A Template Based Continuous Genetic Algorithm for Image Reconstruction

A template-based continuous genetic algorithm (CGA) image reconstruction methodology is proposed for tomographic image reconstruction. For the head and the lung phantoms, the proposed methodology has been found to yield higher image-quality in comparison with the conventional techniques including back-projection, filtered-back-projection, and simulated annealing techniques. Sensitivity of image-quality measures on various crossover operators including uniform, image-row, and block crossover operators has been studied and the image-row crossover has been found to yield the highest PSNR value at large values of crossover probabilities. The CGA, in terms of the PSNR values, has shown about 80% improvement compared with FBP for both head as well as lung phantoms and about 11% in the case of SA for 32 times 32 head phantom.