Table 4B.
Post hoc comparison results for the Individual Binary Threshold method.
| Diagnostic tasks | Tractography algorithm | (I) Threshold | (J) Threshold | Mean difference (I-J) | Sig. | 95% confidence interval |
|
|---|---|---|---|---|---|---|---|
| Lower bound | Upper bound | ||||||
| AD vs. MCI | Tensor-RK2 | 0.05 | 0.30 | -0.09325 | 0.023 | -0.01800 | -0.0065 |
| 0.40 | -0.08961 | 0.035 | -0.1763 | -0.0029 | |||
| Hough | 0.05 | 0.25 | -0.10897 | 0.034 | -0.2142 | -0.0038 | |
| MCI vs. NC | Tensor-TL | 0.05 | 0.35 | -0.10285 | 0.014 | 0.0109 | 0.1948 |
| PICo | 0.05 | 0.15 | 0.11037 | 0.028 | 0.0056 | 0.2151 | |
| 0.35 | 0.12817 | 0.004 | 0.0234 | 0.2329 | |||
| 0.40 | 0.12654 | 0.005 | 0.0218 | 0.2313 | |||
The “Sig.” column shows the SPSS adjusted p-value and only values below 0.05 are treated as nominally significant (Please refer to the footnote for detailed explanation). Only comparisons that passed Bonferroni correction are shown here. 95% confidence interval is on the mean difference (I-J). Using the Individual Binary Threshold as the feature extraction method, the AUCs from some tractography algorithms may be statistically affected by the threshold values chosen for specific diagnostic tasks.