The accuracy and reproducibility of an algorithm to correct for geometric image distortion in quantitative coronary angiography

The authors have developed a method to correct for image distortion (so-called dewarp). The method models the distortion by a set of two-dimensional polynomials of which the degree is optimally adjusted. The model parameters are derived through imaging a cm-grid put against the image intensifier (II) input screen. The derived accuracy of the pixel size (scale factor) was better than 0.006% and the RMS length of the residuals was 0.03-0.05 mm for cine film and 0.04-0.09 mm for on-line video images. The reproducibility for a dewarped position was 0.01-0.04 mm for cine film and 0.04-0.07 mm for video images, depending on the II size (14/17/27 cm). The dewarped size of an object (vessel) reproduced between 0.04% and 0.11% for cine film and between 0.15% and 0.18% for video images, depending on the II size. The main source of variability turned out to be an unstable high-voltage of the electron optics in the II.