A COMPARISON OF SIMILARITY MEASURES FOR USE IN 2-D-3-D MEDICAL IMAGE REGISTRATION

A comparison of six similarity measures for use in intensity-based two -dimensional-three-dimensional (2-D-3-D) image registration is present ed. The accuracy of the similarity measures are compared to a ''gold-s tandard'' registration which has been accurately calculated using fidu cial markers, The similarity measures are used to register a computed tomography (CT) scan of a spine phantom to a fluoroscopy image of the phantom, The registration is carried out within a region-of-interest i n the fluoroscopy image which is user defined to contain a single vert ebra. Many of the problems involved in this type of registration are c aused by features which mere not modeled by a phantom image alone. Mor e realistic ''gold-standard'' data sets were simulated using the phant om image with clinical image features overlaid. Results show that the introduction of soft-tissue structures and interventional instruments into the phantom image can have a large effect on the performance of s ome similarity measures previously applied to 2-D-3-D image registrati on, Two measures were able to register accurately and robustly even wh en soft-tissue structures and interventional instruments were present as differences between the images. These measures were pattern intensi ty and gradient difference, Their registration accuracy, for all the r igid-body parameters except for the source to film translation, was wi thin a root-mean-square (rms) error of 0.54 mm or degrees to the ''gol d-standard'' values. No failures occurred while registering using thes e measures.