Francisca Nijland

Automatic segmentation of echocardiographic sequences by active appearance motion models

A novel extension of active appearance models (AAMs) for automated border detection in echocardiographic image sequences is reported. The active appearance motion model (AAMM) technique allows fully automated robust and time-continuous delineation of left ventricular (LV) endocardial contours over the full heart cycle with good results. Nonlinear intensity normalization was developed and employed to accommodate ultrasound-specific intensity distributions. The method was trained and tested on 16-frame phase-normalized transthoracic four-chamber sequences of 129 unselected infarct patients, split randomly into a training set (n=65) and a test set (n=64). Borders were compared to expert drawn endocardial contours. On the test set, fully automated AAMM performed well in 97% of the cases (average distance between manual and automatic landmark points was 3.3 mm, comparable to human interobserver variabilities). The ultrasound-specific intensity normalization proved to be of great value for good results in echocardiograms. The AAMM was significantly more accurate than an equivalent set of two-dimensional AAMs.

Prognostic Implications of Restrictive Left Ventricular Filling in Acute Myocardial Infarction: A Serial Doppler Echocardiographic Study

OBJECTIVES This study was designed to evaluate the relative prognostic significance of restrictive left ventricular (LV) filling after acute myocardial infarction. BACKGROUND Data regarding the contribution of diastolic dysfunction to prognosis after myocardial infarction are limited, and the additional value over the assessment of systolic dysfunction is not known. METHODS Serial Doppler echocardiography was performed in 95 patients on days 1, 3 and 7 and 3 months after acute myocardial infarction. Patients were classified into two groups: a restrictive group (n = 12) with a peak velocity of early diastolic filling wave (E)/peak velocity of late filling wave (A) ratio > or = 2 or between 1 and 2 and a deceleration time (DT) < or = 140 ms during at least one echocardiographic study; and a nonrestrictive group (n = 83) with an E/A ratio < or = 1 or between 1 and 2 and a DT > 140 ms at all examinations. RESULTS Cardiac death occurred in 10 patients during a mean follow-up interval of 32 +/- 17 months. The survival rate at 1 year was 100% in the nonrestrictive group and only 50% in the restrictive group. After 1 year there was a continuing divergence of mortality, resulting in a 3-year survival rate of 100% and 22%, respectively. Univariate Cox analysis revealed that restrictive LV filling, wall motion score index, ejection fraction and end-systolic and end-diastolic volume indexes, as well as peak creatine kinase, peak MB fraction and heart failure during the hospital course were significant predictors of cardiac death, although restrictive filling was the single best predictor (p < 0.0001). Multivariate analysis showed that restrictive filling adds prognostic information to clinical and echocardiographic variables of systolic dysfunction. CONCLUSIONS Restrictive LV filling after acute myocardial infarction is the single best predictor of cardiac death and adds significantly to clinical and echocardiographic markers of systolic dysfunction.

Prediction of improvement of contractile function in patients with ischemic ventricular dysfunction after revascularization by fluorine-18 fluorodeoxyglucose single-photon emission computed tomography

OBJECTIVES We evaluated the use of fluorine-18 fluorodeoxyglucose (FDG) and single-photon emission computed tomography (SPECT) to predict improvement of left ventricular ejection fraction (LVEF) after revascularization. BACKGROUND FDG SPECT has recently been proposed for assessment of myocardial viability. However, FDG SPECT still awaits validation in terms of predicting improvement of contractile function after revascularization in patients with poor left ventricular (LV) function. METHODS Fifty-five patients with contractile dysfunction (including 22 with LVEF < 30%) underwent FDG SPECT during hyperinsulinemic glucose clamping and early thallium-201 SPECT (to assess perfusion). Improvement of LV function was evaluated 3 months after revascularization with echocardiography and radionuclide ventriculography. RESULTS The 55 patients were arbitrarily classified into two groups: 19 with three or more viable, dysfunctional segments on FDG SPECT and 36 with less than three viable, dysfunctional segments. LVEF increased significantly in the first group, from 28 +/- 8% (mean +/- SD) before to 35 +/- 9% (p < 0.01) after revascularization. In the second group, LVEF remained unchanged after revascularization (45 +/- 14% vs. 44 +/- 14%, p = NS). The 22 patients with severely depressed LV function were similarly classified into two groups: 14 with three or more viable segments on FDG SPECT in whom LVEF improved significantly (25 +/- 6% vs. 32 +/- 6%) and 8 with less than three viable segments in whom LVEF remained unchanged (24 +/- 6% vs. 25 +/- 6%). CONCLUSIONS This study shows that FDG SPECT can identify patients in whom LV function improves after revascularization. Because SPECT is widely available, this technique may contribute to more routine use of FDG for determination of viability.

Long-term Implications of Reocclusion on Left Ventricular Size and Function After Successful Thrombolysis for First Anterior Myocardial Infarction

BACKGROUND Successful thrombolysis can prevent left ventricular dilatation after acute myocardial infarction. However, in almost 30% of patients, reocclusion occurs. The aim of this study was to assess the long-term implications of reocclusion on left ventricular size and function. METHODS AND RESULTS Fifty-six patients were studied with two-dimensional echocardiography at baseline (2 +/- 1.6 days) and 5.0 +/- 1.4 years after first anterior myocardial infarction. All patients (a subset of those enrolled in the APRICOT trial) had a patent infarct-related artery when studied < 48 hours after thrombolysis and underwent repeat coronary angiography at 3 months. Baseline characteristics were comparable in patients with (n = 17) and without reocclusion (n = 39). Left ventricular volume indexes were stable in patients without reocclusion. Patients with reocclusion, however, showed a significant increase in end-diastolic volume index (EDVI; P = .008) and end-systolic volume index (ESVI; P = .039). Furthermore, patients without reocclusion demonstrated improvement in wall motion score index (WMSI; P = .0001) and ejection fraction (EF; P = .016), whereas patients with reocclusion did not. After 5 years, patients with reocclusion had significantly larger volume indexes (EDVI, 99 +/- 41 versus 76 +/- 22 mL/m2, P = .007; ESVI, 59 +/- 40 versus 39 +/- 20 mL/m2, P = .017) and more compromised left ventricular function (WMSI, 1.63 +/- 0.33 versus 1.39 +/- 0.32, P = .013; EF, 45 +/- 13% versus 51 +/- 11%, P = .077) than patients without reocclusion. Multivariate analysis identified baseline WMSI and reocclusion as significant independent predictors of left ventricular dilatation. CONCLUSIONS Reocclusion of the infarct-related artery within 3 months of successful thrombolysis is associated with left ventricular dilatation and is detrimental to functional recovery of left ventricular function 5 years after first anterior myocardial infarction.

Myocardial viability: impact on left ventricular dilatation after acute myocardial infarction

Objective: To evaluate whether the presence of viable myocardium, detected by low dose dobutamine echocardiography, limits the likelihood of left ventricular dilatation in patients with acute myocardial infarction. Patients: 107 patients were studied by low dose dobutamine echocardiography at (mean (SD)) 3 (1) days after acute myocardial infarction. Cross sectional echocardiography was repeated three months later. Patients were divided in two groups based on the presence (n = 47) or absence (n = 60) of myocardial viability. Results: Baseline characteristics were comparable between the two groups, except for infarct location. Left ventricular end diastolic volume index (EDVI) was stable in patients with viability, but end systolic volume index (ESVI) decreased significantly (p = 0.006). Patients without viability had a significant increase in both EDVI (p < 0.0001) and ESVI (p = 0.0007). Subgroup analysis in patients with small and large infarcts (peak creatine kinase ≤ 1000 v > 1000 IU/l) showed that ventricular dilatation occurred only in patients with large infarcts without viability. This resulted in larger ESVI values at three months in that group compared with patients with large infarcts plus viability (p < 0.05). Multivariate regression analysis identified myocardial viability as an independent predictor of left ventricular dilatation, along with wall motion score index on low dose dobutamine echocardiography and the number of pathological Q waves. Conclusions: The presence of viability early after acute myocardial infarction is associated with preservation of left ventricular size, whereas the absence of viability results in ventricular dilatation, particularly in large infarcts.

In-hospital and long-term prognostic value of viable myocardium detected by dobutamine echocardiography early after acute myocardial infarction and its relation to indicators of left ventricular systolic dysfunction

The prognostic value of myocardial viability early after acute myocardial infarction (AMI) is still controversial, depending on the patient under study and the outcome end point considered. Furthermore, the relative prognostic importance of viability compared with indicators of systolic left ventricular (LV) dysfunction is not known. One hundred thirty-eight patients were studied with low-dose dobutamine echocardiography 3 +/- 1 days after AMI. Patients were divided in 2 groups based on presence (n = 55) or absence (n = 83) of myocardial viability and followed up for in-hospital and late cardiac events. During hospitalization, myocardial viability was the only independent predictor for recurrent ischemic events (chi-square 5.0, p = 0.025). End-systolic volume index and ejection fraction were both independent predictors of the occurrence of heart failure, whereas gender and end-systolic volume index emerged as independent predictors of hard cardiac events (death and sustained ventricular tachycardia). After hospital discharge, patients were followed for 19 +/- 7 months. Again, myocardial viability emerged as the only independent predictor of unstable angina (chi-square 7.7, p = 0.005). Age, hypertension, and ejection fraction were the most important independent predictors of hospitalization for heart failure, whereas ejection fraction was the only independent predictor of hard cardiac events. Presence of myocardial viability early after AMI is the single best predictor of recurrent in-hospital ischemic events and unstable angina after discharge. With respect to hard cardiac events and occurrence of heart failure, indicators of LV systolic dysfunction have a higher prognostic value than presence of myocardial viability.

Active appearance motion models for endocardial contour detection in time sequences of echocardiograms

Active Appearance Models (AAM) are suitable for segmenting 2D images, but for image sequences time-continuous results are desired. Active Appearance-Motion Models (AAMM) model shape and appearance of the heart over the full cardiac cycle. Single-beat sequences are phase-normalized into stacks of 16 2D images. In a training set, corresponding shape points on the endocard are defined for each image based on expert drawn contours. Shape (2D) and intensity vectors are derived similar to AAM. Intensities are normalized non-linearly to handle ultrasound-specific problems. For all time frames, shape vectors are simply concatenated, as well as and intensity vectors. Principal Component Analysis extracts appearance eigenvariations over the cycle, capturing typical motion patterns. AAMMs perform segmentation on complete sequences by minimizing model-to-target differences, adjusting AAMM eigenvariation coefficients using gradient descent minimization. This results in time-continuous segmentation. The method was trained and tested on echocardiographic 4-chamber sequences of 129 unselected patients split randomly into a training set (n=65) and a test set (n=64). In all sequences, an independent expert manually drew endocardial contours. On the test set, fully automated AAMM performed well in 97% of cases (average distance 3.3 mm, 9.3 pixels, comparable to human inter- and intraobserver variabilities).

Overview of automated quantitation techniques in 2D echocardiography

Many methods for automated quantitation in two-dimensional echocardiography have been published, but few have gained practical importance. This chapter describes the problems and pitfalls of border detection in cardiac ultrasound, gives an overview of methods described in the literature and categorizes the applied techniques into a hierarchy of abstraction levels. Furthermore, a practical system for automated border detection (ECHO-CMS) and its evaluation will be discussed, and the chapter will be concluded with an overview of the general developments anticipated for the near future.

Early prediction of improvement in ejection fraction after acute myocardial infarction using low dose dobutamine echocardiography

Objective: To evaluate the relation between changes in ejection fraction during the first three months after acute myocardial infarction and myocardial viability. Patients: Myocardial viability was assessed using low dose dobutamine echocardiography in 107 patients at mean (SD) 3 (1) days after acute myocardial infarction. Cross sectional echocardiography was repeated three months later. Left ventricular volumes and ejection fraction were determined from apical views using the Simpson biplane formula. Results: In patients with viability, ejection fraction increased by 4.4 (4.3)%; in patients without viability it remained unchanged (0.04 (3.6)%; p < 0.001). A ≥ 5% increase in ejection fraction was present in 21 of 107 patients (20%). Receiver operating characteristic analysis showed that myocardial viability in ≥ 2 segments predicted this increase in ejection fraction with a sensitivity of 81% and a specificity of 65%. Multivariate logistic regression analysis was used to define which clinical and echocardiographic variables were related to ≥ 5% improvement in ejection fraction. Myocardial viability, non-Q wave infarction, and anterior infarction all emerged as independent predictors, myocardial viability being the best (χ2 = 14.5; p = 0.0001). Using the regression equation, the probability of ≥ 5% improvement in ejection fraction for patients with a non-Q wave anterior infarct with viability was 73%, and for patients with a Q wave inferior infarct without viability, only 2%. Conclusions: Myocardial viability after acute myocardial infarction is the single best predictor of improvement in ejection fraction. In combination with infarct location and Q wave presence, the probability of ≥ 5% improvement can be estimated in individual patients at the bedside.

Fully automated endocardial contour detection in time sequences of echocardiograms by three-dimensional active appearance models

A novel 3-D Active Appearance Model (3-D AAM) is applied to fully automated endocardial contour detection in 2-D + time (2DT) 4-chamber ultrasound sequences, without knowledge of cardiac phase (ED/ES frames). 2DT appearance of the heart is modeled in 3-D by converting the stack of 2-D time slices into a 3-D voxel space. In a training set, an expert defines corresponding endocardial contour points for one complete cardiac cycle (ED to ED). 2DT shape is represented as a 3-D surface. Image appearance is modeled as a vector of voxel intensities in a volume-patch spanned by the 3-D surface. Principal Component Analysis extracts eigenvariations of 3-D shape and appearance, capturing typical cardiac motion patterns. 3-D AAM segments the image volume by minimizing 3-D model-to-target intensity differences, adjusting eigenvariation coefficients and 3-D pose using gradient descent minimization. This provides time-continuous border localization for one beat in both time and space. The method was used on 3-beat sequences from 129 patients split randomly into a training (65) and a test set (64). An independent expert manually drew all endocardial contours. 3-D AAM converged well in 89% of test cases. Average absolute temporal error was 37.0 msec, spatial error 3.35 mm, comparable to human inter-observer variabilities.