Table 2.
Gamma pass-rate values comparing the dose distribution on reference CT with the dose distribution on sCTs.
| Number of centers in the train | Models | Gamma pass rate (%) | ||
|---|---|---|---|---|
| Test C1 | Test C2 | Test C3 | ||
| 1 | Model A (C1) | 96.5 ± 1.4 | 85.9 ± 10.1* | 96.1 ± 2.7 |
| Model B (C2) | 96.2 ± 2.4 | 97.9 ± 1.4 | 96.7 ± 1.7 | |
| Model C (C3) | 96.5 ± 2.0 | 97.3 ± 1.5 | 98.7 ± 0.9 | |
| 2 | Model D (C1+C2) | 96.4 ± 1.2 | 98.1 ± 2.2 | 96.2 ± 3.7* |
| Model E (C1+C3) | 96.5 ± 1.1 | 94.7 ± 4.5* | 92.2 ± 5.6* | |
| Model F (C2+C3) | 95.8 ± 2.7 | 98.0 ± 2.0 | 98.3 ± 1.6 | |
| 3 | Model G (C1+C2+C3) | 96.3 ± 1.3 | 98.1 ± 1.5 | 98.4 ± 1.4 |
The gamma criteria are local, 1%/1 mm, and 10% dose threshold.
The values are expressed as mean ± standard deviation. The Wilcoxon’s test was used to compare the gamma pass rate of the generic model with those of the other models.
*p-value< 0.05—significant difference.
Red cases indicate the monocentric models (model A/test C1, model B/test C2, model C/test C3) and blue cases represent the generic model (model G/test C1 or C2 or C3).