Purpose: Non-contrast images of the head are used for evaluating stroke severity in the acute setting, using the Alberta Stroke Program Early CT Score (ASPECTS). Our purpose was to assess the correlation between the ASPECT score using non-contrast images and Virtual Non-Contrast (VNC) images of the brain, acquired on a dual layer spectral CT scanner.
Materials and Methods: Retrospective assessment of CT scans of the head, between January 2017 and February 2017, was performed. Scans were included if they were indicated for symptoms of acute stroke, within the therapeutic window for thrombolytic or endovascular treatment. Scans were performed on an iQon dual layer 64 slice CT (Phillips, Netherlands). CT Angio protocol for acute stroke patients in our institution includes a tri-phasic scan. Images for evaluation were obtained from the non-contrast scan and from VNC images which were reconstructed from the spectral data of the venous phase scan. The scoring was performed independently by two radiologists using the ASPECTS method, on an axial image at the basal ganglia level and at the supra-ventricular level. Scores were compared and subjected to statistical analysis.
Results: 11 CT scans were included in the study. One scan was excluded due to poor VNC image quality. ASPECT scores ranged between 9-10 for reader 1 and between 8-10 for reader 2. There was poor inter-rater agreement for the scores of the non-contrast scans (r2=0.13), however this was slightly higher for the VNC images (r2=0.26). Additionally, there was poor correlation between the scoring of the non-contrast images and the VNC images for both readers independently (r2=0.05 and r2=0.003 for reader 1 and 2, respectively).
Conclusion: Stroke scoring using VNC images does not correlate with scoring of conventional non-contrast images and thus is not a suitable replacement as the modality for ASPECT scoring in the setting of acute stroke.
