Example: random uncertainties for target OAR
Random variation of target D
90,
e.g. random inter-observer variation
•
∆
D
2cm³
=±10%/fx -> observed morbidity 7.5 %
(vs 7.3% model prediction)
•
∆
D
2cm³
=±20%/fx -> observed morbidity 8.9 %
•
∆
D
2cm³
=±30%/fx -> observed morbidity 10.5 %
•
Model prediction: 10.5% NTCP@ 80Gy EQD2
Random variation of rectum D
2cm³
e.g. random intra-/inter-fraction variation of
organ position or shape
For target – differences in TCP < 0.5%
“Can reduction of uncertainties in cervix cancer brachytherapy potentially
improve clinical outcome?” Nesvacil et al. 2016, submitted to R&O