High-Frequency 4-Dimensional Ultrasound (4DUS): A Reliable Method for Assessing Murine Cardiac Function

Abstract

In vivo imaging has provided a unique framework for studying pathological progression in various mouse models of cardiac disease. Although conventional short-axis motion-mode (SAX MM) ultrasound and cine magnetic resonance imaging (MRI) are two of the most prevalent strategies used for quantifying cardiac function, there are few notable limitations including imprecision, inaccuracy, and geometric assumptions with ultrasound, or large and costly systems with substantial infrastructure requirements with MRI. Here we present an automated 4-dimensional ultrasound (4DUS) technique that provides comparable information to cine MRI through spatiotemporally synced imaging of cardiac motion. Cardiac function metrics derived from SAX MM, cine MRI, and 4DUS data show close agreement between cine MRI and 4DUS but overestimations by SAX MM. The inclusion of a mouse model of cardiac hypertrophy further highlights the precision of 4DUS compared with that of SAX MM, with narrower groupings of cardiac metrics based on health status. Our findings suggest that murine 4DUS can be used as a reliable, accurate, and cost-effective technique for longitudinal studies of cardiac function and disease progression.

Keywords: cardiac disease; cine MRI; hypertrophy; mouse; ultrasound.

Figure 1.

Representative displays of the imaging modalities used on an example mouse: short-axis M-Mode (SAX MM) (A), 4D ultrasound (4DUS) (B), and bright-blood gradient echo magnetic resonance imaging (MRI) (C). The SAX MM row shows the prescribed cursor for sampling (dashed yellow line) along with corresponding data time-synced to ECG signals. Both 4DUS and MRI rows show long-axis (left), short-axis (center), and 4-chamber (right) views at corresponding slice locations (scale bar = 2.0 mm).

Figure 2.

Overview of masking and slice analysis. Example short-axis 4DUS slices with outlines of endocardial (blue) and epicardial (green) borders drawn (A). Adjacent short-axis border definitions are pieced together to create a volumetric mask, as shown in (B) with an example mask (blue) of the left ventricle (LV) in a long-axis view. The epicardial border (green) was also included in (B) for reference. To identify the sensitivity of total mask volume to interpolated borders across skipped slices, a complete masking (ie, every slice across the volume was manually outlined) had equal-sized subsets of slices deleted (C; top), which were subsequently filled using cubic spline interpolation (C; middle/bottom). A slice thickness of 0.0762 mm was used to calculate physical gap sizes. At each gap size, the set of remaining slices—excluding the most proximal and distal slices—were serially shifted to identify variability based on gap positioning. Total volumes were calculated for each paradigm as the sum of all cross-sectional areas. Percent differences in volume from the complete masking across all gap paradigms are shown for a representative wild-type (D; left), early-stage disease (D; middle), and late-stage disease mouse (D; right).

Figure 3.

Scatter dot plots with mean and standard deviation bars comparing measurement differences between each pair of imaging methods. In each plot, the left blue column represents the 4DUS–MRI values, the green center column represents the SAX MM–4DUS values, and the red right column represents the SAX MM–MRI values. The plots of ejection fraction (A), stroke volume (B), and LV mass (C) each represent metrics derived from volume estimates made at end-diastole (D) and peak-systole (E), commonly used to evaluate left ventricular function. One-way ANOVA identified statistical significance between methods as designated by *P < .05 and **P < .01, respectively.

Figure 4.

Representative Masson's Trichrome histology of the various disease stages imaged, with magnifications at 4× (scale bar = 1.0 mm) (A), 10× (scale bar = 200 μm) (B), and 40× (scale bar = 100 μm) (C). The first row shows a representative nonmutated mouse (n = 5), in which wild-type cardiomyocyte size and density are observed. The second row shows an early stage of hypertrophy (n = 3), in which enlarged cardiomyocytes are observed without any noticeable necrosis. The third row shows a late stage of hypertrophy (n = 2), in which enlarged cardiomyocytes and cell necrosis with less stain uptake are both observed.

Figure 5.

Characterization plots of the wild-type, early-stage disease, and late-stage disease mice derived from the 4DUS-based metrics. The top 3 plots of EF versus SV for SAX MM (A), MRI (B), and 4DUS (C) illustrate the relative spread of cardiac function metrics across the study animals. It is clear that the distributions are more closely matched for the MRI and 4DUS techniques. This agreement in metric distributions in also observed for the structural metrics of EDV versus LVM (D–F). The characteristics expected for the CPT2^M−/− model are highlighted in these plots showing relative clusters depending on health status.