A novel explicit 2D+t cyclic shape model applied to echocardiography.

In this paper, we propose a novel explicit 2D+t cyclic shape model that extends the Point Distribution Model (PDM) to shapes like myocardial contours with cyclic dynamics. We also propose an extension to Procrustes alignment that removes pose and subject size variability while maintaining dynamic effects. Our model draws on ideas from Principal Component Analysis (PCA), Multidimensional Scaling (MDS) and Kernel PCA (KPCA) and solves 3 shortcomings of previous implicit models: (1) cardiac cycles in the data set do not each need to have the same number of frames, (2) the required number of subjects for statistically significant results is substantially reduced and (3) the displacement of contour points incorporates time as an explicit variable. We illustrate our method by computing models of the myocardium in the 4 principal planes of 2D+t echocardiography data.