Introduction of non-linear models for characterization of shape and deformation statistics : application to contractility assessment of enzymatically isolated adult cardiocytes

Includes bibliographical references (p. 41-44).This thesis describes a novel approach to cardiocyte contractility assessment based on biomechanical properties of the cardiac cells, energy conservation principles, and information content measures. The measure of cell contraction is defined as being the distance between the shapes of the contracting cell, assessed by the minimum total energy of the domain deformation (warping) of one cell shape into another. To guarantee a meaningful vis-`a-vis correspondence between the two shapes, both a data fidelity term and a regularization term are employed. The data fidelity term is based on nonlinear features of the shapes while the regularization term enforces the compatibility between the shape deformations and that of a hyper-elastic material. The approach was tested by assessing the contractile responses in isolated adult rat cardiocytes and contrasted these measurements against two different methods for contractility assessment in the literature. The results show good qualitative and quantitative agreements with these methods as far as frequency, pacing, and overall behavior of the contractions are concerned. Furthermore, the warping algorithm implemented on a graphics processing unit (GPU) using Nvidia's CUDA. Several strategies for optimizing the kernel calls for the central difference gradient and the convolution are attempted. The strategies specifically attempt to reduce 'read redundancy,' i.e., the number of times that the same memory location is read during execution