Skip to main content
. 2005 Jan;26(1):104–112.

TABLE 2:

Sensitivity, specificity, accuracy, and interobserver agreement in detecting stroke

Imaging Study True-Positive False-Positive False-Negative True-Negative Sensitivity (%) Specificity (%) Accuracy (%) κ
Nonenhanced CT
    Any sign 18 23 8 43 69.2 65.0 66.2 0.502
    Lost cortical ribbon 11 5 15 61 43.4 92.3 78.5 0.528
    Lost insular ribbon 7 3 19 63 26.9 95.0 75.8 0.591
    Basal ganglia hypodensity 15 10 12 56 55.8 85.0 76.7 0.478
    Dense artery 6 10 20 56 23.1 85.0 67.5 0.462
    ATLANTIS/ASPECTS hypoattenuations 115 77 139 1141 45.2 93.7 85.3 0.594
PCT
    TTP review 194 298 60 920 76.5* 75.6 75.7* 0.526
    MTT review 197 288 57 930 77.6* 76.3 76.5* 0.618
    rCBF review 158 111 96 1107 62.2 90.9* 86.0* 0.516
    rCBV review 128 89 126 1129 50.5 92.7* 85.4* 0.505
    Computerized automated method 173 94 81 1124 68.2 92.3* 88.1* NA

Note.—Follow-up CT or MR imaging was the criterion standard. Globally, accuracy was significantly greater with PCT than with nonenhanced CT: TTP and MTT maps are significantly more sensitive, whereas rCBF and rCBV maps were significantly more specific. NA indicates not applicable.

*

Significantly superior to nonenhanced CT (any sign).