1. 1. M D AndersonCancer Center Orlando TomoTherapys Implementation of Image-guided Adaptive RadiationTherapyKatja Langen, PhDResearch supported by TomoTherapy Inc.
2. 2. Todays Lecture Introduction to helical tomotherapy Image guidance with TomoTherapy Adaptive radiation therapy with TomoTherapy
3. 3. Introduction to helicalTomotherapy
4. 4. TomoTherapy Helical tomotherapy process to deliver IMRT Same hardware can be used to image (MVCT) patient (IGRT)
5. 5. TomoTherapyFan beam, continuously Patient rotates Tumor Healthy Organ
6. 6. TomoTherapy Divides beam into 64 beamletseach beamlet can be on/off
7. 7. TomoTherapyFan beam, continuously Patient rotates Tumor Healthy Organ
8. 8. TomoTherapyFan beam, continuously Patient rotatesTumor
9. 9. TomoTherapyFan beam, continuously Patient rotates Tumor Healthy Organ
10. 10. Fan Beamlateral longitudinal400 mm 10, 25, or 50 mmlaterallongitudinal
11. 11. Helical Tomotherapy ?from: W. Kalender, Computed Tomography: Fundamentals,System Technology, Image Quality, Applications
12. 12. Delivery- 51 projections per rotation: 360 /51=7 -Temporal modulationleaf opening timeduring projection
13. 13. Head + NeckDelivery Instructions(Sinograms)PrProstate -Typical gantry rotation:o 15-20 secondsject -Typical treatment time:i Prostate: 5 Mino H+N: 7 Minns64 Leafs
14. 14. The TomoTherapy unit at M D Anderson Orlando
15. 15. The TomoTherapy unit at M D Anderson Orlando
16. 16. Megavoltage CT images6 MV Accelerator(tuned to 3.5 MV for MVCT) Primary Collimator( 4 mm Slice Width) Binary MLC85cm (all leafs open during MVCT)User can adjust:TFOV-Scan rangeC cm40-Pitch Approximately 50cm Courtesy of: TomoTherapy, Inc .
17. 17. Patient Images: ProstateDiagnostic (kV) CT Tomotherapy (MV) CT
18. 18. Patient Images: Head and Neck Diagnostic (kV) CT Tomotherapy (MV) CT
19. 19. Image guidance with the TomoTherapy system
20. 20. TerminologyImage-guided Radiation Therapy (IGRT):Use images to position patient -Portal images, Ultrasound, CT-basedImage-guided Radiation Therapy (IGRT) with TomoTherapy:Megavoltage CT (MVCT) based
21. 21. Image (MVCT) Quality ?
22. 22. Helical Tomotherapy: MVCT Imaging Performance Characterization AAPM CT Phantom [Cardinal Health, Hicksville, NY]. Noise Uniformity Spatial Resolution Dosimetry Multiple Scan Average Dose (MSAD) in 20 cm diameter acrylic phantom Meeks et al., Med Phys, 32, pp 2673
23. 23. Noise Expressed as % of water Mean ROIROI StandardMatrixPitch Noise"CT #"Deviation1.0 102425.4 2.7 2561.6 1026 an worse th 25.4 2.7 -2.4 1021lues ch va25.3 2.7 t of pi1.0 ependent102835.4 3.7nd 512 ix i tr 1.6 102735.5 3.7co n ma with re nners2.4 102936.2 3.8 Increases CT scacdiag n o s ti 1.0 102543.2 4.5 7681.6 102543.8 4.62.4 102643.3 4.6 AcQSim1.0 1005 181.9 512Meeks et al., Med Phys, 32, pp 2673
24. 24. UniformityMaximum Peripheral to Central Deviation MatrixPitch Maximum Uniformity Mean Mean PeripheralIndexCenter Peripheralrs Variation (%)"CT" "CT"T sc anne 256 1.0 0.9999.5 1015 s ti c 1025 Coiagn10271019 d 1.6 0.97 99.6 a ble to 2.4 1.03 99.5 compar 10131022 v a lues 512 1.0 1.00matrix 99.5 1020 1029econ 1.6 h/r 0.9099.6 1019 1028f pitc ndent o 2.4 1.0499.5 1021 1031 I ndepe 768 1.0 0.9499.5 1019 1027 1.6 0.8099.6 1019 1026 2.4 0.8999.6 1019 1027 AcQSim1.0 0.5599.6 1006 1005 512 Meeks et al., Med Phys, 32, pp 2673
25. 25. Spatial Resolution: Qualitative Determination256x256 matrix 512x512 matrixNo holes resolved Visible resolution ~1.5 mm Visible resolution > 1.75 mm 768x768 matrixVisible resolution ~1.25 mmMeeks et al.,Med Phys, 32, pp 2673
26. 26. Meeks et al., Contrast DetailMed Phys, 32,pp 2673 181625.4 mm 14 19.1 mm12 Diameter (mm)12.7 mm 109.5 mm86.4 mm 3.2 mm 6 Visible5% difference 42 Not Visible 00 10 20304050 60 70 80 90 100 110% Electron Density Difference
27. 27. Patient Images
28. 28. MVCT Dose1.10 Multiple Scan Average Dose in 20-cm diameter phantom 1.00 Fine scan ~ 1 cGy0.90 0.80 0.70 MSAD (cGy) Measured Values Norm to Pitch 10.60 0.50Coarse scans < 0.5 cGy0.40 0.30 0.20 0.10 0.511.52 2.53 3.5 44.5 PitchMeeks et al., Med Phys, 32, pp 2673
29. 29. Summary Uniformity is comparable to diagnostic CT scanners while noise is worse Spatial resolution using 512x512 matrix is ~1.5 mm Dose: ~1 cGy for fine setting (pitch of 1) ; dose decreases with looser pitch Image Quality is acceptable for patient alignment, and is also acceptable for delineation of many soft tissue structures. Meeks et al., Med Phys, 32, pp 2673
30. 30. Average In-Room Times 5 minutes for setupIGRT 3-5 minutes MVCT scan 5 minutes reconstruction and registering MVCT to KVCT and shift patient 5-10 minutes beam on time 5 minutes assist patient out
31. 31. Prostate IGRT: Helical Transverse, not aligned Tomotherapy
32. 32. Prostate IGRT: Helical Transverse, aligned Tomotherapy
33. 33. Prostate IGRT: Helical Sagittal, aligned Tomotherapy
34. 34. Image Registration for Prostate cases ?Fiducial marker basedAnatomy basedContour based
35. 35. Prostate alignment study RT: Marker, Anatomy, Contour MD: Marker, Anatomy, Contour 112 image pairs3 patientsLangen et al., IJROBP, 62, pp 1517
36. 36. RT vs MD registration Difference 5 mm MarkerAnatomy Contour A/P: 1 % A/P: 5 % A/P: 17 % S/I: 2 % S/I: 10 %S/I: 31 % Lat.: 1 %Lat.: 0 %Lat.: 3 %Langen et al., IJROBP, 62, pp 1517
37. 37. RT vs MD registration Difference 3 mmMarkerAnatomy Contour A/P: 2 % A/P: 15 % A/P: 48 % S/I: 12 %S/I: 32 % S/I: 52 % Lat.: 2 %Lat.: 3 % Lat.: 24 % Langen et al., IJROBP, 62, pp 1517
38. 38. Conclusions of alignment study Implanted marker based alignment have least inter-user variabilityAnatomy-based alignments outperformcontour-based alignments Langen et al., IJROBP, 62, pp 1517
39. 39. Head and Neck alignment: bony anatomyMVCTSpinal CordBase of skull not alignedaligned
40. 40. IGRT and ART, do we need both ?
41. 41. Daily-IGRT, why ART ?Deformation in the Pelvis Same patient, different daysDosimetric consequence ?
42. 42. Progressive deformation Barker (2004): 14 H+N patients, 13-20 repeat CT, in-room CTObservation of tumor regression Barker et al. , IJROBP, 59, pp 960, 2004
43. 43. Progressive deformation Barker (2004): 14 H+N patients, 13-20 repeat CT, in-room CTObservation of parotid shrinkage and migration Barker et al. , IJROBP, 59, pp 960, 2004
44. 44. kVCT vs. end-of-treatment MVCT Dosimetric consequence ? kVCT: PTV SC
45. 45. Adaptive Radiation Therapy with the TomoTherapy system
46. 46. TerminologyAdaptive Radiation Therapy (ART): Use information from images to change subsequenttreatments -Change margin based on observed setup/organmotion-Use images to evaluate dosimetry (dose-guided, dosecompensation)Adaptive Radiation Therapy (ART) with TomoTherapy:Using MVCT images to evaluate dosimetry on daily basis
47. 47. Adaptive Radiation Therapy: need accurate HU numbersAcquire MVCT Recalculate dose distribution on MVCT Add dose distribution to calculate cumulative dose Compare with plan need deformableAdapt plan image registration
48. 48. Accuracy of MVCT numbers
49. 49. Need MVCT to electron density calibrationkVCTMVCT RMI, Model 467, CT to electron densityphantom
50. 50. MVCT to electron density calibration curves kVCTMVCT
51. 51. MVCT Integrity ?Does the calibration change with time ?Does calibration change with phantom arrangement ? Does the calibration change with MVCT pitch ?
52. 52. MVCT Integrity ? Does the calibration change with time ?
53. 53. MVCT Integrity ? Does the calibration change with phantom arrangement ?
54. 54. MVCT Integrity ? Does the calibration change with phantom arrangement ?
55. 55. MVCT Integrity ? Does the calibration change with MVCT pitch ?
56. 56. Dose re-computation end-to-end testsRigid phantom kVCT scan Fictitious target Generate treatment planMVCT scan phantom Re-calculate dose in MVCT Compare plan DVH with recalculated DVH Expect agreement !!!
57. 57. Dose re-computation end-to-end tests
58. 58. Dose re-computation end-to-end tests all target D95are within 0.5 %
59. 59. Dose re-computation end-to-end tests Deformed anatomy ? Plan 2 Plan 2, Plan 2, Deformation 1 Deformation 2Point of ion chamber measurement
60. 60. Dose re-computation end-to-end testsMeas. Diff (%) Re-calc. Diff (%)Plan 1, Def 1 + 2.7+2.9Plan 1, Def 2 -10.5-9.3Plan 2, Def 1 +3.3 +3.2Plan 1, Def 2 -11.5-10.8
61. 61. Conclusions- MVCT numbers are reproducible MVCT to electron density calibration is reliable- Phantom end-to-end test results are typically within 1 % of plan results MVCT images can be used for reliable dose computations Langen et al.: Use of mega voltage CT (MVCT) images fordose computations. PMB, 50, pp 4259
62. 62. Daily-IGRT, why ART ?Deformation in the Pelvis Same patient, different daysDosimetric consequence ?
63. 63. Evaluate dosimetric consequence: Dose Re-calculation Use pre-treatment MVCTRecalculate plan based on planned MLC patternRecalculate Dose on MVCT image
64. 64. Deformation in the Pelvis Evaluate dosimetric consequenceObtain MVCT MVCT contours Dose recalculation
65. 65. Deformation in the PelvisPlan DVH39 true DVHs
66. 66. Deformation in the PelvisPlan DVH 39 true DVHs 1) Requires manual contouring 2) Need deformable image registration tocalculated the cumulative DVH
67. 67. Deformable image registrationRelate voxel location x to its location in the second image (at a later time t) using a displacement vector u(x,t)x(x,t)=x+u(x,t)Algorithm minimizes u and differences in HU units Lu et al. , Phys Med Biol, 49, pp 3067, 2004
68. 68. MVCT to kVCTdeformable image registration Algorithm requires same HU for same voxel in MVCT and kVCTMVCT are noisier => apply edge-preserving smoothing=> MVCT numbers kVCT numbers intensity histogram calibration Lu et al. , Phys Med Biol, submitted, 2006
69. 69. MVCT to kVCTdeformable image registration Algorithm maps each MVCT voxel to kVCTAlgorithm can map each kVCT voxel to MVCT can be used for automatic contouring of MVCT automatically generated MVCT contours can beused to visualize deformation map Lu et al. , Phys Med Biol, submitted, 2006
70. 70. MVCT to kVCT deformable image registrationLu et al. , Phys Med Biol, submitted, 2006
71. 71. MVCT to kVCT deformable image registration kVCT: PTV SC MVCT:PTVSC
72. 72. Accuracy of automatic H+N MVCT contours ? Spinal Cord: Automatic vs. manual contours, Compare dose-based end points: Dmax, Dmean1 DmaxDmean1.4Dose (Gy)Dose (Gy) 0.8 1.2Automatic ManualAutomatic Manual 1 0.60 510 15 2025 30 35 0 5 10 1520 25 30 35Teatment fraction Treatment fraction 3 patients: Dmax ratio: 1.1 3.5 %Dmean ratio: 0.1 2.5 %Moderated Poster: AAPM 06, Langen et al.
73. 73. Accuracy of automatic H+N MVCTcontours ? Parotids: Visual inspection of automatic contourskVCT contours:MVCT contours:
74. 74. Accuracy of automatic H+N MVCTcontours ? Parotids: Visual inspection of automatic contours - 6 patients, 150 MVCT scans, 2 physicianConclusion: The parotid contours generated by the deformable algorithm correlate with the location of these structures on the daily MVCT images. Submitted to ASTRO 06, Manon et al.
75. 75. Automatic MVCT contouring kVCTTx 1Tx 7 Tx 15kVCT contours:MVCT contours:
76. 76. Automatic MVCT contouring kVCT Tx 1Tx 7 Tx 15Dosimetric consequence ?
77. 77. Dosimetric consequence Tx 7DVH 1.4 Gy
78. 78. Parotid dose during treatment course Numerical Analysis:-6 patients, 150 MVCT scansConclusion: At end of treatment: mean parotid dose was on average 7.5 Gy higher than planned Range:1.6-16.4 Gy Submitted to ASTRO 06, Manon et al.
79. 79. Clinical example 73 year old Head and Neck patient: PTV prescription: 70 Gy in 35 fractions
80. 80. Clinical example First 17 fractions:
81. 81. Plan vs. deformable registration DVH after 17 fractions Rt. Lt. Parotid Parotid 15 Gy 17 Gy
82. 82. Generate adaptive plan-Obtain new kVCT scan of patient-How to transfer dosimetry information to new kVCT ?
83. 83. Generate adaptive plan-Use kVCT-kVCT deformable registration to transfer original structures to second kVCT-Physician reviewed structures-Generate new plan Limit dose to over-dosed parotid region
84. 84. Rt. Parotid DVH (after 17 Txs) Medium doseHigh dose10-20 Gy > 20 Gy Low dose 0-10 GyDose difference mostly in10-20 Gy region
85. 85. Parotid DVH Generate contour that corresponds to 10-20 Gy region
86. 86. Generate second plan First PlanTARGETSecond Plan PAROTID
87. 87. Patient completes treatmentDaily scans Daily deformable registrationFULL COURSE
88. 88. WHAT IF SCENARIOS.compare -Original Plan P1(35) -Plan 1 delivered for 17 fx + Plan 2 for 18 fxPlanned to deliver with adaptive RTD1(17)+P2(18)-Actual delivery (Plan 1-17 fx and Plan 2-18 fx)D1(17)+D2(18)-without adaptive RT, Plan 1 only D1(35)
89. 89. What if .. (Rt. Parotid DVH)100 90 80 P1(35)70 60 50 40 30 20 100 0 10 2030 40 5060 70 Dose (Gy)
90. 90. What if .. (Rt. Parotid DVH)100 90P1(35)80 70D1(17)+P2(18)60 50 40 30 20 100 0 10 2030 4050 6070 Dose (Gy)
91. 91. What if .. (Rt. Parotid DVH)100 90P1(35) 80D1(17)+P2(18)70 60D1(17)+D2(18) 50 40 30 20 100 0 10 2030 4050 6070 Dose (Gy)
92. 92. What if .. (Rt. Parotid DVH)100 90P1(35) 80D1(17)+P2(18) 70D1(17)+D2(18) 60D1(35)50 40 30 20 100 0 10 2030 4050 6070 Dose (Gy)
93. 93. Conclusion Opportunities Challenges - Assess consequence- When to adapt plan ? of deformation - How often ? - Correct differences between delivered and- Verify accuracy of planned dose in adaptive deformable registration Plan - Shrinking volumes ?
94. 94. Thanks ! MD Anderson:TomoTherapy: Sanford Meeks Gustavo OliveraTom WagnerJason HaimerlTwyla WilloughbyKen RuchalaOmar Zeidan Weigu LuAmish Shah Sam JeswaniArati LimayeQuan ChenPatrick KupelianEric SchnarrRafael MaonDaniel BuchholzAlan ForbesWayne Jenkins