Measurement of Carotid Stenosis on Computed Tomographic Angiography: Reliability Depends on Postprocessing Technique

Purpose We previously demonstrated the validity of axial source (AxS) image quantification of computed tomographic angiography (CTA) visualized carotid stenosis. There is concern that AxS images may not accurately measure stenosis in patients with obliquely orientated stenosis and that measurements on axial oblique (AxO) multiplanar reformats (MPR), maximum intensity projections (MIP) images, or Doppler ultrasound (DUS) are superior. We tested the performance of AxS images against AxO MPRs, MIPs, and DUS techniques for stenosis quantification. Methods A total of 120 consecutive patients with CTA and DUS detected carotid disease were enrolled; carotids with occlusion, near occlusion, or stenosis <40% were excluded. Proximal and distal carotid diameters and North American Symptomatic Carotid Endarterectomy Trial (NASCET) style ratios were measured independently by 2 neuroradiologists on AxS, AxO, and MIP images on separate occasions in a blinded protocol. Intra- and interobserver agreements were determined for all measurements. The performance of different image types to identify ≥70% stenosis was assessed against a NASCET-style reference standard. Results Intra- and interobserver reliabilities for stenosis measurements were higher for both AxS (interclass correlation coefficients [ICC], 0.87–0.93 and 0.84–0.89) and AxO images (ICCs, 0.82–0.89 and 0.86–0.92) than for MIPs (ICCs, 0.66–0.86 and 0.79–0.82), respectively. Intra- and interobserver agreements on the NASCET ratio tended to be lower than proximal stenosis measurements. AxS and AxO image proximal stenosis measurements most accurately distinguished patients with ≥70% stenosis (0.90), followed by DUS (0.83) and MIP images (0.76). Conclusions A single AxS image stenosis measurement was highly reproducible and accurate in the estimation of carotid stenosis, which precluded the need for AxO MPRs.