Purpose: Original TomoTherapy systems may involve a trade-off between conformity and treatment speed, the user being limited to three slice widths (1.0, 2.5, and 5.0 cm). This could be overcome by allowing the jaws to define arbitrary fields, including very small slice widths (<1 cm), which are challenging for a beam model. The aim of this work was to incorporate the dynamic jaws feature into a Monte Carlo (MC) model called TomoPen, based on the MC code PENELOPE, previously validated for the original TomoTherapy system. Methods: To keep the general structure of TomoPen and its efficiency, the simulation strategy introduces several techniques: (1) weight modifiers to account for any jaw settings using only the 5 cm phase-space file; (2) a simplified MC based model called FastStatic to compute the modifiers faster than pure MC; (3) actual simulation of dynamic jaws. Weight modifiers computed with both FastStatic and pure MC were compared. Dynamic jaws simulations were compared with the convolutionsuperposition (CS) of TomoTherapy in the cheese phantom for a plan with two targets longitudinally separated by a gap of 3 cm. Optimization was performed in two modes: asymmetric jaws-constant couch speed (running start stop, RSS) and symmetric jaws-variable couch speed (symmetric running start stop, SRSS). Measurements with EDR2 films were also performed for RSS for the formal validation of TomoPen with dynamic jaws. Results: Weight modifiers computed with FastStatic were equivalent to pure MC within statistical uncertainties (0.5% for three standard deviations). Excellent agreement was achieved between TomoPen and CS for both asymmetric jaw openingconstant couch speed and symmetric jaw openingvariable couch speed, with deviations well within 2%/2 mm. For RSS procedure, agreement between CS and measurements was within 2%/2 mm for 95% of the points and 3%/3 mm for 98% of the points, where dose is greater than 30% of the prescription dose (gamma analysis). Dose profiles acquired in transverse and longitudinal directions through the center of the phantom were also compared with excellent agreement (2%/2 mm) between all modalities. Conclusions: The combination of weights modifiers and interpolation allowed implementing efficiently dynamic jaws and dynamic couch features into TomoPen at a minimal cost in terms of efficiency (simulation around 8 h on a single CPU).
|Number of pages||9|
|State||Published - Sep 2011|
Bibliographical noteFunding Information:
E. Sterpin is a research fellow of the Belgian ‘Fonds National pour la Recherche Scientifique’ (Charge de recherches du F.R.S—FNRS FC 73512). This work was also supported by the Herb Attix fund through the University of Wisconsin Foundation and TomoTherapy, Inc., (Madison, WI). T. R. Mackie and W. Lu have financial interest in TomoTherapy, Inc.
- Monte Carlo
ASJC Scopus subject areas
- Radiology Nuclear Medicine and imaging