Table 1.
The OAR anatomical features analyzed in the algorithm implemented by Yuan et al. There were 11 features for each OAR relative to one planning target volume (PTV). One primary PTV and one boost PTV could be included. In this study, only the primary PTV was included
| Anatomical features |
|---|
| Distance‐to‐target histogram (DTH) principal component 1 (PC1) |
| DTH PC2 |
| DTH PC3 |
| Fraction of OAR volume overlapping with the PTV |
| Fraction of OAR volume outside the treatment field |
| OAR volume |
| PTV volume |
| OAR wrap angle around the PTV |
| (DTH PC1)2 |
| PTV dose volume point 1 (PTV D2%)a |
| PTV dose volume point 2 (PTV D50%)a |
a
PTV dose volume points were included to take into consideration of the OAR sparing variation among plans, since overly spared OAR can result in less homogenous PTV dose. This variation was adjusted by standardizing the dose volume point of training cases and set 0 for the new cases.