Use of Monte Carlo modelling in radiotherapy linac design David Roberts, PhD Senior Physicist Elekta
Contents Overview of Elekta What we do Where we use Monte Carlo Codes and resources Example : Agility head leakage IEC60601-2-1 modelling and testing Other general examples of use of Monte Carlo Conclusion
Overview of Elekta
Elekta overview Elekta Neuroscience Elekta Oncology Global medical technology (3350 staff) Group within oncology and neurosurgery Close to 1,000,000 patients receive treatment with radiation therapy and radiosurgery equipment from Elekta Elekta Brachytherapy Solutions Elekta Software
Elekta overview Elekta Neuroscience Elekta Oncology Elekta Brachytherapy Solutions Monte Carlo used throughout the business Neuroscience (Sweden) Gamma Knife modelling Software (USA) Treatment planning systems Oncology (Crawley, UK and China) Linear accelerator system design Elekta Software
Elekta overview Linear accelerators that produce electron energies between 4 and 22 MeV Treatment either using electrons or photons Radiation collimation systems to conform the radiation dose to the tumour volume Versa HD
Codes Primarily two codes used Photons/Electrons BEAMnrc/EGSnrc Used for beam line physics Pegasos (Penelope engine) Used for beam line and out of field simulations
Why we use it From a linac development perspective Optimise design Reduce development iterations Increase reliability Reduce cost Reduce nasty surprises during the testing phase The main thing it reduces is trial and error experiments. Understanding the process and why things happen enables more effective designs. In house Monte Carlo mainly based on optimising design for compliance. Strong history of customer collaborations on projects (Ghent/Leeds Agility leaf bank design, Velindre kv CBCT dose)
Uses of Monte Carlo Improving design Electron applicators Improve reliability - Leakage to electronics Optimising shielding
Agility is an MLC collimation with the primary function that it accurately conforms the radiation beam to the treatment target. One of the important features is to reduce radiation leakage. This is the un-desired dose outside the treatment target. This increases the dose to organs at risk potentially limiting the effectiveness of the treatment and/or increasing side effects
Leakage occurs in 3 areas: Diaphragm and leaves (inside area M) Area M Inter-tip leakage Leakage outside area M Leakage is quoted as a percentage of the dose at isocentre for a 10 cm x 10 cm field
Diaphragm and leaves (inside area M) Ghent and Leeds modelled MLC design to minimise leakage through leaves Off focus MLC sides and tall (9 cm leaves) reduce leakage to less than 0.5% C Thompson et al.. Further Dosimetric Specification and Monte Carlo Modelling of a New Elekta Radiation Head with Integrated 160-leaf MLC, Estro 2010
Inter-tip leakage(inside area M) Important if island fields (two or more separate shapes) are being delivered Leakage determined by gap between tips and off axis position C Thompson et al.. Further Dosimetric Specification and Monte Carlo Modelling of a New Elekta Radiation Head with Integrated 160-leaf MLC, Estro 2010
Leakage outside Area M Our system are required to meet the IEC60601-2-1 safety standard. One clause involves ensuring that leakage radiation is below set values Inside patient plane (4 m x 4 m area) <0.2% max and 0.1% average Outside patient plane (1m radius cylinder and on covers) <0.5%
Measuring leakage is complex and very time consuming (weeks) Inside patient plane Measured used a 2 m x 2 m in-air scanner Moved over 4 quadrants A5 chamber (100 cc) used for coarse scanning Farmer chamber (0.6 cc) for hotspots Low signal high beam on times Test always at 10 and 25 MV (worst case leakage) Large x-y scanner to map 4 m x 4 m area
Measuring leakage is complex and very time consuming (weeks) Outside patient plane Measured used a 2 m x 2 m in-air scanner Additionally use film to identify hotspots on covers Long irradiation times for film (hours) Gafchromic film to detect leakage positions around the head
Requires a prototype system Iterations can have long lead times (manufacturing of new parts) From experiments it is not easy to predict where the leakage is coming from. Time consuming In-efficient use of material Most likely sources are target and flattening filter. Old process: Produce device and iteratively test and fix leakage (multiple iterations) New process Reduce iterations by early versions of device and identify leakage positions
T------G (metres) Modelling of head leakage for Agility First beta head simulated and showed possible high leakage when leaves in overtravel. Leaves are over travelled to the left. Results for 25 MV photons. -2 0.12-1 0 0.1 0.08 0.06 0.04 % Leakage Leakage ~0.15% (>0.1% limit) 1 0.02 2 Monte Carlo -2-1 0 1 2 A ---B (metres) 0
Confirmation of leakage on first prototype head Monte Carlo Experiment
Leakage point identified from 3D geometry
First beta head simulated ans showed possible high leakage when leaves in overtravel. Leakage point identified from 3D geometry
First beta head simulated ans showed possible high leakage when leaves in overtravel. Leakage point identified from 3D geometry
First beta head simulated ans showed possible high leakage when leaves in overtravel. Leakage point identified from 3D geometry
First beta head simulated ans showed possible high leakage when leaves in overtravel. Leakage point identified from 3D geometry
Shielding designed to fix issue. Experiment
First initial beta head experimental data showing high leakage when leaves in full overtravel Re-run simulation to confirm fix Experiment
First initial beta head experimental data showing high leakage when leaves in full overtravel Experiment (before) MC (After)
Conclusion Monte Carlo widely used for machine design Head leakage modelling Placement of Tungsten blocks on leaf bank reduces leakage from 0.15% to around 0.08%. Only one design iteration of shielding required The earlier modelling can be performed the early mechanical designs can be completed. Experiment (before) MC (After)