Eligible patients
This retrospective study included patients who were referred to stress echocardiography from January 2013 to February 2014 at the Karolinska University Hospital in Huddinge (The heart clinic), Sweden. This patient group was selected as the image protocol at our institution routinely involves analysis of both greyscale and contrast-enhanced images both at rest and during stress, regardless of image quality. This enables a comparison of the diagnostic outcome between greyscale and contrast-enhanced images at rest in the subset of patients beyond the current recommendations for contrast agent usage in standard echocardiography. Note that the greyscale images routinely are acquired with such quality that deformation analysis using speckle tracking is feasible during rest and stress. The image acquisition was performed with a GE Vivid E9 (GE Healthcare, Wisconsin, USA), at a transmitted frequency of 1.5–1.7 MHz. All contrast-enhanced images were acquired at a low MI index using contrast specific image sequences, such as pulse inversion. SonoVue® (Bracco Diagnostics Inc., Switzerland) (n = 187) and Optison® (GE Healthcare, Wisconsin, USA) (n = 5) were used as contrast agents.
The exclusion criteria were based on the recommendations for contrast agent usage during standard echocardiography [7, 8]. Patients were excluded if they fulfilled the criteria for contrast-enhanced echocardiography, i.e. two or more contiguous LV segments not visualized on a greyscale image or suspicion of LV structural abnormalities, such as apical hypertrophic cardiomyopathy, ventricular non-compaction, thrombus or ventricular pseudoaneurysm. Experienced echocardiographer (n = 4, 3–15 years of experience) of contrast-enhanced echocardiography visually evaluated whether patients should be included in the study or not. Additionally, patients were excluded if any apical 2 chamber (ch), 3ch or 4ch images in either greyscale or contrast mode were missing. The study protocol was approved by the regional ethical review board in Stockholm, Sweden.
Image analysis
The flow chart of the image analysis is shown in Fig. 1. For all patients included, two image sequences (greyscale images, greyscale + contrast images) were prepared for each patient. The latter image sequence contained both modes to support the evaluation of LV structural abnormalities, but only the contrast images in this sequence were used during WMSI and EF assessment. Before image analysis was initiated, all patient data were anonymised. The image analyses were allocated between four experienced readers and the readers evaluated the two image sequences of one patient with a minimum of two days apart (greyscale images, greyscale + contrast images). Furthermore, repeated analyses were performed on five patients for each of the experienced readers, who were blinded to previous results. To assess the interobserver variability, these five patients were also analyzed once by another experienced reader. Moreover, the variability between readers with different experience level was evaluated by letting two inexperienced readers do repeated analyses on the same twenty patients as the ones being re-evaluated by the experienced readers. The inexperienced reviewers had less than 0.5 years of experience of analyzing echocardiographic images. Prior to initiation, the inexperienced readers received a demonstration of the image analysis process.
Assessment of wall motion
From apical 2ch, 3ch and 4ch views, WMSI was obtained by dividing the LV using an 18 segment model [10]. Each segment was assigned a wall motion score (WMS) depending on the myocardial thickening and movement pattern. A normally contracting segment was graded as 1, hypokinesia as 2, akinesia as 3 and dyskinesia as 4. A segment was given NA if evaluation of that specific segment was not applicable due to insufficient image quality.
LV volume measurements
Left ventricular end-diastolic volume (EDV), left ventricular end-systolic volume (ESV) and EF were measured in one cardiac cycle using the biplane method of disks summation (the modified Simpson’s rule) in EchoPAC (GE Healthcare, Wauwatosa, WI, USA). End-diastole was visually defined as the maximal cavity area following mitral valve closing, whereas end-systole was defined as the minimal cavity area preceding mitral valve opening. In accordance with the recommendations from EACVI and ASE, papillary muscle and trabeculations were excluded from the cavity during tracing [10].
Detection of LV structural abnormalities
The contrast-enhanced images (3 images/patient) were screened for LV structural abnormalities such as apical hypertrophic cardiomyopathy, ventricular non-compaction, thrombus and ventricular pseudoaneurysm. The screening for LV structural abnormalities was only conducted by the experienced readers. Note that patients, who showed structural abnormalities on the greyscale images, were already excluded from the study since they had indication for contrast-enhanced echocardiography.
Statistics
Data analysis was performed using the statistical software IBM SPSS Statistics 19 (Armonk, New York, USA). Age, blood pressure, LV volumes and EF were classified as continuous data and expressed with mean and standard deviation. Paired Student t test (95 % confidence level) was used to compare numerical groups. WMSI for each patient was determined by calculating the sum of all WMS divided by the number of visualized segments [10]. Paired Student t test (95 % confidence level) was used to compare WMSI before and after contrast. Patients were categorized according to reference values of EF: ≥ 55 % reference range, 45–54 % mildly abnormal, 30–44 % moderately abnormal, < 30 % severely abnormal [10]. Additionally, the difference in EF classification between greyscale and contrast image analysis was investigated using McNemar test. Intra– and interobserver variability for WMSI and EF measurements was assessed by using intra-class correlation coefficient (ICC) (95 % confidence level).