Diagnostic performance per patient for different imaging strategies
| Strategy | Sensitivity (95% CI) (TP/TP+FN) | Specificity (95% CI) (TN/TN+FP) |
| CTA1-150 | 0.80 (0.65–0.90) (36/45) | 0.91 (0.82–0.97) (63/69) |
| MRA1-150 | 0.71 (0.56–0.84) (32/45) | 0.97 (0.90–1.00) (67/69) |
| TCD | 0.73 (0.58–0.85) (33/45) | 0.91 (0.82–0.97) (63/69) |
| CTA+TCD | 0.83 (0.66–0.93) (29/35) | 0.98 (0.91–1.00) (58/59) |
| MRA+TCD | 0.76 (0.59–0.88) (28/37) | 1.00 (0.94–1.00) (61/61) |
| CTA+MRA | 0.79 (0.64–0.91) (31/39) | 1.00 (0.94–1.00) (61/61) |
↵1-150 Results used for CTA and MRA were those of the “better” observer, although the results for both observers were very similar (see note in discussion). For the “poorer” observer, sensitivity per patient (PP) was 0.82 and 0.71 for CTA and MRA respectively. For combination strategies, sensitivity PP was 0.86, 0.79, and 0.85 for CTA+TCD, MRA+TCD, and CTA+MRA respectively.
Where tests disagreed, the result was classified as “uncertain”, necessitating confirmatory IADSA rather than being classified as a true positive (TP), true negative (TN), false positive (FP), or false negative (FN) result. For CTA+TCD there were 20 “uncertain” cases, 16 for MRA+TCD and 14 for CTA+MRA.