Table 1.
Classification results according to SVM, PLS-DA and CART for discrimination of fascicular tissue of the nerve from the surrounding tissues
| Classification method | Feature selection | MCC | ACC | SENS (%) | SPEC (%) | PPV | NPV | TP | FN | FP | TN |
|---|---|---|---|---|---|---|---|---|---|---|---|
| SVM | Fit | 0.711 | 0.854 | 82.6 | 88.8 | 0.901 | 0.806 | 580 | 122 | 64 | 508 |
| SVM | PCA | 0.793 | 0.897 | 89.9 | 89.5 | 0.913 | 0.878 | 631 | 71 | 60 | 512 |
| SVM | Segments | 0.779 | 0.890 | 88.6 | 89.5 | 0.912 | 0.865 | 622 | 80 | 60 | 512 |
| SVM | Combined | 0.826 | 0.914 | 91.3 | 91.4 | 0.929 | 0.896 | 641 | 61 | 49 | 523 |
| PLSDA | 10PC’s | 0.814 | 0.907 | 92.5 | 89.5 | 0.864 | 0.943 | 494 | 40 | 78 | 662 |
| CART | Fit | 0.615 | 0.808 | 81.2 | 80.4 | 0.836 | 0.777 | 570 | 132 | 112 | 460 |
For SVM different feature selection methods are used: fit parameters, PCA, segments and a combination of the last three. For PLSDA, 10 principal components have been used (10PC’s). For the CART analysis, the fit parameters have been used as features
Matthews correlation coefficient (MCC see Eq. 2 text), accuracy (ACC = [TP + TN]/[TP + FN + FP + TN]), sensitivity (SENS = TP/[TP + FN]), specificity (SPEC = TN/[FP + TN]), positive predictive value (PPV = TP/[TP + FP]), negative predictive value (NPV = TN/[TN + FN]), true positive (TP), false negative (FN), false positive (FP), true negative (TN)