Table 4.
Diagnostic performance of quantitative assessment of vertebral compression fractures (VCF)
| AUC [95% CI] | Optimal cutoff for malignancy1 | Sensitivity2 (%) [95% CI] | Specificity2 (%) [95% CI] | +LR2 | −LR2 | ||
|---|---|---|---|---|---|---|---|
| ROI1 | Signal drop | 0.94 [0.87, 0.98] | ≤ 18% | 71.9 [53.3, 86.3] (23/32) | 98.4 [91.5, 100] (62/63) | 45.3 [6.4,320.3] | 0.3 [0.2, 0.5] |
| FF from W and F images | 0.84 [0.76, 0.91] | ≤ 21% | 81.3 [63.6, 92.8] (26/32) | 77.8 [65.5, 87.3] (49/63) | 3.7 [2.2, 6.0] | 0.2 [0.1, 0.5] | |
| FF map | 0.82 [0.73, 0.89] | ≤ 16% | 65.6 [46.8, 81.4] (21/32) | 93.7 [84.5, 98.2] (59/63) | 10.3 [3.9, 27.6] | 0.4 [0.2, 0.6] | |
| ROI2 | Signal drop | 0.97 [0.91, 0.99] | ≤ 20% | 93.8 [79.2, 99.2] (30/32) | 93.7 [84.5, 98.2] (59/63) | 14.8 [5.7,38.3] | 0.1 [0.0, 0.3] |
| FF from W and F images | 0.94 [0.87, 0.98] | ≤ 13% | 90.6 [75.0, 98.0] (29/32) | 84.1 [72.7, 92.1] (53/63) | 5.7 [3.2, 10.2] | 0.1 [0.0, 0.3] | |
| FF map | 0.91 [0.84,0.96] | ≤ 14% | 90.6 [75.0, 98.0] (29/32) | 85.7 [74.6, 93.3] (54/63) | 6.34 [3.4, 11.7] | 0.1 [0.0, 0.3] |
1Optimal cutoff values in order to differentiate benign from malignant VCFs corresponding with the Liu method
2Sensitivity, specificity, +LR, and −LR calculated for optimal cutoff for malignancy
AUC, area under the curve; 95% CI, 95% confidence interval; +LR, positive likelihood ratio; −LR, negative likelihood ratio; FF, fat fraction; ROI, region of interest. Raw data are reported in parentheses wherever applicable