Videodensitometric quantitation of stenosis: in vitro and in vivo validation

Percentage of stenosis of a vascular lumen was quantitated using a digital subtraction angiography system with videodensitometric analysis. To validate the algorithm and methods, we used precisely drilled Lucite rods of three inner diameters and various reduced diameters to give known luminal reduction. Both in vitro and in vivo results of stenosis measurements resulted in an excellent correlation between actual and measured values (correlation coefficient greater than 0.9 for all trials). Consistent underestimation of stenosis of about 5% was attributed to the various image degradations inherent with the imaging procedure and equipment. Errors were greatest with midrange stenoses and less at the extremes of the lesion values. An insufficient signal-to-noise ratio was caused by low photon flux and/or low contrast material concentration and overlying vasculature filled with contrast material. Overall, the videodensitometric technique is accurate, easy to implement, objective, and relatively free of errors associated with the geometric/edge detection method.