Performance comparison of parallel finite element and Monte Carlo methods in optical tomography

Optical tomography is a promising medical imaging method which uses visible light. For optical tomography, the ability to solve the “forward problem” plays an important role. This problem involves predicting the distribution of light intensities inside and on the surface of the object using the knowledge of the optical structure of the object and the characteristics and position of the light source. A commonly used model for light propagation in turbid media a´s the diffusion equation. Two numerical methods, the Finite Element Method and the Monte Carlo Method, for solving the diffusion equation in optical tomography are presented, analyzed and compared in terms of their properties when being executed on a parallel computer system. To obtain performance predictions for both methods, processor and application performance models are developed permitting the determination of the conditions under which either one or the other method is superior