Reproducibility of dynamic computed tomography brain perfusion measurements in patients with significant carotid artery stenosis

Abstract

Perfusion computed tomography (PCT) determination is a minimally invasive and widely available technique for brain blood flow assessment, but its application may be restricted by large variation of results. To determine the intraobserver, interobserver, and interexamination variability of brain PCT absolute measurements in patients with significant carotid artery stenosis (CAS), and to evaluate the effect of the use of relative perfusion values on PCT reproducibility. PCT imaging was completed in 61 patients before endarterectomy, and in 38 of these within 4 weeks after treatment. Cerebral blood flow (CBF), cerebral blood volume (CBV), time to peak (TTP), and peak enhancement intensity (PEI) were calculated with the maximum slope method. Interexamination variability was evaluated based on perfusion of hemisphere contralateral to the treated CAS, from repeated examinations. Interobserver and intraobserver variability were established for the untreated side, based on pretreatment examination. Interobserver and intraobserver variability were highest for CBF measurement (28.8% and 32.5%, respectively), and interexamination variability was the highest for CBV (24.1%). Intraobserver and interobserver variability were higher for absolute perfusion values compared with their respective ratios for CBF and TTP. The only statistically significant difference between perfusion values measured by two observers was for CBF (mean 78.3 vs. 67.5 ml/100 g/min). The interexamination variability of TTP (12.1%) was significantly lower than the variability of other absolute perfusion measures, and the interexamination variability of ratios was significantly lower than absolute values for all the parameters. In longitudinal studies of patients with chronic cerebral ischemia, PCT ratios and either TTP or CBV are more suitable measures than absolute CBF values, because of their considerably lower inter- and intraobserver variability. Differences in CBF between two examinations as high as 30% may be considered as significant in such patients.