Table 2.
Performance in glioma grading based on 2021 WHO classification.
| Parameters | AUC | 95% CI | p-Value * |
|---|---|---|---|
| Dynamic | |||
| TAC | 0.80 | 0.69–0.91 | NA |
| Conventional static | |||
| SUVmax | 0.58 | 0.42–0.74 | 0.010 |
| SUVmean | 0.54 | 0.38–0.70 | 0.0052 |
| TBRmax | 0.56 | 0.39–0.73 | 0.0045 |
| TBRmean | 0.56 | 0.40–0.73 | 0.0071 |
| TLG | 0.57 | 0.40–0.74 | 0.0078 |
| Texture features | |||
| GLCM Homogeneity | 0.41 | 0.24–0.58 | 0.0003 |
| GLCM Energy | 0.44 | 0.26–0.62 | 0.0024 |
| GLCM Contrast | 0.60 | 0.43–0.76 | 0.037 |
| GLCM Correlation | 0.39 | 0.21–0.58 | <0.0001 |
| GLCM Entropy | 0.55 | 0.37–0.73 | 0.012 |
| GLCM Dissimilarity | 0.60 | 0.43–0.77 | 0.046 |
| NGLDM Coarseness | 0.73 | 0.57–0.89 | 0.54 |
| NGLDM Contrast | 0.66 | 0.49–0.83 | 0.22 |
| NGLDM Busyness | 0.40 | 0.22–0.58 | 0.0002 |
| Indices from shape | |||
| Sphericity | 0.63 | 0.46–0.80 | 0.11 |
| Surface | 0.41 | 0.22–0.60 | 0.0001 |
| Compacity | 0.51 | 0.30–0.71 | 0.0034 |
| Volume | 0.55 | 0.37–0.72 | 0.0044 |
| First-order features from Histogram | |||
| Skewness | 0.40 | 0.24–0.55 | 0.0001 |
| Kurtosis | 0.39 | 0.23–0.54 | 0.0001 |
AUC, area under the curve from receiver operating characteristics; 95% CI, 95% confidence interval; TAC, time–activity curve; SUV, standardized uptake value; TBR, tumor-to-background ratio; TLG, total lesion glycolysis; GLCM, grey-level co-occurrence matrix; NGLDM, neighboring gray-level dependence matrix. Test for grade 2 (n = 20) versus grade 3–4 (n = 41). * p-value in comparison to TAC type. ROC comparison did not find significant differences between TAC, NGLDM contrast, NGLDM coarseness and sphericity (p > 0.05) but TAC had better performance than all the others parameters (p < 0.046).