Fig 1.
Diagram showing the steps in our cardiac imaging pipeline.
We have bold-faced the items that relate to new aspects unique to this work (compared to our previous study [12]).
Table 1.
Summary of cohort statistics grouped by genotype, sex, and exercise participation.
Table 2.
Overview of experimental group composition with percentages of the HN and exercise factors in each genotype. This table has been included primarily to communicate the relative counts for each experimental factor (i.e., sex, HN factor, and exercise).
Table 3.
Test set quantitative segmentation performance summary. The upper half of the table shows the performance of the network when trained using CT images as input while the bottom half shows performance when trained with I maps as input. We show mean values of accuracy, Jaccard index, Dice coefficient, precision, and recall for each anatomical segment. We also show an overall mean value for each performance metric across all segments.
Fig 2.
Test set qualitative segmentation performance overview.
Within the images labelled “Label” and “Prediction,” the outline of each segmentation is shown as an overlay on either a CT image (upper half of figure) or a decomposed I image (bottom half of figure). The CT images are shown as attenuation maps with units of 1/voxel size while the I images represent concentrations with units of mg/mL. Each segmentation color represents a different anatomical region. We reiterate that the segmentations are voxelated 3D semantic segmentations; we show only the 2D outline of the segmentations in these images for visual simplicity.
Table 4.
Averages and standard deviations for quantitative metrics grouped by sex, genotype, and exercise state. Standard deviations are shown in parentheses below each average value. From left to right the quantitative metrics are mass, heart rate (HR), stroke volume (SV), ejection fraction (EF), cardiac output (CO), cardiac index (CI), right ventricular stroke volume (RVSV), right ventricular ejection fraction (RVEF), and myocardial mass (MM).
Fig 3.
Violin plots of mass across genotype, sex, and exercise groups.
The lines with asterisks indicate significant differences (p<0.05) identified by the Kruskal-Wallis test and (when needed) Dunn’s post hoc test. These plots coupled with other findings indicate that mass differences in our cohort can be attributed to sex and genotype but not to exercise.
Table 5.
A summary of all significant findings from the multi-factor ANOVA. For brevity, we have included only the metrics and effects that were statistically significant (p<0.05). In the final column, we have provided a basic interpretation of those results. Effect sizes are described relatively with the values 0.01, 0.06, and 0.14 representing small, medium, and large respectively.
Table 6.
A summary of the significant differences (p<0.05) identified by the stratified analysis of exercise within sex and genotype subgroups.
Fig 4.
Collection of violin plots showing the differences between exercised and nonexercised mice subdivided by sex.
The bars with asterisks indicate statistically significant differences revealed by a Mann Whitney U test (P<0.05).
Fig 5.
Collection of violin plots showing the differences between exercised and nonexercised mice subdivided by.