Table 3.
Classification of FTLD and controls based on DR, GMV, and WMV measures from various ROIs using logistic regression and area under a receiver operator characteristic curve (AUC). Specificity, sensitivity, and accuracy were evaluated using 4-fold cross-validations of the logistic regression results. p values are adjusted by false discovery rate and indicate the significance of the logistic regressions
| Measures | Sensitivity (%) | Specificity (%) | Accuracy (%) | AUC | p |
|---|---|---|---|---|---|
| Measures in ROI_1 | |||||
| GMV | 80.1 ±7.0 | 48.7 ±8.2 | 65.7 ±4.0b | 0.655 | 0.004 |
| WMV | 77.2 ±7.2 | 34.6 ±7.6 | 59.2 ±4.1 | 0.627 | 0.03 |
| DR | 79.9 ±6.8 | 72.3 ±9.8 | 76.0± 4.2a | 0.853 | <0.001 |
| Measures in ROI_2 | |||||
| GMV | 77.0 ±7.0 | 46.6 ±8.9 | 63.9 ±4.5b | 0.722 | 0.001 |
| WMV | 71.5±7.7 | 36.4 ±6.6 | 58.1 ±3.3 | 0.657 | 0.002 |
| DR | 80.7 ±8.4 | 80.5 ±9.8 | 81.4± 3.3a | 0.877 | <0.001 |
| Measures in ROI_3 | |||||
| GMV | 74.2 ±9.5 | 5.4± 3.1 | 45.7 ±4.2 | 0.566 | 0.22 |
| WMV | 79.8 ±8.2 | 5.3± 2.9 | 47.4 ±4.1 | 0.606 | 0.24 |
| DR | 73.3 ±6.9 | 58.6 ±8.6 | 67.6± 4.1a | 0.722 | <0.001 |
Bold: Differences in classification accuracy between modalities were significant by Wilcoxon signed rank tests.
a
DR showed better accuracy compared to GMV and WMV.
b
GMV showed better accuracy compared to WMV.