Abstract
This paper proposes an algorithm which maps the position of a catheter tip on a fluorograph to the 3D position in magnetic resonance angiography (MRA) data. This algorithm was assessed for its accuracy. We designed an algorithm consisting of a registration step and a recognition step. The registration step registers MRA and fluorography data using a digital subtraction angiography (DSA) image. The recognition step recognizes the position in the MRA data corresponding to the catheter tip position on a fluorograph. We checked the accuracy of the recognition step by employing an artificial data set consisting of 3D image data (64 × 64 × 64 matrix) and its projection image (92 × 92 matrix) and the accuracy of the registration step with the aid of three of the 3D time-of-flight MRA data sets (256 × 256 matrix and 60 slices) and their projection images in the form of DSA images. The accuracy of the recognition step depended upon that of the registration. When there was no misregistration, all of the mean errors were less than 0.2 mm. The mean errors of the registration step were 0.273 mm and 0.226 mm, respectively, for the longitudinal shift along the X and Y axes, 0.478°, 1.203° and 0.208°, respectively, for the rotation angles around the X, Y and Z axes and 0.020 times for the magnification. The mean image error between the projection image of the registered MRA data and that of the MRA data which were employed as the DSA image was 0.034 mm.