Al-Khawarizmi Engineering Journal (Jan 2010)
Finite Element Based Solution of Laplace's Equation Applied to Electrical Activity of the Human Body
Abstract
Computer models are used in the study of electrocardiography to provide insight into physiological phenomena that are difficult to measure in the lab or in a clinical environment. The electrocardiogram is an important tool for the clinician in that it changes characteristically in a number of pathological conditions. Many illnesses can be detected by this measurement. By simulating the electrical activity of the heart one obtains a quantitative relationship between the electrocardiogram and different anomalies. Because of the inhomogeneous fibrous structure of the heart and the irregular geometries of the body, finite element method is used for studying the electrical properties of the heart. This work describes the implementation of the Conjugate Gradient iterative method for the solution of large linear equation systems resulting from the finite element method. A diagonal Jacobi preconditioner is used in order to accelerate the convergence. Gaussian elimination is also implemented and compared with the Precondition Conjugate Gradient (PCG) method and with the iterative method. Different types of matrix storage schemes are implemented such as the Compressed Sparse Row (CSR) to achieve better performance. In order to demonstrate the validity of the finite element analysis, the technique is adopted to solve Laplace's equation that describes the electrical activity of the human body with Dirichlet and Neumann boundary conditions. An automatic mesh generator is built using C++ programming language. Initially a complete finite element program is built to solve Laplace's equation. The same accuracy is obtained using these methods. The results show that the CSR format reduces computation time compared to the order format. The PCG method is better for the solution of large linear system (sparse matrices) than the Gaussian Elimination and back substitution method, while Gaussian elimination is better than iterative method.
