Purpose: Rigid registration is usually performed to compare measured and calculated dose distributions. A precise spatial correlation does not exist between the two dose distributions, therefore comparison could be biased if one searching for a combination of dose registration and normalization that provides the highest gamma passing rate. Here, we propose a technique to correlate the two dose distributions precisely using fiducial-based registration for a fair quantitative evaluation of the overall uncertainty in the delivery process.
Methods: Four CT skin markers (Beekley MedicalÂ®) with 1.5 mm lead-free pellet were placed on a cylindrical 3D dosimeter (PRESAGEÂ®) for computed tomography (CT) simulation. Subsequently, the images were transferred to EclipseTM treatment planning system to create a treatment plan. After irradiation, the 3D dose distribution of the irradiated dosimeter was read out by an optical-CT scanner (OCT) developed at our institution. An automatic localization algorithm was developed in MATLAB to register the markers in the OCT images to those in the CT images. Three detected markers were used as registration points and the other one was used to evaluate the Target registration error (TRE). Due to the precise localization, dose normalization was performed without selection of a normalization point or a â€˜best matchâ€™. An IMRT plan was also delivered to demonstrate the process.
Results: The accuracy of the point-based registration was estimated by the mean and standard deviation of the Fiducial localization error (FLE), Fiducial registration error (FRE) and Target registration error (TRE) were estimated to be 0Â±0.22, 0.44Â±0.22 and 0.44Â±0.22 mm. In addition, the gamma passing rate (%GP) of an IMRT plan using the proposed registration were found to be 93%, 4% lower than the %GP using regular â€˜best matchâ€™ registration.
Conclusion: In this study, we improved the formalism for gamma index comparison to perform a more rigorous and fair dose comparison.