Simulation Studies on a Genetic Algorithm Based Tomographic Reconstruction Using Time-of-Flight Data from Ultrasound Transmission Tomography

Results of simulation studies on the application of genetic algorithms (GA) for solving an inverse problem, tomographic reconstruction, using time-of-flight (TOF) data from ultrasound transmission tomography are presented. The TOF data is simulated without taking into consideration the diffraction effects of ultrasound which is reasonably valid when the impedance mismatch in the specimen under consideration is small. The proposed GA based reconstruction algorithm is described and the results for a number of cases are discussed. The sensitivity of the proposed algorithm is studied for various GA parameters viz. the population size, maximum number of generations, crossover probability, and mutation probability. A time complexity analysis of the proposed algorithm shows that the reconstruction times and number of unknowns bears a near quadratic relation enabling the prediction of reconstruction times when dealing with higher resolutions. The performance of proposed algorithm to the reconstruction when TOF data is contaminated with noise is also analyzed and presented. The results obtained are found to be consistent for a wide range of resolutions, type, size, and shape of inclusions.