Morphometry versus densitometry--a comparison by use of casts of human coronary arteries

Geometric and densitometric methods for quantitative coronary arteriography have generally been compared by use of phantoms simulating arteries with circular lumina ('Hole phantoms'). We have used more adequate phantoms obtained by casting disease-free and atheromatous human coronary arteries. The phantoms, filled with contrast medium, were imaged digitally (1024 x 1024 x 10 matrix) under experimental conditions simulating routine coronary angiography. The angiographic 'diameters' and the densitometric cross-sectional areas of 59 marked lumina were determined in single plane and orthogonal biplane raw images. Geometric calibration was performed by help of a 7F coronary catheter. For the densitometric calibration, we used a 'hole phantom' attached to the image intensifier. The obtained luminal areas were compared to their true values determined previously by planimetry. The mean absolute error of single plane cross-sections obtained geometrically was 1.53 mm2. Biplane imaging reduced it by a factor 2.4 to 0.64 mm2. The corresponding mean absolute errors for densitometry were 0.56 mm2 and 0.51 mm2. Single plane 'diameter' measurements appear thus of very limited value for hemodynamic conclusions. In contrast, biplane geometric quantification was not markedly inferior to single plane and biplane densitometry.