Noise in stenosis measurement using digital subtraction angiography

This paper examines statistical errors in the measurement of arterial stenoses by digital videodensitometry. Images of vessel phantoms were acquired using digital subtraction angiographic techniques with low concentrations of an iodine contrast medium and low levels of x-ray exposure. Effects of the spatial and temporal averaging of image information on signal-to-noise ratios in the stenosis measurement were of primary interest. The influences of iodine concentration, x-ray scatter, veiling glare, x-ray energy spectrum, x-ray exposure, and detective quantum efficiency of the system were also included in the theoretical analysis. The agreement between theoretical calculation and experimental measurement of a simulated vessel was verified using measured values of the imaging system parameters. With a 14.2 mg/ml iodine concentration, using 20 mR per image at the entrance to a 13-cm water phantom, and averaging over a 6-mm length of a vessel 6.2 mm in diameter, the standard deviation in a measurement of a vessel's relative cross-sectional area was about 0.05. The extension of these results to practical applications in vivo is discussed.