AAPM Summer School on IMRT Colorado Springs 2003 Commissioning and Quality Assurance For IMRT Treatment Planning M.B. Sharpe, Ph.D. Department of Radiation Oncology Princess Margaret Hospital University of Toronto Toronto, Ontario, CANADA M5G 2M9 michael.sharpe@rmp.uhn.on.ca IMRT: A Clinical Perspective • • • • • • Unique power to create and manipulate dose gradients Detailed quantitative treatment objectives Collimator movement during treatment. Increased number of treatment monitor units Need to control and determine treatment margins through objective measurement of set-up uncertainties internal organ motion. All aspects of the radiotherapy process should be reexamined under more stringent requirements for accuracy and precision. AAPM Summer School 2003 IMRT: A Technical Perspective • • • • IMRT is an extension of “3D Conformal” practices. Existing recommendations for RTP QA are applicable. Quality of IMRT/RTP plans depends on “up- and down- stream” quality processes; i.e., QA of CT, MR, as well as R&V, accelerator. QA procedures for planning are linked closely with commissioning, where base-line performance criteria are established. AAPM Summer School 2003 Commissioning and QA • • • • • Information management (networking) MLC: alignment, position, leaf speed, leaf leakage Small-field dosimetry Beam stability for short irradiations Because of inter-play between dose rate and leaf velocity, tolerances are stricter for dynamic MLC. • Disease-specific commissioning • feasibility studies • inter- & intra- fraction setup variation & organ motion, and potential interplay with multi-segments delivery • procedure validation AAPM Summer School 2003 Commissioning and QA • ISO Précis: • “Everything required to make everything right”. • Components: • • • General validation Procedure validation Routine checks (Monitor for change/deviations) • Confidence comes from evidence. • Errors often come from unforeseen events. AAPM Summer School 2003 RTP Commissioning & QA • • • • • • • • Accept hardware and software. Measure and enter basic dosimetric data, machine geometry, other operational parameters. Tune algorithm for best performance in anticipated clinical situations. Verify dose algorithms and associated configuration parameters. Configure import interfaces for imaging systems. Verify image quality and geometry after data transfer. Configure export interfaces to treatment machines Learn how to interact with the system and apply it to clinical cases. AAPM Summer School 2003 1 IMRT Planning System QA • • • • Geometry • Image and Structure Import • Multiplanar reconstructions • Beam and DRR geometry, display, and export Image Segmentation • 3-D display • Automatic tools - auto-contouring,auto-margin, etc • Dose volume histograms (DVH) Dose Calculation Algorithms Plan Evaluation Tools AAPM Summer School 2003 Image and Structure Import • Vendor-specific (proprietary) interfaces have given way to the DICOM Standard. • DICOM main capabilities, including: • Network Information Transfer • Open Media Interchange • Integration within the Healthcare Environment • Open standards should be endorsed, used, and reinforced. AAPM Summer School 2003 Image and Structure Import • • • IMRT is driven by volumetric segmentation. • Greater need to assure diligent care and accuracy of the segmented structures. Segmentation guidelines outlined by TG 53 The interaction of segmentation with the processes of IMRT depends on • staff awareness, • adequate time • Resources for adherence to clinical protocols • • AAPM Summer School 2003 Patient Orientation AAPM Summer School 2003 Evaluating Transfer of Contours Images may (or may not) be manipulated, e.g., • • • Contours and IMRT resample on import into RTP system. Slices reordered or mirrored. Examine effect of slice thickness and image resolution on target volumes and their effects on contour position. Potential error in the position of the isocentre AAPM Summer School 2003 4 AAPM Summer School 2003 2 Slice Thickness & Image Resolution • • • • • Image & Contour Import QA CT Slice Thickness 2.5mm, DFOV 25 cm, 5 mm phantom grid Phantom with 5 mm grid spacing CT scan with progressively larger Display Field of View Draw contours to coincide with the grid. Transfer images and contours from Virtual Simulator to RTP system. Measure mismatch of the contour with the image grid. After DicomRT transfer CT Sim AAPM Summer School 2003 AAPM Summer School 2003 Image & Contour Import QA CT Slice Thickness 2.5mm, DFOV 50 cm, 5 mm phantom grid Image & Contour Import QA DFOV cm 25 40 Area cm2 Target 1 6.57 6.61 6.45 6.6 Area cm2 Target 2 6.12 6.10 6.08 6.2 Area cm2 Ext. cont. 150.16 152.03 151.03 154 Contour Shift mm <1 1 2 0 After DicomRT transfer CT Sim AAPM Summer School 2003 Calculated value 50 AAPM Summer School 2003 Van Dyk & Craig Auto Segmentation: External Surface Beam Geometry Phantoms Dependence of Segmentation on Auto threshold values. Threshold values Contour Area cm2 From Image Actual -150 150.56 6.92 7 -500 152.54 6.97 7 -800 153.57 6.99 7 AAPM Summer School 2003 Isocentre depth cm Rotatable component • Beam geometry • Gantry rotation • Couch rotation Body component • Quantify volumes • Contour expansion • CT number conversion AAPM Summer School 2003 3 Reconstructed Image Verification e bl Ta m n io ot Oblique CT reconstruction Digitally reconstructed radiograph AAPM Summer School 2003 AAPM Summer School 2003 CT HU’s to Electron Density Conversion Dose Calculation Algorithms Needs: System A System B System C Battista and Bronskill 2000 1500 General Flexible Accurate Fast CT Number 1000 500 0 -500 -1000 0.0 0.5 1.0 1.5 Relative Electron Density 2.0 2.5 electron density phantom AAPM Summer School 2003 • Relate dose calculation in patient to beam calibration conditions AAPM Summer School 2003 Dose Calculation Methods Data measured in water Dose: Convolution/Superposition D( x , y, z ) = ∫ ∫ ∫ Φ ( x' , y' , z')K( x − x' , y − y' , z − z')dxdydz Parameterize water data Reconstitute water data Calculate corrections to water data Calculate dose directly based on beam and phantom configurations “Correction” based methods “Model” based methods AAPM Summer School 2003 AAPM Summer School 2003 4 Dose: Convolution/Superposition D( x , y , z ) = A Simple Algorithm Check ∫ ∫ ∫ Φ ( x' , y' , z)K ( x − x' , y − y')dxdy • 20 X 20 cm2 field, 18MV • 50 X 50 X 50cm3 water phantom • 200cGy to 22cm depth z AAPM Summer School 2003 • • Introduce air inhomogeneities, 1cm wide mediastinum, 2cm surface layer • Contour correction: 1cm2 wide “spike” • Contour correction: 25cm2 wide “spike” AAPM Summer School 2003 A Simple Algorithm Check: MU’s System A System B homo/hetero homo/hetero 242.7 / 242.0 244 / 244 246.8 / 260.7 244 / 244 321.7 / 321.0 244 / 244 279.7 / 278.8 244 / 244 Dose Computation: Why Model Based? • • • IMRT = Σ small fields Dose = function(penumbra+leakage+head scatter) Need accurate treatment head model to get this right IMRT IMRT beam profile AAPM Summer School 2003 XRT XRT beam profile AAPM Summer School 2003 IMRT Can Include Small Fields IMRT Can Include Small Fields Intensity Profile possibly unconstrained intensity levels • • Accuracy of dose model at small field sizes is a consideration Convolution-superposition or Monte Carlo desirable Measurement Conventional RTP Reconstituted from Beam Segments Intensity Grouping includes MLC constraints limit delivery to a few discrete intensity levels AAPM Summer School 2003 IMRT AAPM Summer School 2003 5 Boyer et al, al, Med. Phys. 24: 757 Extend the Beam Model • • X-Ray Field Edge Minimum Equivalent Square • XRT: ~4x4cm2 • IMRT: ~0.5x1cm2 - 2x2cm2 How to get small field data? • Direct measurement – difficult, errors • Extrapolation of existing data – assumes an underlying X-Ray Field Edge = 1 HVL physical model • Either case will have uncertainty; evaluate its impact on a clinical IMRT distribution generated by the inverse planning system AAPM Summer School 2003 AAPM Summer School 2003 Boyer et al, al, Med. Phys. 24: 757 Light and X-Ray Field Edges AAPM Summer School 2003 AAPM Summer School 2003 Curved Leaf Face Correction Without correction AAPM Summer School 2003 Curved Leaf Face Correction With correction Density Scaling Approximation • terma and kernel are computed for water and scaled by the average density computed along raylines. AAPM Summer School 2003 6 Calculated Data • Inhomogeneity Corrections Woo, Cunningham et al 1990 Sharpe, Battista 1996, Superposition AAPM Summer School 2003 White et al (1996) AAPM Summer School 2003 Dose Calculation Dose Calculation: Primary Rays Only Beam j φkj represents intensity of ray k from beam j. 1 Primary only φk-1j φkj M di 3D scatter Number of step&shoot fields >120 for tolerance level = 7% φk+1j AAPM Summer School 2003 AAPM Summer School 2003 Dose Calculation: Primary + Scatter Dose Algorithms Continue to Evolve Monte Carlo Number of step&shoot fields <50 for tolerance level = 5% AAPM Summer School 2003 AAPM Summer School 2003 7 Dose Algorithms and IMRT • • AAPM Summer School 2003 AAPM Summer School 2003 • 5 mm 1 3 10 mm Actual AAPM Summer School 2003 Big Large Small Big Large 100 80 60 40 20 0 2 Small 120 Medium 3 mm Medium 140 1 mm dose computation and dose evaluation. The volume chamber is mounted sagittally, replacing the grid in Lucy phantom. Scans performed with varying slice thickness and volumes are contoured and compared with actual volumes Big • Interplay of imaging and Large Effect of Slice Thickness Small Effect of Slice Thickness • Contours are visualized and used for dose scoring. For IMRT, fractional organ volumes are part of the prescription. Verify volume reporting. Medium • • Dose algorithms use during optimization may (or may not) be the same in different modes or modules, e.g.: • 3DCRT • IMRT • Post-IMRT evaluation. IMRT is still a maturing technology Normalized Volume • Plan Evaluation CT Small volume 1.75 MR CADPLAN V o lu m e D e lin e a tio n M o d a lity Big volume Large volume 12.25 126.75 Medium volume 5.25 Actual volume in cc Slice thickness in mm AcQSim CT 5 10 1 3 5 10 1 3 5 10 1 3 5 10 1.74 1.8 1.79 2.08 5.53 5.48 5.72 6.46 12.0 12.6 12.6 12.8 125.7 124.7 122.9 132.5 AcQSim MR Cadplan - 1.75 2.09 2.38 - 5.04 5.77 6.94 - 12.1 13.6 13.5 - 124.4 124.5 145.3 1.92 1.84 1.76 1.17 5.82 5.26 4.96 4.46 13.8 12.1 12.2 12.2 - 132.0 128.1 123.6 1 3 AAPM Summer School 2003 Van Dyk & Craig Effect of Slice Thickness Dose Volume Histograms • MRI scans performed for one patient using three slice thicknesses. • Volumes contoured for each of the scans. • An IMRT plan generated using 3mm slice thickness • Re-calculated for other image sets. Critical Structures in Head and Neck Treatment 100 100 3mm 6mm 12mm PTV 60 Nodal Bed 40 20 0 40 Spinal Cord 80 % Volume % Volume 80 45 50 55 60 Dose (Gy) 65 70 75 3mm 6mm 12mm 60 Calculated Measured 40 20 0 0 60 20 40 60 80 100 120 Dose (%) 40 Rt.Parotid 20 AAPM Summer School 2003 80 Volume (%) Target Structures for Head and Neck Treatment 100 0 Lt.Parotid 0 10 20 30 Dose (Gy) 40 50 60 AAPM Summer School 2003 8 Volume Analysis for DVH Polystyrene Cylinder Plan Evaluation Tools for IMRT Polystyrene Cube • 150 140 130 120 110 2 1.5 1 0.5 0 A B C A D Lucite Cube B C D Air Wedge 60 40 20 40 30 20 10 0 B C AAPM Summer School 2003 A D B C visualization Check for simultaneous DVH tally Behavior may depend on organ “priority” System may allow overlap, and attempt to resolve competing objectives in overlapping Region of structures overlap IMRT Commissioning: Optimization Manual Weights RT 25.0 AP 25.0 LT 25.0 PA 25.0 Preserve simple symmetry overcome contour variations deal with skin “flash” inhomogeneity corrections AAPM Summer School 2003 IMRT Commissioning: Optimization AAPM Summer School 2003 OAR AAPM Summer School 2003 Test inverse-planning algorithms Test MLC sequencing Can the system be driven to sensible solutions? • • • • PTV PTV D Maximum variation +42% to -44% Optimization • • • • • • 0 A Points belonging two structures: • Check if permitted • Check for simultaneous IMRT Weights RT 24.52 AP 25.48 LT 24.83 PA 25.16 AAPM Summer School 2003 Unopposed Fields AAPM Summer School 2003 9 (Van Esch et al, 2002). Contour Correction, Single Field Deliver uniform dose to a plane Dose Ratio P1 /P2 OPEN: 0.76 20: 0.68 TAR10 IMRT intensity P1 /P2 =0.73 Heterogeneities and Surfaces (a) (b) a) b) c) d) 20cm Different doses Equal dose Contour variation Heterogeneities 20cm 10cm P1 P2 (c) (d) target regions for uniform dose air or cork heterogeneities AAPM Summer School 2003 AAPM Summer School 2003 (Van Esch et al, 2002). Equal depth, Different doses Routine QA • • • • • AAPM Summer School 2003 Frequency of RTP QA AAPM Summer School 2003 Summary • Assess software upgrades and • • • • patches. Perform QA prior to release of any software Refer to AAPM TG53 and TG40 • When developing and validating • techniques, patient-specific QA is required. Learn what is necessary from patient-specific QA, and generalize. AAPM Summer School 2003 Monitor software and hardware modifications at all stages (CT, MR, CT-Sim, RTP, R&V, Linac). Verify patient geometry routinely Identify several cases of increasing complexity as “standards”. Capture base-line output during commissioning. Repeat plans “de-novo” at regular intervals, and post software modification. • • DICOM dose not guarantee problem-free image and structure transfer. Calculation and evaluation of dose-volume objectives depend on image slice thickness and resolution. Specify and monitor protocols to assure consistent image data from all sources. Dose calculations continue to evolve, with increasing accuracy, but may differ for regular and IMRT treatment planning. AAPM Summer School 2003 10 Summary • • • • It is difficult to “decouple” all components of IMRT planning software. IMRT is a technical evolution. The dependence of IMRT on segmentation requires adherence to clinical protocols Commission planning system, and validate each treatment planning procedure. AAPM Summer School 2003 11