P2A-5 Guided Automatic Segmentation of the Murine Left Ventricle Using Conservation of Myocardial Volume

Computation of important metrics of cardiac function - specifically left ventricular end systolic and end diastolic volume - necessitate accurate segmentation of myocardial boundaries in 3D image data sets. Automatic segmentation is highly dependent on appropriate selection of segmentation algorithm parameters. We propose the use of conservation of myocardial volume - that is, the difference in volumes between the epicardial and endocardial surface segmentations - as a metric of efficacy of a particular set of model parameters for the gradient vector flow field algorithm. The metric was found to be an effective predictor for the efficacy of the gradient vector field parameter - models generated from parameter values which exhibited strong conservation of volume yielded more accurate segmentations of both synthetic and in vivo data.