In vivo dosimetry measurements are accepted when the difference between measured and calculated dose is under 5%. A statistical analysis has been conducted to determine whether this tolerance matched the clinical practice for the studied localizations: pelvis, thorax, head and neck, breast.
The technical characteristics of the detectors were checked before being used in clinical practice. Then an automatic statistical analysis was implemented using the 2450 in vivo dosimetry measurements obtained during 1 year.
The global average is 1.10%, the standard deviation 2.46%and the percentage of out of level measurements 4.09%. By distinguishing the localizations, the 5%tolerance appeared to be too narrow for the breast localization.
Several investigations were initiated to justify the modification of the tolerance for the breast localization. They highlighted an underestimation of the calculated dose when high beam angles are set: a new correction factor was defined to take account this error. A specific tolerance was also specified for the breast localization.