Real-Time 3D Echocardiography in Pediatric Cardiology: Head-to-Head Comparison of 3D Quantification Software (QLab and TomTec) and CMRI

Background: Determining left ventricular (LV) volume is essential for therapy and follow-up in pediatric cardiology. Real-Time-3D-Echocardiography (RT3DE) is a promising method for LV assessment. There is evidence neither for quantification software to be interchangeable nor for one software to be more consistent with results obtained by CMRI.

Methods: A total of 370 healthy children (1 day to 216 months) underwent RT3DE imaging of the LV (IE33, Philips). 3D data was quantified using two different border detection software programs (Qlab 9.0, Philips and TomTec, LV 2.7), detecting end systolic (ESV), end diastolic (EDV), stroke volume (SV) and ejection fraction (EF). Using TomTec, influence of contour finding activity was tested by high (TomTec₇₅) as well as low (TomTec₃₀) sensitivity. Results of 22 subjects were compared to CMRI derived data.

Results: Analysis of identical 3D-data with different software had a significant impact. Comparing QLab to TomTec₃₀, Bland-Altman analysis showed small biases ([Table 1]). Comparing QLab to TomTec₇₅, using a higher contour sensitivity, the bias was significant higher (ESV: 9.8%). 95% limits of agreement (LOA) in both cases showed a wide range. To examine the impact of image quality, only images rated excellent were compared (104 patients). In contrast to data of all probands, 95% LOA were smaller and deviations over 50% did not occur (ESV: bias 0.1%; LOA: 39.6–39.7%). Concordant to previous studies QLab and TomTec underestimated volumes derived by CMRI. Intra- and interobserver-variability were excellent in both Qlab and TomTec.

Table 1 Bland Altman analysis

Parameter	Bias (mL)	95% LOA (mL)	Bias (%)	95% LOA (%)
Agreement between QLab and TomTec30
ESV	0.7	−12 to 13.3	0.8	−45 to 47
EDV	−1.2	−26.3 to 23.8	−2.2	−35.5 to 31.2
Agreement between CMRI and RT3DE-QLab
ESV	2.3	−8.9 to 13.6	4.8	−24 to 33.6
EDV	8.3	−14.5 to 31.1	10.6	−15.4 to 36.6
Agreement between CMRI and RT3DE- TomTec30
ESV	−1.9	−21 to 17	−2.2	−38.1 to 33.6
EDV	8.9	−19 to 37	12.2	−16.2 to 40.5

Conclusion: RT3DE by semi-automated border detection software allows noninvasive assessment of LV volumes in pediatric cardiology. However, results differ, depending on software algorithms of the individual software as well as on changes of contour sensitivity within the software itself. Therefore, quantification programs should not be used interchangeably. Suboptimal image quality impairs border-tracing acutance.