A quantitative method for the evaluation of spectral analysis patterns in carotid artery stenosis.

Duplex scanning in conjunction with Fast Fourier Transform Spectral Analysis offers a means of identifying stenotic internal carotid arteries, but few quantitative algorithms for interpretation of the spectral analysis patterns have been described. We first developed two objective parameters for analyzing carotid artery spectral analysis patterns which included a total of three peak frequencies in diastole and a ratio of spectral pattern areas between the internal and common carotid vessels. We then devised a formula for interpretation of the test results based on a computerized, best-fit, step-wise, discriminate analysis performed on 196 angiographically studied carotid vessels. This formula was then prospectively applied to 154 other carotid arteries with independently interpreted angiograms. 103/109 (94%) of the carotid arteries with less than 50% angiographic stenosis were correctly identified as were 40/50 (89%) of the vessels with 50% or more narrowing. Combining these groups with 16/17 (94%) properly classified total occlusions resulted in an overall accuracy rate of 93%. Application of the formula to smaller subgroups also showed significant differences in the formula means as they related to the degree of angiographic stenosis. Duplex scanning in conjunction with spectral analysis offers a reliable screening test for the evaluation of patients with suspected carotid artery stenosis.