Radiographic assessment of the extracranial internal carotid artery.

The accuracy of new (non-invasive) techniques of carotid artery assessment is classically established by comparing the results of the method under study with those of conventional contrast angiography. It is, therefore, important to establish the reproducibility of data acquired from contrast angiograms. Categories of carotid lesion must be chosen such that good reproducibility is likely both on an inter and intra-observer basis. This paper suggests that the smallest useful number of categories be used. It a two category discriminating system (less than 25% and greater than or equal to 25% stenosis), inter-observer agreement was found to be 95% and intra-observer agreement 96%. The principal observer variation occurred at the 0% and greater than 0% level of discrimination and this should be avoided in future studies comparing new arterial assessment techniques with conventional angiography.