Alexander MacArthur MD, Presenter: Nothing to Disclose

Jean Jeudy MD, Abstract Co-Author: Nothing to Disclose

Charles S. White MD, Abstract Co-Author: Nothing to Disclose

Left ventricular ejection fraction (EF) is commonly calculated from computed tomography (CT) images using Simpson’s method from the short-axis (SAX) stack of images. However, with this technique, error may arise when evaluating the base of the left ventricle, due to slice irregularity from the mitral valve and left ventricular outflow tract. Using three dimensional models from CT studies as a reference standard, we compared the EF measured using Simpson’s method with SAX images to EF measured using the Simpson’s method with horizontal long-axis (HLA) images.

With IRB approval, we analyzed 28 retrospectively-gated cardiac CT scans acquired between July 2010 and March 2011. Each study contained 10 cardiac phases from which end-diastolic (ED) and end-systolic (ES) phases were chosen. Contiguous image stacks were reconstructed in SAX and HLA orientations. EF was then calculated in each series using Simpson’s method. Each calculated EF was compared to the reference EF based on three-dimensional models of the left ventricle at ED and ES.

The mean difference between the SAX calculated EF and the volumetric EF was 3.48%, standard deviation (SD) = 8.96%, while the mean difference between HLA calculated EF and the volumetric EF was 0.30%, (SD= 4.83%; p < 0.05). After Bland-Altman analysis, there was no significant correlation between ejection fraction and Simpson’s method estimation error. SAX EF and HLA EF were linearly correlated with volume EF with R2 values of 0.80 and 0.92, respectively. Separate regressions show that both SAX EF and HLA EF are significant predictors of volumetric EF. SAX EF predicts 79% of the variance in volumetric EF. HLA predicts 91% of the variance. However, once HLA has been entered as a predictor, SAX does not add significantly to the accuracy of the prediction. The combined regression still predicts 91% of the variance, and the p-value for SA is non-significant.

These results support the horizontal long-axis Simpson’s method as a convenient and perhaps more accurate alternative to the traditional short-axis Simpson’s method when calculating left ventricular ejection fraction from CT images.

Although Simpson’s method using SAX slices is conventionally used to calculate EF, this study demonstrates that similar Simpson’s method using HLA slices provides more significantly reliable results.

MacArthur, A, Jeudy, J, White, C, Comparison of Ejection Fraction Calculation Using Simpson’s Method in Long Axis and Short Axis CT Images.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11016461.html