Novel indices for left-ventricular dyssynchrony characterization based on highly automated segmentation from real-time 3-d echocardiography

Abstract

Cardiac resynchronization therapy (CRT) using a biventricular pacemaker is an invasive and expensive treatment option for left ventricular mechanical dyssynchrony (LVMD). The CRT candidate selection is a crucial issue due to the unreliability of the current standard CRT indicators. Real-time three-dimensional (3-D) echocardiography (RT3DE) provides four-dimensional (4-D) (3-D+time) information about the LV and is suitable for LVMD assessment. In this article, the complex left ventricle (LV) shape and motion of 50 RT3DE datasets are represented by novel 4-D descriptors - 4-D sphericity, volume and shape, from which novel indices were derived by principal component analysis (PCA) and subsequently analyzed by a support vector machine (SVM) classifier to assess their capability of LVMD characterization and CRT outcome prediction. These novel indices outperformed clinical indices and have promising capabilities in disease characterization and great potential in CRT outcome prediction. To enable efficient quantitative RT3DE analysis, a segmentation method was developed to combine the powers of active shape models and optimal graph search. Various aspects of the method were designed to handle varying RT3DE image quality among datasets and LV segments. An application with graphical user interface was developed to provide the user with simple and intuitive control. The developed method was robust to inter-observer variability and produced very good accuracy - 3.2±1.1 mm absolute surface positioning error, <1 mm mean signed error and <5% mean volume difference. The computer method's classification performance was compared with the independent standard, showing that the 4-D shape modal indices were not only the most capable of all tested options when employed for disease characterization but also the least sensitive to segmentation imperfections.

Figure 1

Sixteen left ventricular segments.

Figure 2

Meshes of (a) full LV, (b) apical (#13–16) segments, and (c) basal (#1–6) segments.

Figure 3

Graph construction.

Figure 4

The profile vector p associated with graph node υ.

Figure 5

Results of PCA on profile vectors of LV segment 3: mean profile (solid lines) and variations associated with four strongest components (dashed lines correspond to $\mathit{p̅}\pm 2\sqrt{{\lambda }_i}{\mathit{\varphi }}_i$).

Figure 6

Manual tracing of reoriented RT3DE dataset.

Figure 7

Screenshot of the RT3DE segmentation application. Yellow dashed lines show the user-specified region of interest. Three segmentation results using different sets of cost weights and their corresponding 4D volumes are shown (red: w_g=1, green: w_i=1, blue: w_p=1).

Figure 8

Variations of 4D volumes of LV segments 3 (a1–a3) and 8 (b1–b3) associated with varying first three PCA scores (top to bottom) by ±2σ (dashed lines) and the mean volumes (solid lines) of models derived from the independent standard of 50 LVMD datasets.

Figure 9

Shape variations at phases 0 (top row) and 7 (bottom row) introduced by varying the first PCA score of the 4D LV shape model derived from the independent standard of 50 LVMD datasets. The values of the first PCA score are: −2σ (left), 0 (m ean shape, middle), and 2σ (right).

Figure 10

Examples of segmented RT3DE data. Six datasets with increasing image quality measured by CNR for the whole LV are shown, where dataset 1 has the lowest CNR among all 50 datasets tested.

Figure 11

(a1–2) Distributions (μ ± σ, blue +s’ and lines) of image quality (CNR of LV segment) for individual RT3DE dataset (a1) and individual segment (a2). Red line in (a1) shows the sorted CNRs of the whole LV. (b1–2) Distributions of absolute positioning errors (measured for each LV segment) for individual RT3DE dataset (b1) and individual LV segment (b2).

Figure 12

Distributions (μ ± σ) of temporal segmentation errors. (a1–3) absolute positioning errors; (b1–3) percent volume differences; (c1–2) base and apex offsets.

Figure 13

Comparison of volume curves derived from the manual tracing and computer segmentations of the RT3DE datasets shown in Figure 10.