Subtraction CT angiography using non-rigid registration: The impact of similarity measure and image pre-processing

Subtraction CT Angiography is susceptible to motion artifacts especially when abdomen is the area under examination. The purpose of this study was to compare two non-rigid registration algorithms, one, using maximization of mutual information and the other, using sum of squared differences as a similarity measure and to evaluate their registration performance. Furthermore the impact of pre-processing of the images (native images versus edge weighted images) on the registration quality for registration of the abdomen prior to subtraction was assessed. Maximization of the mutual information performed better than sum of squared differences and edge weighted images performed better than native images especially when plaques were considered. Edge weighted images preserve the volume of plaques and bones, thus increasing the registration performance for subtraction CT angiography.