Commissioning an IMRT System for MLC Delivery Gary A. Ezzell., Ph.D. Mayo Clinic Scottsdale Taking the broad view of commissioning Commissioning elements • • • Validating the dosimetry system Commissioning the delivery system Commissioning the planning/delivery systems for dosimetric accuracy • Learning how to use the planning system for particular clinical problems Caveat: our experience is with Corvus + Varian Step 0: the dosimetry system • Decide on the system that will be used to measure IMRT dose distributions • Determine the measurement uncertainty - Compare measurements of unmodulated fields with this system to ion chamber scans • Needed to establish acceptance criteria Delivery system tests • Create test patterns independent of planning system - We use a text editor to create Varian files • Need to check - Effect of rounded leaf ends (if applicable) - MLC positioning accuracy - MLC dynamic motion accuracy (if applicable) - MU linearity over range of interest ... Radiation field “offset” for rounded leaf ends • For rounded MLC leaf ends, there is an offset between the light and radiation field edges: ~0.6 mm • Important in IMRT Light Radiation Important for IMRT because there are so many junctions within a field Measuring the offset • Irradiate a series of strips that create matchlines • Use different values of offset • Select offset value that gives best uniformity Measuring the offset No offset 0.6 mm offset 1.0 mm offset Here, 0.6 appears best, i.e. subtract 0.6 mm from MLC settings What is really needed?? most uniform at junction? Integrated dose? Check planning system requirements. Matchlines are very sensitive to positioning Offset 0.5 mm Offset 0.7 mm Offset 0.9 mm MLC accuracy • Tenths of millimeters are important for IMRT • Use patterns of matchline strips for a sensitive test - Test at different gantry and collimator angles - Select subset for routine QA DMLC tests • For dynamic MLC motion, leaf speed and dose rate also need to be controlled • Need to test for range of leaf speed and dose rates, as well as gantry and collimator angles • Create patterns that move a 1 cm gap using all leaves at same or variable rates Sweep all leaves at same rate Chamber reading at center should be proportional to MU Film should show uniformity Sweep leaves at different rates e.g. Travel 5, 7, 9, 11, 13 cm in same MU Check relative dose Special problem of matching accelerators • In ’99 Mayo Clinic Scottsdale had two “matched” Varian 2100C linacs, but IMRT doses differed by ~2.5% • Needed to adjust MLC calibration on one machine by 0.13 mm (Leaf gap error parameter) Delivery system QA • Sweep a 5mm gap across a 10 cm span with Farmer chamber at center • Take ratio to 10x10 open field • 2.5% for 0.2 mm change Ratio to stan dard 5m m g ap 1.100 1.050 1.000 0.950 0.900 -0.4 -0.2 0 0.2 G ap error 0.4 0.6 Routine QA checks for positioning accuracy • • • Matchline films 1 mm gap films Check at different gantry, collimator angles • Decide QA frequency Mayo monthly includes matchline films at 5 combinations of gantry, collimator Check visually. Patterns include programmed “deviations” 0.5 mm closed 0.5 mm open orientation 0.5 mm open 0.5 mm closed Mayo monthly includes one profile analyzed quantitatively Mayo monthly includes ratio of swept 5 mm gap to 10x10 open Similar test used with daily QA device Summary for the delivery system • Commission the delivery system independently of the planning system • Establish QA tests to monitor performance …on to the planning system Planning system: Dosimetric accuracy • For IMRT, the MLC leaves move through the area of interest • Final distribution is created by summing many beamlets • New things become important - Leakage through MLC leaves - Penumbra defined by MLC leaves - Small fields MLC leakage - measure average value • Leakage through leaf (~2%) • Between neighboring leaves (~5%) • Measure using a pattern that fully closes all leaves careful not to be under carriage or jaw Penumbra • Measure with film, diode, or microchamber, conventional scanning chamber too wide • Subtle effects make a difference in IMRT Beam model based on penumbra measured with chamber Beam model based on penumbra measured with film Dosimetric validation (1) Start with the basics • Define simple open fields irradiating a simple phantom - check calculated output, PDD, profiles against standard measurements - (just like commissioning any planning system) • And shaped static fields also 6MV 21EX Profiles 10x30 100 90 80 70 Relative dose 60 Meas D=5 50 Peregrine D=5 Corvus D=5 40 30 20 10 0 -25 -20 -15 -10 -5 0 Off axis 5 10 15 20 25 6MV 21EX Profiles 10x30 alternating 100 90 80 70 Relative dose 60 Meas D=5 Peregrine D=5 50 Corvus D=5 40 30 20 10 0 -20 -15 -10 -5 0 Off axis 5 10 15 20 Dosimetric validation (2) • Define simply-modulated fields with wide regions that can be measured with chamber and film (or equivalent) • Check: - Leakage/transmission - Basic modulation - Penumbra - Out of field dose “Leak100” – “Leak0” patterns Corvus/Meas Peregrine/Meas Leak100 0.979 0.987 Leak50 0.978 0.977 Leak20 0.976 0.970 Leak10 0.960 0.955 Leak0 0.000 0.984 “Ridge” and “Trench” patterns Useful for deciding minimum segment width 100% centers, 10% wings 10% centers, 100% wings “Incline” patterns 50% centers, 100% -10% wings Ridge10 Ridge40 180 160 160 140 140 120 120 Dose (cGy) Corvus Peregrine 80 Chamber Film 60 Dose (cGy) 100 100 C hamber Film 40 20 20 0 0 -5 Peregrine 60 40 -10 C orvus 80 0 5 -10 10 -5 -20 0 5 10 -20 cm off axis cm off axis Trench40 Trench10 100 100 90 80 80 70 60 Corvus Peregrine Chamber 50 40 Film C orvus Dose (cGy) Dose (cGy) 60 Peregrine C hamber 40 Film 30 20 20 10 0 -10 0 -10 -5 -10 0 cm off axis 5 -5 0 10 -20 cm off axis 5 10 5mm strips In c lin e 140 120 100 80 Dose (cGy) C o rvus P e re g rine 60 C ha m b e r F ilm 40 20 0 -1 0 -5 0 -2 0 c m o ff a x is 5 10 10mm strips In c lin e 140 120 100 Dose (cGy) 80 C orvus P eregrine C ham b er F ilm 60 40 20 0 -10 -5 0 -20 cm o ff axis 5 10 “Band100” – “Band0” patterns Bands0 140 120 100 C o rvu s 80 Dose (cGy) P e re g ri n e C ha m b e r F i lm 60 F i lm R e p e a t 40 C ha m b e r 20 0 -1 0 -5 0 -2 0 c m o ff a x is 5 10 10 60 0% 80 20 Finally, highly modulated single fields Random array What to do about differences? • • May need to adjust the beam model May need to live with it - That is, take known deficiencies into account when evaluating plans - Very important to know about it, especially for critical structures And once you are over the simple stuff … Dosimetric validation (3) • Define targets and structures in a simple phantom: mock clinical situations • Create inverse plan • Measure doses with chamber, film, … • Evaluate dose/MU in low gradient region • Evaluate isodoses on many planes Mock prostate Colors: film, Gray: plan Mock Head/Neck Colors: film, Gray: plan Mock spine Critical structure plane Target plane Colors: film, Gray: plan Another question • What is the role of quantitative assessment of individual IMRT fields vs. composite doses? Some problems show up more clearly on individual fields than composite distributions e.g. Dynamic delivery with Corvus/Varian Step/Shoot Dynamic Dosimetric validation (4) • Test calculations in the presence of inhomogeneities Measured/Corvus: 0.985 Dose/MU at 5 cm below 12 cm inhomogeneity Measured/Corvus Bone 1.002 Air 0.988 Lung 0.985 Water 0.978 LUNG 140 120 100 Meas d=1.25 Meas d=5.0 Meas d=10.0 Meas d=20.0 Corvus d=1.25 Corvus d=5.0 Corvus d=10.0 Corvus d=20.0 Peregrine d=1.25 Peregrine d=5.0 Peregrine d=10.0 Peregrine d=20.0 80 60 40 20 0 -15 -10 -5 0 -20 5 10 15 Dosimetric evaluation: Acceptance criteria • What kind of agreement to expect in high dose, low gradient regions? - ~2 - 4% is reasonable (chamber) • What kind of agreement to expect for isodoses? - Much harder to specify - ~2 - 4 mm for 50% - 90% lines How to be more quantitative? • Acceptance criteria should be stated statistically, such as: “within the region of interest, 95% of the calculation points should agree with the measurement to within 5%, which combines the acceptable degree of agreement with reality (3%) with the two-sigma uncertainty in the measurement technique (4%).” Questions • What is the region of interest? • What is the acceptable degree of agreement with reality? • What is the uncertainty in the measurement technique? • What percentage of the points should agree to the specified tolerance? Questions • Should the agreement be expressed as a percent of the local dose, prescription dose, or some other dose? • For the points that do not agree to that tolerance, is there an outer limit of acceptability? • Should distance to agreement be incorporated? Dosimetric validation - Summary • Know your measurement uncertainties • Start with simple situations that can be more easily analyzed • Design single beam tests that focus on specific parameters, e.g. - Transmission - Small or narrow fields - Penumbra and dose outside field • Finish with multiple fields and mock clinical situations Welcome to IMRT … there is a way out