Abstract #3359

Huang W, Sullivan J

An image registration using optimization of voxel similarities was presented, coupled with a hybrid genetic algorithm. It was shown to be a robust and accurate registration strategy. The registration quality was superior to the conventional alignment techniques. Significantly, the GA was not strongly sensitive to the initial start location nor was it susceptible to local minima/maxima. One of the greatest benefits of the GA is the lack of a required quality start location. The automated centroid mask coupled with the sequential application of GA alignments mimicked manual registration sequences. The GA demonstrated excellent alignments with an accuracy of over 98 percent. The mask strategy was proved to be an efficient method to substantially enhance both speed and alignment accuracy.