Quantitative analysis of 3D mitral complex geometry using support vector machines

Quantitative analysis of 3D mitral complex geometry is crucial for a better understanding of its dysfunction. This work aims to characterize the geometry of the mitral complex and utilize a support-vector-machine-based classifier from geometric parameters to support the diagnosis of congenital mitral regurgitation (MR). The method has the following steps: (1) description of the 3D geometry of the mitral complex and establishment of its local reference coordinate system, (2) calculation of geometric parameters and (3) analysis and classification of these parameters. With a control group of 20 normal young children (11 boys, 9 girls, mean age 5.96 ± 3.12 years) and with the normal structure of mitral apparatus, 20 patients (9 boys, 11 girls, mean age 5.59 ± 3.30 years) suffering from severe congenital MR are studied in this study. The average classification accuracy is up to 90.0% of the present population, with the possibility of exploring quantitative association between the mitral complex geometry and the mechanism of congenital MR. (paper)