Skip to main content

Table 2 Diagnostic performance of software detecting large vessel occlusion

From: A multicenter validation and calibration of automated software package for detecting anterior circulation large vessel occlusion on CT angiography

Threshold

of 0.50

Confusion matrix

Prediction

LVO

No LVO

Ground truth, LVO

173

20

Ground truth, no LVO

58

545

Sensitivity (95% CI)

0.896 (0.845–0.936)

Specificity (95% CI)

0.904 (0.877–0.926)

PPV (95% CI)

0.749 (0.688–0.804)

NPV (95% CI)

0.965 (0.946–0.978)

Optimal threshold

Youden (J) index (95% CI)

0.806 (0.741–0.843)

Jmax cutoff point

0.405

Jmax Sensitivity

0.912 (0.863–0.948)

Jmax Specificity

0.894 (0.867–0.917)

Fixed sensitivity of 0.90

Sens90 Specificity (95% CI)

0.902 (0.876–0.925)

Sens90 PPV (95% CI)

0.747 (0.686–0.801)

Sens90 NPV (95% CI)

0.966 (0.948–0.980)

Sens90 cutoff point

0.473

  1. Jmax represents, across all thresholds, the maximum Youden index (sensitivity + specificity − 1). As a secondary reference point, Jmax provides an optimality criterion with equal weighting for sensitivity and specificity. LVO = large vessel occlusion; CI = confidence interval; PPV = positive predictive value; NPV = negative predictive value
Back to article page

BMC Neurology

ISSN: 1471-2377

Contact us