AIM: To examine the feasibility and reliability of measuring global and segmental longitudinal strain (LS) compared to visual assessment of wall motion (WM).

METHODS: Assessment of segmental (17 left ventricular segments) LS using automatic function imaging (AFI) in 55 patients (60.0 ± 8.7 years, 73% male) divided into 2 groups: group I included 35 patients with WM abnormalities and/or impaired ejection fraction and group II included 20 patients with normal WM and ejection fraction. Visual analysis of WM abnormalities was performed using 2-dimensional echocardiography (2DE) and WM score was calculated. Both modalities were analyzed by one expert reader at 2 different sessions.

RESULTS: Analysis of 935 left ventricular segments was completed in 94.1% and 96.3% by visual assessment and AFI, respectively. There was a strong positive linear relationship between the WM score and global LS in all patients. Intra-observer agreement for calculation of WM score was excellent for group I patients (kappa: 0.97) and very good for group II patients (kappa: 0.92). Intra-observer agreement for AFI showed excellent agreement with very small mean difference in both group I and II (-0.0 ± 2.3 and -0.0 ± 1.9, respectively).

CONCLUSION: The interpretation of global and segmental LS using AFI is a more feasible and reliable technique for the quantification of myocardial deformation than visual assessment of WM scores.

Echocardiographic evaluation of segmental and global myocardial function by visual assessment is the most used method[1]. The visual assessment relies mainly on myocardial radial performance, and is limited by high inter- and intra-observer variability. Furthermore, it provides a subjective evaluation of endocardial thickening and excursion[2]. Tissue Doppler imaging was expected to overcome these limitations, but its sensitivity to measure the longitudinal and radial deformation was limited due to the angle-dependancy with inability to differentiate between active and passive myocardial motion[3,4].

Speckle-tracking echocardiography (STE) is a new noninvasive imaging technique that quantitatively analyzes global and regional myocardial function. Its evaluation is based on tracking natural acoustic reflections and interference patterns within an ultrasound window[5]. A large amount of published data has described the accuracy and clinical applications of STE with the ability to elaborate the myocardial deformation in longitudinal, radial and circumferential directions[6-8].

Automated function imaging (AFI) is a novel algorithm calculating the myocardial deformation from ultrasound speckles that is used to measure myocardial strain. It tracks the percentage of wall lengthening and shortening which helps to assess myocardial deformation based on grayscale images, similar in concept to magnetic resonance imaging tagging. It presents an objective, semi-automatic, and angle-independent analysis of longitudinal peak systolic strain (LS) based on speckle tracking and provides a single bull’s-eye summary of the left ventricular (LV) segmental wall strain[9].

The objective of this study was to assess the feasibility and reliability of AFI compared with visual assessment by 2-dimensional echocardiography (2DE), regardless of underlying cardiac disorders.

The study included 55 consecutive patients who were referred for echocardiography for assessment of LV function and segmental wall motion (WM) abnormalities. The inclusion criteria were adequate image quality and sinus rhythm. Conventional 2DE was performed according to guidelines using a commercial ultrasound system (Vivid 7, GE Health Medical, Milwaukee, WI, United States) supported by a multi-frequency transducer (M3S 1.7/3.4 MHz). After the completion of the 2DE study, the 17 LV segments were assessed by visual analysis and AFI. Both visual analysis and AFI were performed by one experienced observer at two different settings. The first analysis was performed immediately after the end of each study. The second re-analysis of all studies was performed by the same observer after collection of cases. The interval between the 2 analysis was 1 mo. In each study, the observer was blinded to the previous readings.

Visual analysis: Visual analysis of the contractile function of all the 17 segments was interpreted according to the American Society of Echocardiography criteria[10] using a four-point score: (1) normal; (2) hypokinetic; (3) akinetic; or (4) dyskinetic. The sum of the WM scores, averaged over the number of segments with interpretive scores, gave the WM score index.

AFI protocol: The AFI was performed through an off-line analysis of 3 digitally stored 2D images (apical long-axis, 2- and 4-chamber) with high frame rates (> 60 frames/s) using commercial imaging analysis software (EchoPAC 6.1.0, GE Health Medical)[11]. The peak systolic strain values in a 17-segment LV model were used in the present study. End-systole was defined as aortic valve closure in the apical long-axis view by continuous Doppler wave recording. Automated delineation of endocardial borders was obtained through marking the mitral annulus level and at the apex on each digital loop. The area of interest was manually adjusted if automated delineation was not optimal. Segments with poor image acquisition or artifacts were excluded due to inability to measure LS. Segmental LS was calculated as the percentage of lengthening or shortening and the results for each plane were displayed (Figure 1). The results for all 3 planes were then combined in a single bull’s-eye summary (Figure 2). The sum of LS, averaged over the number of segments with interpretive scores, gave the global LS. The normal values of segmental and global LS were considered (-17.4% for females and -15.9% for males) based on the Hunt study in Norway[12] which presented the reference values in 1266 healthy individuals according to age and sex.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Segmental longitudinal strain displayed in 3 image planes. A: Apical 4-chamber; B: Apical 2-chamber; C: Apical long axis views.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Measurements of longitudinal strain in all 17 left ventricle segments represented in bull’s eye map in normal (right) and abnormal (left) subjects.

This study demonstrated that AFI analysis of global and segmental LS was well correlated with visual assessment of WM abnormalities. Compared to visual assessment, AFI showed very small intra-observer variability regardless of the presence or absence of WM abnormalities and impaired LV systolic function.

The visual interpretation of WM abnormalities with conventional 2DE is based on the assessment of myocardial thickening and endocardial excursion. This method is widely used for assessing LV global and segmental systolic function. However, it is observer-dependant and requires experience[1]. Several reports have described the reliability and feasibility of STE to evaluate global and segmental myocardial deformation throughout the cardiac cycle in both normal and abnormal subjects. It is able to investigate LV function (circumferential, radial and longitudinal) without angle dependency[15-17].

The AFI algorithm is a novel method based on 2D strain imaging which tracks acoustic pixels equally distributed within the myocardial wall. It enables the quantification of myocardial strain simultaneously in different LV segments with ultrasound beam angle-independency. Previous studies reported that AFI is a fast, simple and high reproducible tool for semi-automatic assessment of LV function that supports clinical decision-making[9,18]. The current study aimed to evaluate the reliability and feasibility of AFI to detect segmental WM asynergy compared with visual assessment.

In agreement with the studies of Reisner et al[18] and Munk et al[19], global LS was correlated better with WM score index than with LV ejection fraction (R = 0.86, P < 0.0001; R = 0.62, P < 0.01, respectively).This finding can be explained by the direct effect of LV volume and heart rate on calculation of ejection fraction, while this effect is minimal on WM score index and nil on global LS.

By AFI, complete analysis was obtained for 96.3% of segments while by visual analysis, it was less (94.1%). This was explained by the higher percentage of un-interpretable segments by visual analysis due to poor endocardial delineation, while AFI (which depends on objective tracking of pixels) originated predominantly from the sub-endocardial fibers[20]. Among the missed segments, basal antero-lateral and basal anterior had the higher incidence by both AFI and visual analysis, with tendency of better intra-observer correlation for AFI than visual analysis.

Theoretically speaking, the semi-automated calculation improves reproducibility of the technique when compared with objective assessment. This can explain the good intra-observer variability in measurement of global LS and LV rotation by 2D speckle tracking which was demonstrated in previous studies[18,21]. In the current study, excellent intra-observer correlation and agreement was shown by both techniques in normal subjects. In abnormal subjects, AFI showed better intra-observer correlation and agreement than visual assessment.

Reclassification of our patients based on global LS measurements showed that a considerable percentage (25%) of patients who were considered as normal due to absent segmental WM abnormalities had reduced global LS. This was in agreement with previous studies[19,22] which demonstrated that quantification of segmental and global LS may detect subclinical LV dysfunction earlier than visual analysis of WM in asymptomatic patients with apparently normal LV ejection fraction.

The number of patients included is relatively small; however, it was enough to reach the target of the study. Only one software application (AFI) was applied for measurement of LS; therefore, our results must be taken in consideration towards the AFI only and not for other available software.

The AFI algorithm enabled a comprehensive assessment of global and segmental LS of the LV. It overcame the subjective and semi-quantitative analysis of LV obtained by visual assessment and showed better intra-observer agreement in both normal and abnormal subjects.