A new approach for the quantification of complex lesion morphology: the gradient field transform; basic principles and validation results.

OBJECTIVES: This report describes the basic principles and the results from clinical evaluation studies of a new algorithm that has been designed specifically for the quantification of complex coronary lesions.
BACKGROUND: Currently used edge detection algorithms in quantitative coronary arteriography, such as the minimum cost algorithm, are limited in the precise quantification of complex coronary lesions characterized by abruptly changing shapes of the obstruction.
METHODS: The new algorithm, the gradient field transform, is not limited in its search directions and incorporates the directional information of the arterial boundaries. To evaluate its accuracy and precision, 11 tubular phantoms (sizes 0.6 to 5.0 mm), were analyzed. Second, angiographic images of 12 copper phantoms with U-shaped obstructions were analyzed by both the gradient field transform and the minimum cost algorithm. Third, 25 coronary artery segments with irregularly shaped obstructions were selected from 19 routinely acquired angiograms.
RESULTS: The plexiglass phantom study demonstrated an accuracy and precision of -0.004 and 0.114 mm, respectively. The U-shaped copper phantoms showed that the gradient field transform performed very well for short, severe obstructions, whereas the minimum cost algorithm severely overestimated the minimal lumen diameter. From the coronary angiograms, the intraobserver variability in the minimal lumen diameter was found to be 0.14 mm for the gradient field transform and 0.20 mm for the minimum cost algorithm.
CONCLUSIONS: The new gradient field transform eliminates the limitations of the currently used edge detection algorithms in quantitative coronary arteriography and is therefore particularly suitable for the quantification of complex coronary artery lesions.