Tools for IMRT QA N. Dogan, Ph.D Department of Radiation Oncology Virginia Commonwealth University Medical College of Virginia Hospitals Richmond, VA, USA
Objectives To identify the QA tasks involving IMRT To describe the QA tools for all aspects of IMRT process To explain the limitations of the current IMRT QA tools To compare the IMRT QA tools and techniques
QA tasks for IMRT Machine QA- Acceptance and routine QA of the MLC for IMRT delivery - dosimetric and geometric characteristics Algorithm QA for IMRT - QA of planning system and data consistency with machine Patient Specific QA prove plan works 1D and 2D dosimetry of treatment components such IM beams and segments 3D dosimetry of entire treatment delivery Post Treatment QA Log-file analysis
Detectors Phantoms Scanners Dosimetric Analysis Tools
Detector Requirements for IMRT QA Geometric and dosimetric accuracy Volumetric simultaneously integrating dosimeter to faithfully quantify the dose delivered over the total time of treatment Good spatial resolution, tissue equivalent response Ability to provide 3-D information Portability to multiple phantoms Ease of use Sufficiently large dynamic range and be insensitive to photon energy spectrum and dose rate response which is independent of the energy spectrum
Detectors Many of them available for IMRT measurements Necessary to characterize the detector response for both static and dynamic fields for linearity Need to be calibrated for absolute measurements Need to determine stem and cable effects
Detectors IMRT QA Tools Need to determine energy dependence and angular response Small field detectors required for small field characterization Sensitive to position Detector should be smaller than homogeneous region of dose to be measured Assess electrometer response
Detectors, cont. Need to determine necessary resolution depends on the resolution of the beamlet grid that is used for planning and sequencing fields for delivery Chambers with the smaller volumes are more sensitive to position and will have a higher response when positioned at an opposing leaf pair junction and between adjacent leaves More stable measurement point Courtesy of Jean Moran, UofM Dose (cgy) 70 60 50 40 30 20 10 Poor detector position
1-D and 2-D Detectors Ion chamber (1-D) TLDs and MOSFETs (1-D) Detector arrays (2-D) Film (2-D) Radiographic Radiochromic Gels (3-D)
Small 1-D Detectors Detector Volume (cm 3 ) Diameter (cm) Disadvantages Microchamber 0.009 0.6 Poorer resolution than diodes Pinpoint chamber p-type Si diode 0.015 0.3 0.2 0.4 Over-respond to low energy photons Martens et al. 2000 Stereotactic diode NA 0.45 MOSFET NA NA Non-linear dose response for <30 cgy Diamond 0.0019 0.73 < resolution than diodes, dose rate dependence, expensive
Ion Chamber Advantages IMRT QA Tools Available in different shape and sizes Dosimetric response is well understood. Absolute dose measurements theory is well establish, they can be used as a benchmark standard Easy to calibrate
Ion Chamber, cont. Disadvantages Only one measurement point for each irradiation does not yield sufficient information to evaluate the dose throughout the target and/or critical structures Volume averaging the measurements are to be considered as an average throughout the chamber s active volume - does not yield significant errors if the ion chamber is placed in a low dose-gradient region even for relatively large chambers
Ion Chamber volume averaging, cont. Micro cham: 0.009cc PTW: 0.125cc Farmer:0.65cc D.A. Low et al. Ionization chamber volume averaging effects in dynamic intensity modulated radiation therapy beams, Med. Phys.30(7): 1706-1711 (2003
TLDs Advantages Multiple measurement points in a single irradiation Reusable Easy to use in multiple phantoms Small size and versatility in placement Readily available readout equipment Achievable accuracy: 2-3%
TLDs Disadvantages Requires calibration to determine calibration factor for each TLD chip Requires calibration of subset of TLD chips for each measurement TLD reader response and oven temperature should me routinely monitored to maintain consistent TLD response Automatic reader recommended for IMRT field verification due to large number of TLDs required for verification in a plane (60 or more) inefficient for routine IMRT QA
TLDs, cont. IMRT QA Tools D.A. Low et al. Phantoms for IMRT Dose Distribution Measurement and Treatment Verification, Int J Radiat Oncol Biol Phys 40: 1231-1235 (1998).
MOSFET systems Advantages Excellent spatial resolution small size (~0.04mm 2 ) Multiple detectors can be irradiated simultaneously Automatic and immediate readout Can be re-used immediately Linear dose response > 30 cgy Response independent of depth Commercially available phantoms to accommodate the small detectors
MOSFET systems Disadvantages Decrease linearity for < 30 cgy limited to high dose applications Over-response for the phantom scatter factor for small fields Specific application and measurement conditions should be carefully assessed and the detector should be used in the appropriate dose range
MOSFET systems IMRT QA Tools Bias Box Reader MOSFET TNRD50 system Courtesy of Cynthia Chuang, UCSF An axial image of MOSFET phantom
MOSFET systems IMRT QA Tools 1400 MOSFET Linearity Mosfet Consistency 1200 MOSFET1 MOSFET2 MOSFET3 MOSFET4 4.0 3.0 MOSFET1 MOSFET2 MOSFET3 1000 2.0 1.0 800 0.0 600-1.0-2.0 400-3.0 200 0 0 100 200 300 400 Radiation (cgy) -4.0 0 2 4 6 8 10 12 14 16 18 20 Number of Measurements Courtesy of Cynthia Chuang, UCSF
MOSFET systems IMRT QA Tools 1.2 1 Percent Depth Dose Comparison Ion Chamber MOSFET 2.0 1.5 1.0 Angular Dependence Percentage (%) 0.8 0.6 0.4 0.5 0.0-0.5-1.0-1.5-2.0 0.2 0 0 5 10 15 20 25 30 35-2.5-3.0 0 20 40 60 80 100 120 140 160 180 Degrees Depth (cm) Courtesy of Cynthia Chuang, UCSF
Cal. 1.64 Gy Meas. 1.72 Gy Diff 4.6 % Cal. 0.70 Gy Meas. 0.68 Gy Diff - 2.8 % Courtesy of Cynthia Chuang, UCSF Calc. 2.18 Gy Meas. 2.09 Gy Diff 4.35% Calc. 1.37 Gy Meas. 1.42 Gy Diff 3.52% Calc. 0.81 Gy Meas. 0.78 Gy Diff 3.45%
Current IMRT QA Tools 2-D Detectors Film Radiographic Radiochromic Beam imaging system, CCD, SLIC, AMFPI 2-D Detector arrays Diode array (Mapcheck) Ion chamber Active matrix flat panel detector (AMFPD)
Radiographic Film Advantages IMRT QA Tools Readily available (XV, EDR2, ) Can be cut into any desired shape Excellent spatial resolution (<1mm) Less expensive than other 2-D systems
Radiographic Film, cont. Disadvantages Over-response to low energy x-rays high atomic number of the active material not good for absolute dosimetry Dependent on QA of film batch Dependent on processor and digitizer Sensitive to storage conditions Need to measure the response to dose for each experiment H&D curve each time Proper normalization is critical
Current IMRT QA Tools Radiographic Film, cont. Other issues Store in a cool and dry place Make sure that the temperature for the film processor is stable Film digitizer pixel spacing, integrity of OD, beware of artifacts Verify spatial and optical density accuracy
Rapid Film Calibration 120 MU 90 MU 60 MU 30 MU 240 MU 210 MU 180 MU 150 MU Multiple dose levels per film-3x3 cm 2 fields of different dose levels Step-and-shoot or SMLC delivery Different dose values required for XV and EDR2 film (15-120MU for XV and 30-240MU for EDR2) Saves both time and film Childress et al Med Phys 29(10), 2002.
XV vs. EDR Film Depth-corrected H&D Net Optical Density 3.5 3.0 2.5 2.0 1.5 1.0 0.5 XV EDR XV2, 6 MV XV2, 15 MV EDR2, 6 MV EDR2, 15 MV Optical Density 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 Co60-EDR2 10MV-EDR2 6MV-EDR2 18MV-EDR2 0.0 0.0 100.0 200.0 300.0 400.0 500.0 600.0 Dose (cgy) Chetty and Charland 2002 PMB 47: 3629-3641 0 0 50 100 150 200 250 300 350 Dose (cgy) Dogan et al. 2002 PMB 47: 4121-4130
XV Depth-corrected H&D curves 1.6 1.4 6MV-EDR2-Depth corrected 6MV-EDR2-Regular EDR 1.6 1.4 18MV-EDR2-Depth corrected 18MV-EDR2-Regular Optical Density 1.2 1 0.8 0.6 0.4 0.2 0 0 50 100 150 200 250 300 350 Dose (cgy) Optical Density 1.2 1 0.8 0.6 0.4 0.2 0 0 50 100 150 200 250 300 350 Dose (cgy) Dogan et al. 2002 PMB 47: 4121-4130
Ion Chamber Film- depth corrected H&D Film regular H&D Dogan et al. 2002 PMB 47: 4121-4130 (a) (b) Ion chamber and EDR2 film depth-dose curves for a) 6 x 6 cm2, b) 14 x 14 cm2 films for 10 MV beam. Films were positioned parallel to the beam and OD to dose conversion was done using regular and depth-corrected H&D curves.
Childress et al. Med. Phys. 32(2) 2005
As compared to XV film, EDR2 film has less dependence on the processor, field size less response to low energy photons have better reproducibility and agreement with ion chamber measurements can be used to measure a complete fraction of an IMRT treatment
Radiographic Film: 2-D Dosimetric Measurements Intensity map from Opt System Calculated Leaf Sequencer Calc-Meas Courtesy of Jean Moran, UM Measured
Radiographic Film: Routine DMLC QA Using radiographic films Intensity-modulated pattern field Check leaf position, acceleration, motion stability Check for hot and cold density Visual check DMLC field 14x14 cm 2 at SSD =100 cm, 2 cm separated strips
Film Processor issues Should do routine maintenance and quality assurance verify spatial intensity, characteristic response, noise due to large changes in optical density ( Dempsey et al, Med Phys, 26; 1721-1731, 1999). Should be warmed up prior to use Should have appropriate amount of chemicals - Several films should be run in advance Should have stable temperature Should have a consistent rate of feeding into the processor
Film Other Issues Accurate positioning of the film in the phantom for the registration with treatment planning system Minimized errors by using a solid-water slab designed for film Have pins between slabs that puncture the film
Radiochromic Film (RCF) Advantages No significant energy dependence decreased sensitivity to low-energy photons Insensitivity to visible light Very high spatial resolution - well-suited for measurements in high-dose gradient fields Self-developing no developer or fixer is required Easy to handle Tissue equivalent
Radiochromic Film, cont. Disadvantages Takes a couple of hours for the color change to stabilize, and it may be necessary to wait up to two days before evaluating the film Sensitive to the air temperature and humidity Ultraviolet light may cause a color change without exposure to ionizing radiation Size, availability, and cost Non-uniform response to radiation double exposure technique minimizes this effect Issues with thermal history, wavelength dependence, and local sensitivity of the film
RCF (Gafchromic HS and MD55-2) vs radiographic films (XV and EDR2) O. Zeidan et al., Med. Phys., 31 (10):2730-2737 (2004)
RCF profiles vs. Ion chamber J. Dempsey et al., Med. Phys., 27 (10):2462-2475 (2000)
RCF digitizer issues Response of the digitizer Light source characteristics Design Gluckman et al, Med Phys, 29(8); 1839-1846, 2002.
Other 2-D systems IMRT QA Tools Beam imaging system, CCD, SLIC, Amorphous silicon flat panel detector (AMFPD) EPID systems attached to gantry Investigated more for pre-treatment QA currently 2-D Detector arrays Diode array (e.g; MapCheck) Ion chamber (e.g; LA48 linear array)
EPID Systems IMRT QA Tools Charged coupled device (CCD) camera systems Scanning liquid ion chambers (SLICs) Amorphous silicon flat panel detector (AMFPD) Active matrix flat panel imagers (AMFPIs) Transit Dosimetry Patient or Phantom Pre-Tx 2-D Measurements Film Replacement
EPID Systems as500 EPID 1 mm copper plate Phosphor scintillating layer (Kodak Lanex Fast B Gd2O2S:Tb, 70 mg/cm3) Array of photodiodes Amorphous Silicon panel each pixel consists of: IMRT QA Tools Light sensitive photodiode Thin film transistor 16-bit ADC Munro et. al, Med. Phys. 25, 1998
EPIDs Advantages Many centers have installed EPIDs and being primarily used for patient-specific pretreatment field verification and MLC QA Logical extension to investigate dosimetric applications Mounted to linear accelerator - known geometry with respect to the beam Detector sag must be accounted for at different gantry angles Positioning reproducibility important Real time digital evaluation No processor, data acquisition takes less time
EPIDs - Challenges EPIDs were primarily designed for patient localization High resolution, good contrast images Additional dose to the patient should be minimized The conversion of imager response to dose is complex Imaging system dependent Other problems Ghosting Lag
EPIDs Dose determination Imager response must be calibrated to a standard Absolute calibration to ion chamber at a point over a ROI E.g. ion chamber in a mini-phantom or slab at same SDD as EPID 2-D calibration to actual beam distribution at the imager plane Can be measured with film or a diode array
Factors for EPID Response Water-equivalent depth of the detector Field size dependence and scatter properties within the imager Short- and long-term reproducibility Dose rate Energy dependence Spatial integrity
EPID: DMLC measurements Overall: Good agreement + Predicted EPID Ion Chamber 10 MV 25 MV Discrepancies in the penumbra region (up to 10%) Pasma Med Phys 26: 2373-2378 (2376) 1999
Linear Diode Array in water vs. CCD Without short range penumbra correction Courtesy of Jean Moran, UofM
Dose Determination using EPID (SLIC) Chang et al., Int J Radiat Oncol Phys 47: 231-240 (p. 233)
Calculation vs. measured using AMFPD for DMLC Calculated (Calculated Measured) Agreement : Within +/- 2 cgy Courtesy of Jean Moran, UofM
EPIDs can provide a much-needed replacement for pre-tx QA film dosimetry Only if proper QA of the EPID is established Need better understanding of regions where EPIDs are inadequate for dosimetry Systems must be verified at more centers against accepted QA methods such as film and ion chamber Additional software is required before more facilities can do proper validation of the methods (Software must be commissioned) Can be part of a comprehensive QA program in conjunction with other methods such as computational checks (monitor programs, log file analysis, etc.)
Gel Dosimeters Advantages 3-D information in one irradiation Energy and dose-rate independent High sensitivity and linear response Cumulative Gel density can be changed - Ideal for anthropomorphic phantoms Near tissue equivalent Multiple readout techniques (MR, optical-ct) New gel formulations and readers commercially available
Gel Dosimeters Disadvantages Sensitive to time, preparation, temperature Cylindrical container required for optical readers - less accurate readout at gel/container interface MR time is often limited and expensive - long scan times for accurate readout, e.g. 5% accuracy over 10 hr scan time (Gum et al. 2002) Relative dosimeter -require cross-calibration technique batch to batch they are different Cost
Gel dosimetry IMRT QA Tools 8cm In-house optical CT scanner cost is less Oldham and Kim, Med. Phys. 31 (5), 1093-1104. Upgraded motors, motion control, and user interface. (Pacific Scientific: step motors. National Instruments: motion control and Labview.)
Gels: Optical Density to Dose Calibration 6 Beam calibration irradiation BANG gel phantom diameter 17.4cm Courtesy of Mark Oldham, Duke University
Gel Dosimeters IMRT QA Tools Five Field Prostate IMRT Courtesy of Mark Oldham, Duke University Re-computed for a 3 L BANG gel dosimeter. Dmax scaled to 1.8 Gy to fit dynamic range of optical scanner BANGkit TM from MGS Research. Optical-CT @ 1x1x3mm, 5hours
Gel Dosimeters IMRT QA Tools Isodose comparison: Pinnacle (red), Gel-dosimetry (black) Courtesy of Mark Oldham, Duke University
Gel Dosimeters IMRT QA Tools Gum, et al. Preliminary study on the use of an inhomogeneous anthropomorphic Fricke gel phantom and 3D magnetic resonance dosimetry for verification of IMRT plans, Phys Med Biol 47; N67-77 2002.
Phantoms for IMRT Measurements multiple phantoms for commissioning Fiducials for reproducible setup of phantom and detectors User-customized for different detectors allow special holders Simple vs. anthropomorphic Homogeneous or heterogeneous
Current IMRT QA Tools Simple Geometric Phantoms Water tank Accommodate different ion chambers Use for measurements of depth dose and profiles Output, flatness, symmetry, and linearity assessment Cylindrical mini-phantom Use with ion chamber to assess dependence of output on gantry angle Water-equivalent plastics: slab w/ custom chamber inserts 1-D and 2-D measurements Detector position can be varied with depth Cylindrical phantoms (plastic or water filled) Straightforward geometry Ion chamber at single position Plastic phantoms may hold films
Water-equivalent square IMRT Verification Phantom D.A. Low et al. Phantoms for IMRT Dose Distribution Measurement and Treatment Verification, Int J Radiat Oncol Biol Phys 40: 1231-1235 (1998).
Current IMRT QA Tools A Cylindrical Phantom containing movable ion chamber L. Xing et al. Dosimetric verification of a commercial inverse treatment planning system, Phys. Med. Biol. 44: 463-478 (1998).
A cylindrical Plastic Phantom Detector
Plastic Cylindrical Phantom with MOSFETs Calc. 1.37 Gy Meas. 1.42 Gy Diff 3.52% Calc. 0.81 Gy Meas. 0.78 Gy Diff 3.45% Courtesy of Cynthia Chuang, UCSF
Spiral Phantom IMRT QA Tools Paliwal et al A spiral phantom for IMRT and tomotherapy treatment delivery verification Med Phys (2000).
Anthropomorphic: RPC Head Phantom Target Volumes Water Critical Structure Removable Dry Insert Water TLDs in Target Volumes Courtesy of Jean Moran, UofM Radiochromic film through multiple plans Delivery is required by RTOG for participation in IMRT trials
Dosimetric Analysis Tools Provide a comprehensive and quantitative comparison between two dose distributions Different ones available Important to know the limitations
Dosimetric Analysis Tools Overlay of isodoses 2-D dose difference displays with colorwash Dose difference histograms Distance-to-agreement (DTA) Gamma evaluation Normalized agreement test (NAT)
Isodose lines and Dose Difference Display 70 cgy 60 cgy 50 cgy 20 cgy 10 cgy Calcs Film +/- 10% Courtesy of Jean Moran, UofM
Dose difference display Useful in shallow dose gradients Overly sensitive in steep dose gradients e.g.; a small spatial shift (due to experimental measurement errors) between two dose distributions yield large dose differences
Dose difference histogram and profiles
Distance to Agreement (DTA) Is the distance between a reference point and the nearest point in the compared dose distribution that exhibits the same dose Is not overly sensitive in steep dose gradients In shallow dose gradients, a large DTA value may be computed even for relatively small dose differences May be hard to interpret
Combination of dose difference and DTA Identify regions where the dose difference and DTA are simultaneously by greater than a pre-selected criteria points that fail both criteria are identified on a composite distribution The display of the dose difference may emphasize the impression of failure in high dose gradient region Provides no information on the magnitude of the failure
Gamma Analysis- Generalization of composite distribution Measures the closest distance between each reference point and evaluated dose distribution after scaling by D and d r ( r, r ) δ ( r, r ) Γ ( r, r ) = + = Γ r r r d D 2 2 e r e r e r γ (r 2 2 r ) m in ( e, r ) e { } { } rr (,): e r spatial distance between evaluated and reference dose points D : Dose difference criteria d : DTA The point with the smallest deviation from reference point is a quantitative measure of the accuracy of the correspondence -> the quality index, γ (r r ) of the reference point γ (r r ) : 1 ->correspondence is within the specified acceptance criteria Low et al, Med Phys 30(9) 2455-64 (2003).
Dose Difference and DTA Dose Difference and DTA Analysis Summary Dose Diff and DTA criteria : 2% of Dmax and 2mm Points Checked = 5348 Points Passed DTA = 5312 Points Passed DD = 4363 Points Passed Either = 5343 Points Passed Both = 4332 99.3269 % of the points passed DTA 81.5819 % of the points passed DoseDiff 99.9065 % of the points passed either Either 81.0022 % of the points passed Both Dose Difference Statistics Summary Mean Dose Diff = 0.488805 0.877915 DTA Summary Mean DTA = 0.0477486 0.0747123
Gamma Analysis Gamma Analysis Summary Dose Diff and DTA criteria : 2% of Dmax and 2mm Points Checked = 5348 Points Passed = 5348 100 % of the points passed Gamma Gamma Statistics Summary GammaBar = 0.0406743 0.0620086 Dose Diff and DTA criteria : 3% of Dmax and 3mm Points Checked = 5348 Points Passed = 5348 100 % of the points passed Gamma Gamma Statistics Summary GammaBar = 0.0271162 0.0413391
Normalized Agreement Test (NAT) Is based on a 2D array of calculated image of NAT values derived from comparisons of measured and computed doses. Assumes that two dose distribution images are registered each other and NAT is calculated using NAT NAT index scale ( δ 1) = D = A ve( NAT ) Ave( D ) scale δ : lesser of Abs( D/ D m ) or d/ d m D scale : D i /D max NAT index represents the average deviation from the D m and d m criteria for every dose pixel, ignoring the ones less than the set criteria N. Childress et al, The design and testing of noval clinical parameters for dose comparison, Int J. rad. Oncol Biol Phys 56(5) 1464-1479 (2003).
NAT Index N. Childress et al, The design and testing of noval clinical parameters for dose comparison, Int J. rad. Oncol Biol Phys 56(5) 1464-1479 (2003).
Other Analysis Tools MU check software In-house dose calc Commercial packages (e.g; Radcalc) Monte Carlo (e.g; Peregrine, EGS4, ) Patient QA Software for Post-treatment QA Analysis of IMRT delivery log files (e.g; inhouse analysis software, Argus IMRT QA package)
MC verification IMRT QA Tools Superposition Monte Carlo =10%
Summary Multiple detectors and phantoms are typically required for IMRT QA Quantitative dose analysis tools are necessary for proper evaluation of delivery - identify the cause of discrepancies between delivery and measurements Treatment planning vendors are starting to provide dosimetric evaluation tools Aware of the limitations of each tool
Summary Verify that all equipment is functioning properly Film processor, digitizer Detectors, cables, electrometers (automatic leakage correction) TLD reader, ovens Input/output to treatment planning system Standardize measurement setup when possible Monitor software and hardware changes and QA
Summary Measurements may show dosimetric differences that planning systems may not model at this time curved leaf ends Need to know the limits of the mechanical systems and interactions with controller and accelerator software for delivery Continued need for improvements to software for delivery system, measurement devices, phantoms, and dose analysis tools
Acknowledgements Jean Moran U of Michigan Cynthia Chuang UCSF Mark Oldham Duke University