An alternative approach to GTV margin determination in stereotactic body radiotherapy
José Bea-Gilabert, M. Carmen Baños-Capilla, M. Ángeles García-Martínez, Enrique López-Muñoz and Luis M. Larrea-Rabassa
Purpose: This study aims to estimate a realistic margin in stereotactic body radiotherapy (SBRT)
through examining the determination uncertainties of gross tumour volume (GTV).
Methods: Three computed tomography (CT) scans were performed on each patient in different sessions as a treatment simulation. Registration of the different CT image sets was based on the fiducial marks from two stereotactic guides. GTV was defined in each one of them, as well as both the encompassing (UNI) and overlapping (INT) volumes. This protocol was altered following imaging guided radiotherapy (IGRT) implementation, so tumour displacements could be corrected for. The patient was scanned without repositioning solely considering tumour intrafraction variations. In addition, isocentre and dimension variations were obtained for each patient and cohort. A Monte Carlo code was developed to simulate tumour volume, considering them as ellipsoids in order to study their behaviour. Lastly, the equivalent radius (Req) was defined for each of these volumes, experimental and simulated, and both RUNIeq and RINTeq values were derived by simple linear regression to the mean value RGTVmeq.
Results: The global margin M can be defined as this systematic error plus an additional residual random uncertainty, with values M = 3.4 mm for Body Frame, M = 2.3 mm for BodyFIX and M = 2.1 mm without repositioning. The experimental results obtained are in good agreement with simulated values, validating the use of the Monte Carlo code to calculate a margin formula.
Conclusions: Introducing IGRT is not enough to obtain a zero margin; consequently, the safety margin, dependent on tumour shape and size dispersion, can be evaluated using this formulation.
Keywords: uncertainty; margins; stereotactic body radiotherapy; image registration, ellipsoidal tumour shape; Monte Carlo simulation