Evaluation strategies for determination of left ventricular indices: Pros and Cons of model vs. non-model based quantification software

Abstract

A variety of different contour-finding evaluation tools are used for assessment of functional ventricular indices. The aim of this study was to estimate the agreement of EDV, ESV and EF between non-model based (SW_{P − mri} = reference) and model-based (SW_{P − mri}) evaluation software applied to the same CMR datasets, acquired in 13 healthy children (mean age=10.3±3.5y; range: 5.3-16.2y). LV indices were compared with those gathered by realtime 3D echocardiography applying a similar model-based analysis strategy (SW_{T − echo}). Using Bland-Altman statistics, acceptable agreement was observed between SW_{P − mri} and SW_{T − echo} [EDV (-2.0%; -26.3 to 22.3%), ESV (-7.8%; -46.8 to 31.2%), EF (3.1%; -12.1 to 18.2%)] whereas a somewhat lower agreement with a higher scatter was found comparing SW_{P − mri} and SW_{T − echo} [EDV (-11.5%; - 37.4 to 14.4%), ESV (-5.2%; -52.0 to 41.6%), EF (-3.6%; -24.7 to 17.4%)]. EDV was underestimated by the model-based SWTmri by 1̃0% and EF by 7̃% when same CMR data were used. The non-model based algorithm for assessment of LV functional parameters is a flexible and universal applicable tool, whereas the model-based evaluation software might be helpful to additional detect cardiac motional disorders.