian strain showing the relationship between peak negative strain of free wall of the left ventricle (a) and right ventricle (b) and gestational age. Mean and 95% reference ranges are shown.
Copyright 2011 ISUOG. Published by John Wiley & Sons, Ltd.
Ultrasound Obstet Gynecol 2011; 37: 150–157.
Matsui et al.
154 (a)
Table 1 Comparison of high-frame-rate (HFR) and low-frame-rate (LFR) storing for offline analysis
End diastole
Heart rate (bpm)* End systole
Frame rate per second Frame rate per beat Biphasic diastolic wave
HFR storing
LFR storing
145.8 ± 9.7 (120.0–169.5) 79.4 ± 29.4 (27.4–167.2) 32.5 ± 11.4 (11.0–67.1) 44/171 (26)
145.8 ± 9.7 (120.0–169.5) 25 10.3 ± 0.7 (8.9–12.5) 0 (0)
Data given as mean ± SD (range), absolute value or n (%). *Data identical for HFR and LFR as the same initial loops were stored using each technique. Table 2 Effect of frame rate storing on Lagrangian strain Lagrangian strain (%) Ventricle RV LV
HFR storing (n = 93)
LFR storing (n = 93)
P
−22.3 (−14.2 to −30.1) −21.6 (−13.5 to −31.0)
−23.2 (−4.5 to −42.5) −19.6 (−7.3 to −39.0)
0.289 0.007
Data given as mean (range). HFR, high-frame-rate; LFR, low-frame-rate; LV, left ventricle; RV, right ventricle. Table 3 Paired comparison of high-frame-rate (HFR) vs. low-frame-rate (LFR) storing for the calculation of time domain parameters using vector velocity imaging
Diastole
Systole
Figure 5 Schematic diagram (a) and M-mode echocardiography (b) illustrating measurement of maximal displacement of the atrioventricular annulus, i.e. distance between end-systole and ) calculated from X-Y coordinates. end-diastole (
HFR storing
LFR storing
Pairs (n)
P*
79 82
< 0.001 < 0.001
79 81
< 0.001 < 0.001
Peak diastolic velocity at AV ring (cm/s) 2.86 ± 1.07 1.53 ± 0.68 LV 3.08 ± 1.46 1.64 ± 0.69 RV Peak positive strain rate at base (1/s) LV 4.08 ± 1.43 1.95 ± 0.80 3.60 ± 1.90 1.77 ± 0.77 RV
Data shown as mean ± SD or n. *Wilcoxon signed rank test. AV, atrioventricular; LV, left ventricle; RV, right ventricle.
Comparison of natural and Lagrangian strain acceptable: SD 1.5% for LV and 1.9% for RV strain with nearly all observations lying within two SDs on the Bland–Altman plot. Intraclass correlation coefficients were high at 0.928 (95% CI, 0.832–0.970) for LV and 0.894 (95% CI, 0.768–0.954) for RV strain.
Frame rates and velocities and strain rates Both peak diastolic velocities and peak positive Lagrangian strain rates were significantly higher at HFR (Table 3). Bland–Altman analysis demonstrated an increase in difference between LFR and HFR velocities with increasing values.
Copyright 2011 ISUOG. Published by John Wiley & Sons, Ltd.
There was a strong correlation between the methods for both LV (rS = 0.883, P < 0.001) and RV (rS = 0.792, P < 0.001) but the wider spread of natural strain values in RV resulted in relative differences ((Lagrangian RV strain − natural RV strain)/Lagrangian RV strain)) of up to 20%, suggesting reduced reliability of natural strain. Unlike measures of Lagrangian strain, which decreased in both ventricles, measures of natural strain showed a decrease with gestational age only in RV (P = 0.018).
Reference ranges for Lagrangian strain Variables considered for inclusion in the model included heart rate and frame rate (quality of speckle tracking
Ultrasound Obstet Gynecol 2011; 37: 150–157.
Fetal speckle tracking
155
represented by frames per beat). As these both correlated with gestational age their inclusion alongside gestational age did not add to the quality of the regression model. Reference ranges for strain were created using robust regression (Figure 4): LV strain = 2.502E−1 − 1.663E−4 × GA RV strain = 2.766E−1 − 3.041E−4 × GA where GA = gestational age in days. The 95% reference ranges are calculated as above ± 1.96 × 0.038, where 0.038 is the robust residual standard error for both models.
Comparison of TDI and VVI Peak systolic and diastolic basal myocardial velocities were significantly higher using TDI (Table 4). Additionally a biphasic diastolic wave was present in more than 95% of TDI recordings, but in only 26% of HFR and no LFR-VVI recordings (Table 1). The spike of the isovolumic acceleration was almost invariably seen in TDI waveforms but never in VVI traces, even at HFR.
Comparison of annular displacement using VVI and M-mode echocardiography There was strong correlation (LV slope = 1.046, r = 0.90, P < 0.001; RV slope = 1.170, r = 0.88, P < 0.001) (Figure 6) and no significant difference between these methods in either ventricle.
DISCUSSION Previous reports have demonstrated the feasibility of VVI speckle tracking in the fetus. Because the technique is apparently simple and offline strain analysis automatic, fetal strain and strain rate have been reported without validation of the method18,19 . Our novel approach was to use a dummy ECG signal during image acquisition, which enabled us to store DICOM images at the original HFR, and our study design allowed for paired analysis of data obtained from the same fetus stored sequentially at HFR followed by standard video at LFR. Moreover, Table 4 Paired comparison of time domain parameters evaluated using vector velocity imaging with high-frame-rate storing (HFR-VVI) vs. tissue Doppler imaging (TDI) HFR-VVI
TDI
Peak systolic velocity at AV ring (cm/s) 2.28 ± 0.77 4.20 ± 1.02 LV 3.04 ± 0.84 5.78 ± 1.64 RV Peak diastolic velocity at AV ring (cm/s) 2.78 ± 0.99 8.61 ± 2.14 LV 3.86 ± 1.48 10.03 ± 2.39 RV
Pairs (n)
P*
28 29
< 0.001 < 0.001
28 29
< 0.001 < 0.001
Data shown as mean ± SD or n. *Wilcoxon signed rank test. AV, atrioventricular; LV, left ventricle; RV, right ventricle.
Copyright 2011 ISUOG. Published by John Wiley & Sons, Ltd.
we compared the X,Y tracking co-ordinates in our calculations as well as the values generated automatically by the Syngo software to assess whether the two were sufficiently similar to give the same clinical information. In accord with others, we confirmed that speckle tracking is feasible in about 85% of cases14,17,18,22 . While Lagrangian and natural strain methods gave similar mean values for the RV but not LV free wall and the measurements correlated strongly, the segmental point strain values provided by the Syngo v1 software varied widely, suggesting reduced reliability of natural strain measurements. Some have proposed that differing measures of ventricular strain may contain physiological information, perhaps related to different segmental fiber orientation23 , but as natural strain is derived from adjacent myocardial segments at distances often less than the discriminative capacity of commonly used ultrasound probes, we believe the wide spread of natural strain values questions whether natural strain provides robust and valid information in the evaluation of fetal myocardial function. Our concern regarding the loss of peak strain data from LFR-VVI analysis was confirmed, at least in the LV, which is in agreement with other studies17 , and as both RV and LV strain were similar using HFR we conclude that the interventricular differences reported previously in the fetus may be due in part to lower frame rate storage and may not reflect physiological differences. Most work on strain in the fetus7,17,18,24 , children25 and animal models26,27 reports no change, although an increase with advancing gestation was reported by one group12 . Our results suggest that the real decrease in strain with gestational age may have been missed using the LFR provided in some systems. We would anticipate a change in myocardial strain with gestation as it is a loaddependent index estimating cardiomyocyte shortening28 , and as loading conditions and myocardial function alter with gestation, so might strain. Therefore our results lead us to question the conclusions drawn by others that intrinsic myocardial properties such as wall stress are established early in pregnancy and strain remains constant in the second and third trimesters17,18 . We compared VVI with pulsed-wave TDI (which measures peak instantaneous myocardial velocities at high temporal resolution of 250–300 samples/s) and M-mode (that samples at 3–4 ms5 ). We have previously published reference ranges for long-axis function using these techniques in the fetus1 . Relatively low frame rates will reduce the capacity of VVI to record peak myocardial velocities but it is also important to recognize that speckle tracking measures the mean modal velocity from the X and Y coordinates and not the peak velocity. So while our modified technique allowed us to analyze 33 frames per fetal heart beat on average compared with 10 available frames per beat using standard video storing, resulting in significantly higher myocardial velocities, the performance of VVI was inferior to TDI, which could demonstrate biphasic diastolic early and late waveforms and the short 50 ms spike
Ultrasound Obstet Gynecol 2011; 37: 150–157.
Matsui et al. (a) 8
(b) 8
Maximal displacement (mm): M-mode
Maximal displacement (mm): M-mode
156
6
4
2
0
0
2
4
6
Maximal displacement (mm): VVI
6
4
2
0
0
2
4
6
Maximal displacement (mm): VVI
Figure 6 Comparison of maximal displacement using vector velocity imaging (VVI) and M-mode echocardiography in the mitral annulus (a) (y = 1.046x (r = 0.90, P < 0.001)) and tricuspid annulus (b) (y = 1.170x (r = 0.88, P < 0.001)).
of isovolumic acceleration in almost all recordings. Biphasic diastolic profiles were only seen in VVI storing above 30 frames per beat but no isovolumic acceleration waveform was present. This suggests that frame rates are too low using most current speckle-tracking algorithms in the fetal heart, particularly if harmonic imaging is required to obtain good-quality 2D images later in gestation14 . An important limitation of this study is that it is not possible to compare VVI with an invasive technique that provides concurrent information on human fetal cardiac loading conditions or to record pressure–volume loops as a ‘gold standard’4,28 . LFR velocities are lower than HFR velocities, but as they are obtained with a standard sampling rate they might represent a constant fraction of the underlying myocardial velocities, in contrast to HFR velocities, which are obtained with a higher but variable sampling rate. As peak HFR velocities could vary according to the sampling rate, care should be taken to ensure that the frame rate between consecutive HFR studies is similar in serial studies of function. Although the feasibility of HFR-VVI in our prospective (but not strictly consecutive) cohort of normal fetuses was good, lower performance has been reported when imaging conditions are unfavorable, including fetuses with cardiac decompensation and polyhydramnios29 , which were excluded in this study. However, as all available cases, irrespective of image quality, were stored for further analysis we believe no selection bias was present in our study and that HFR-VVI tracking should be successful in the majority of normal fetuses. Although Syngo VVI allows functional measurements in any orientation, there are certain pitfalls in the assessment of fetal cardiac function. Gating will increase frame rate storage and reliability of measurements and the use of the tracking points to assess Lagrangian, rather than point strain calculated from LFR acquisition, will improve the
Copyright 2011 ISUOG. Published by John Wiley & Sons, Ltd.
reliability of offline assessment of myocardial function in the fetus.
ACKNOWLEDGMENTS This work was supported by the Tiny Tickers charity, www.tinytickers.org, the Richard and Jack Wiseman Trust, the NIHR Biomedical Research Centre funding scheme and the Institute of Obstetrics and Gynaecology Trust Fund, Queen Charlotte’s and Chelsea Hospital, London, UK.
REFERENCES 1. Gardiner HM, Pasquini L, Wolfenden J, Li W, Kulinskaya E, Henein MY. Myocardial tissue Doppler and long axis function in the fetal heart. Int J Cardiol 2006; 113: 39–47. 2. Fogel MA, Weinberg PM, Gupta KB, Rychik J, Hubbard A, Hoffman EA, Haselgrove J. Mechanics of the single left ventricle: a study in ventricular–ventricular interaction II. Circulation 1998; 98: 330–338. 3. Suga H, Sagawa K, Shoukas AA. Load independence of the instantaneous pressure-volume ratio of the canine left ventricle and effects of epinephrine and heart rate on the ratio. Circ Res 1973; 32: 314–322. 4. Derrick GP, Narang I, White PA, Kelleher A, Bush A, Penny DJ, Redington AN. Failure of stroke volume augmentation during exercise and dobutamine stress is unrelated to load-independent indexes of right ventricular performance after the Mustard operation. Circulation 2000; 102: 154–159. 5. Van de Veire NR, De Sutter J, Bax JJ, Roelandt JR. Technological advances in tissue Doppler imaging echocardiography. Heart 2008; 94: 1065–1074. 6. Nii M, Shimizu M, Roman KS, Jaeggi ET. Doppler tissue imaging in the assessment of atrioventricular conduction time: validation of a novel technique and comparison with electrophysiologic and pulsed wave Doppler-derived equivalents in an animal model. J Am Soc Echocardiogr 2006; 19: 314–321. 7. Perles Z, Nir A, Gavri S, Rein A. Assessment of fetal myocardial performance using myocardial deformation analysis. Am J Cardiol 2007; 99: 993–996.
Ultrasound Obstet Gynecol 2011; 37: 150–157.
Fetal speckle tracking 8. Nii M, Roman KS, Kingdom J, Redington AN, Jaeggi ET. Assessment of the evolution of normal fetal diastolic function during mid and late gestation by spectral Doppler tissue echocardiography. J Am Soc Echocardiogr 2006; 19: 1431–1437. 9. Chan LY, Fok WY, Wong JT, Yu CM, Leung TN, Lau TK. Reference charts of gestation-specific tissue Doppler imaging indices of systolic and diastolic functions in the normal fetal heart. Am Heart J 2005; 150: 750–755. 10. Harada K, Tsuda A, Orino T, Tanaka T, Takada G. Tissue Doppler imaging in the normal fetus. Int J Cardiol 1999; 71: 227–234. 11. Kukulski T, Voigt JU, Wilkenshoff UM, Strotmann JM, Wranne B, Hatle L, Sutherland GR. A comparison of regional myocardial velocity information derived by pulsed and color Doppler techniques: an in vitro and in vivo study. Echocardiography 2000; 17: 639–651. 12. Di Salvo G, Russo MG, Paladini D, Pacileo G, Felicetti M, Ricci C, Cardaropoli D, Palma M, Caso P, Calabro R. Quantification of regional left and right ventricular longitudinal function in 75 normal fetuses using ultrasound-based strain rate and strain imaging. Ultrasound Med Biol 2005; 31: 1159–1162. 13. Larsen LU, Petersen OB, Norrild K, Sorensen K, Uldbjerg N, Sloth E. Strain rate derived from color Doppler myocardial imaging for assessment of fetal cardiac function. Ultrasound Obstet Gynecol 2006; 27: 210–213. ¨ 14. Suhling M, Jansen C, Arigovindan M, Buser P, Marsch S, Unser M, Hunziker P. Multiscale motion mapping: a novel computer vision technique for quantitative, objective echocardiographic motion measurement independent of Doppler: first clinical description and validation. Circulation 2004; 110: 3093–3099. 15. Helle-Valle T, Crosby J, Edvardsen T, Lyseggen E, Amundsen BH, Smith HJ, Rosen BD, Lima JA, Torp H, Ihlen H, Smiseth OA. New non-invasive method for assessment of left ventricular rotation. Speckle tracking echocardiography. Circulation 2005; 112: 3149–3156. 16. Burns AT, McDonald IG, Thomas JD, MacIsaac A, Prior D. Doing the twist: new tools for an old concept of myocardial function. Heart 2008; 94: 978–983. 17. Younoszai AK, Saudek DE, Emery SP, Thomas JD. Evaluation of myocardial mechanics in the fetus by velocity vector imaging. J Am Soc Echocardiogr 2008; 21: 470–474. 18. Ta-Shma A, Perles Z, Gavri S, Golender J, Tarshansky S, Shlichter C, Bar Tov H, Rein AJ. Analysis of segmental
157
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
and global function of the fetal heart using novel automatic functional imaging. J Am Soc Echocardiogr 2008; 21: 146–150. Barker PCA, Houle H, Li JS, Miller S, Herlong JR, Camitta MGW. Global longitudinal cardiac strain and strain rate for assessment of fetal cardiac function: novel experience with velocity vector imaging. Echocardiography 2009; 26: 28–36. Reisner SA, Lysyansky P, Agmon Y, Mutlak D, Lessick J, Friedman Z. Global longitudinal strain: a novel index of left ventricular systolic function. J Am Soc Echocardiogr 2004; 17: 630–633. R Development Core Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing: Vienna, Austria, 2004; ISBN 3-900051-00-3. ¨ H, Katoh M, Kramann R, Franke A, Becker M, Bilke E, Kuhl ¨ Bucker A, Hanrath P, Hoffmann R. Analysis of myocardial deformation based on pixel tracking in two-dimensional echocardiographic images enables quantitative assessment of regional left ventricular function. Heart 2006; 92: 1102–1108. Sanchez-Quintana D, Anderson RH, Ho SY. Ventricular myoarchitecture in tetralogy of Fallot. Heart 1996; 76: 280–286. Peng QH, Zhou QC, Zeng S, Tian LQ, Zhang M, Tan Y, Pu DR. Evaluation of regional left ventricular longitudinal function in 151 normal fetuses using velocity vector imaging. Prenat Diagn 2009; 29: 1149–1155. Kutty S, Deatsman SL, Nugent ML, Russell D, Frommelt PC. Assessment of regional right ventricular velocities, strain, and displacement in normal children using velocity vector imaging. Echocardiography 2008; 25: 294–307. Taber LA, Sun H, Clark EB, Keller BB. Epicardial strains in embryonic chick ventricle at stages 16 through 24. Circ Res 1994; 75: 896–903. Tobita K, Keller BB. Right and left ventricular wall deformation patterns in normal and left heart hypoplasia chick embryos. Am J Physiol Heart Circ Physiol 2000; 279: H959–H969. Freeman GL, LeWinter MM, Engler RL, Covell JW. Relationship between myocardial fiber direction and segment shortening in the midwall of the canine left ventricle. Circ Res 1985; 56: 31–39. Van Mieghem T, Giusca S, DeKoninck P, Gucciardo L, Don´e E, Hindryckx A, D’Hooge J, Deprest J. Prospective assessment of fetal cardiac function with speckle tracking in healthy fetuses and recipient fetuses of twin-to-twin transfusion syndrome. J Am Soc Echocardiogr 2010; 23: 301–308.
SUPPORTING INFORMATION ON THE INTERNET The following supporting information may be found in the online version of this article: Appendix S1 Explanation of the difference between natural and Lagrangian strain.
Copyright 2011 ISUOG. Published by John Wiley & Sons, Ltd.
Ultrasound Obstet Gynecol 2011; 37: 150–157.
