Motion Management Strategies Gig S Mageras, PhD, FAAPM Peng Peng Zhang, PhD Memorial Sloan‐Kettering, New York aapm ss/mageras 1 Disclosure Support from: • NIH/NCI award R01 126993 • Varian Medical Systems aapm ss/mageras 2 Outline Topics in this talk: • Focus on intra‐fx motion (respiratory & prostate) • Effects of motion on RT • Motion mitigation in TP process • Studies of comparative motion management strategies Additional topics in chapter: • Types of motion management • Real‐time imaging: achievable accuracy • Residual uncertainties w current motion management aapm ss/mageras 3 EFFECTS OF MOTION ON RT aapm ss/mageras 4 Motion affects CT • Standard CT appears distorted: scan times long wrt respiratory motion time scales Chen SRO 2004: • Max object lengthening or shortening ~ motion extent NSCLC example “Free breathing” CT • Max displacement of object center ~ ½ motion extent Static Φ = π/4 Φ = 5π/4 Chen 2004 aapm ss/mageras 5 Motion affects CBCT • Motion during ~1 min acquisition image blurring, streaking, reduced tissue boundary visibility Respiration aapm ss/mageras Peristalsis (figure courtesy Michael Lovelock) 6 Motion during radiation delivery • “Blurs” dose, broadens dose gradients • Interplay effect with moving MLC: – Significant over single field/fx – Similar to physical compensator over multiple fields/fx planned profile **** delivered profile, peak-trough motion 10mm Breast IMRT (Chui 2003) Single 2Gy fx aapm ss/mageras Average over 30 fx 7 Motion affects on particle beams Proton treatment of liver: Equivalent path length changes (Lu 2007) Interplay with scanned particle beams (Bert 2008) aapm ss/mageras 8 Motion in prostate: Intra‐fx motion trajectory from EM tracking (Calypso) LAT Case 1 Case 3 SI AP Case 2 Case 4 (Data courtesy Dennis Mah) aapm ss/mageras 9 MOTION MITIGATION IN THE TREATMENT PLANNING PROCESS: RESPIRATORY SITES aapm ss/mageras 10 Respiration‐correlated CT: • Cine acquisition (Low 2003, Pan 2004) • Acquisition > 1 resp. cycle each couch position • Retrospective image sorting wrt resp. phase Abdominal displacement Pan et al 2004 • Helical acquisition (Ford 2003, Vedam 2003, Keall 2004, Fitzpatrick 2006) mskcc/gsm 11 Internal target volume • ICRU 62: – Internal Margin (IM): Variations in size, shape, and position of CTV relative to anatomic reference points; e.g., movements of respiration – Internal Target Volume (ITV): Volume encompassing the CTV and IM (ITV = CTV + IM) • Common approach: ITV encompasses GTVs in RCCT, then margin added for CTV • Commonly available s/w tools: • Loop through RCCT images while delineating ITV • Max intensity projection: each voxel set to max CT# for that voxel in RCCT set aapm ss/mageras (Underberg, 2005 ) 12 Example clinical process: Gated treatment of pancreas at MSKCC (1/3) • Fiducials: implanted markers or stent • Plan CT: BH helical CT + IV contrast • RCCT: cine scan + record respiration, 20 repeat img/position, phase sorted GE Discovery 8-row PET/CT, Varian RPM Slide courtesy Karyn Goodman aapm ss/mageras 13 Gated treatment of pancreas at MSKCC (2/3) • • • • Select gate from GTV motion, usually 30‐70% GTV/CTV delineation on plan CT ITV delineation: loop over in‐gate RCCT images Lung HFx: ITV includes all RCCT images 30 aapm ss/mageras 50 70 14 Gated treatment of pancreas at MSKCC (3/3) • Fiducials delineated in RCCT: – At gate onset DRR, compare with radiographs – In‐gate ITVfid DRR, compare with fluoro DRR aapm ss/mageras Radiograph 15 Uncertainties related to RCCT (1/3) • RCCT phase sorting: amplitude variations cause discontinuity artifacts (Rietzel 2005, 0% phase Lu 2006, Fitzpatrick 2006, Abdelnour 2007) aapm ss/mageras Artifacts in phase-sorted RCCT 16 Uncertainties related to RCCT (2/3) • Deviations in GTV volume/shape caused by RCCT artifacts (Persson 2010) – Variation in lung GTV larger than delineation error – Variation largest in bins adjacent to end inspiration 8 patients, 2 delineations/pt EE Bin 4 (EE) aapm ss/mageras 17 Uncertainties related to RCCT (3/3) • RCCT is a single sample: – Breathing pattern may vary (Purdie 2006, Britton 2007) – Tumor baseline variation (Sonke 2008) – Tumor size & shape changes (Kupelian 2005) Motion estimates require verification at treatment aapm ss/mageras 18 Uncertainties in ITV definition (1/2) • Max intensity projection from RCCT may underestimate lung tumor extent (left) • Alternatives: – Verification of GTV on each RCCT image (middle) – DIR + propagation of GTV contours (right) MIP MIP + GTV verification in RCCT Union of GTVs (Ezhil 2009) aapm ss/mageras 19 Uncertainties in ITV definition (2/2) • Gating window ITV can be underestimated in RCCT with irregular breathing – 2D dynamic MRI, lung ca, n=8 – “RedCAM” = simulated RCCT from dMRI – “dSGP” = moving phantom tumor using dMRI trajectories aapm ss/mageras (Cai 2010) 20 Motion‐encompassing ITV may overestimate internal margin Dosimetric analysis: Internal margin < implied by motion extent – Wolthaus IJROBP 70:1229, 2008 – Mutaf IJROBP 70:1561, 2008 aapm ss/mageras 21 Incorporating respiratory motion into dose calculation • Enables evaluation of dose to CTV & normal tissues • Makes use of RCCT + deformable image registration: – Dose calculation on each 3D image in RCCT set – DIR between each image & selected ref image – Dose grids warped to ref image & summed aapm ss/mageras 22 Studies of cumulative dose (1/3) (Admiraal, RO 86:55,2008) • Stage I NSCLC n=10, ITV from MIP, PTV=ITV, dose calc on avg CT • Small differences between PTV static dose & CTV cumulative dose Static dose Cum. dose aapm ss/mageras PTV CTV 23 Studies of cumulative dose (2/3) (Wu, MP 35:1440, 2008) • SBRT liver n=5, ITV=union of DIR‐prop. GTV, PTV=ITV • CTV D95/D99 <static PTV in 2/5 cases • Small motion effects on normal liver, kidneys 8 mm motion Static Cum. Dose difference, Pt 5 aapm ss/mageras 17 mm 24 Studies of cumulative dose (2/3) (Starkschall, IJROBP 5:1560, 2009) • Stage III NSCLC n=15, 0‐2cm, ITV = union of rigid‐propagated CTVs, PTV=ITV+(5‐to‐10 mm), model‐based DIR, 3DCRT/IMRT • 6/15 cases CTV D99 differs >3%, no cases >5% • Comparison cum. dose to ITV not assessed • Negligible effect on normal organs Small motion Large motion aapm ss/mageras 25 Uncertainties related to cumulative dose • DIR inaccuracies dose mapping errors – Self‐consistency errors – Absence of landmarks, RCCT artifacts – Tumor growth/regression TP dose distribution SD inconsistency in DIR vectors Dose SD error (Salguero MP 38:343, 2011) aapm ss/mageras 26 MOTION MITIGATION IN THE TREATMENT PLANNING PROCESS: PROSTATE aapm ss/mageras 27 Intra-fraction prostate motion modeling and dosimetric analysis (Hossain et al, 2008) • Motion tracking • Single fraction prostate SBRT, 50‐70 min • Record intrafraction motion via Cyberknife dual x‐ray system • Track fiducial markers • Motion modeling • Displace isocenter using marker trajectory • Recalculate dose • Investigate the dosimetric effects of intrafraction motion • Finding: • DVH shows clinically important changes when motion > 5mm aapm ss/mageras 28 Paradigm for dosimetric action levels in hypofractionated prostate VMAT (Zhang 2011) Hypofractionated prostate RT benefits from VMAT due to shortened Tx time Intra-fraction motion management: how to intervene? aapm ss/mageras 29 Motion in prostate: Intra‐fx motion trajectory from EM tracking (Calypso) LAT Case 1 Case 3 SI AP Case 2 Case 4 (Data courtesy Dennis Mah) aapm ss/mageras 30 Characteristics of motion trajectories Trajectories of EM transponder centroid in prostate aapm ss/mageras (Zhang 2011) 31 Cumulative dose from trajectories 180 D j Dij ( MLCi , D i , vo vi ) i 1 aapm ss/mageras (Zhang 2011) 32 Effect of intra‐fraction motion on Tx plan Static plan: CTVD95% = 95% Static plan: UDmax = 99.3% Static plan: RWD1cc = 86.5% aapm ss/mageras Static plan: BWD1cc = 98.8% (Zhang 2011) 33 Setting Action Thresholds Patient specific aapm ss/mageras (Zhang 2011) Generic 2mm 34 Sensitivity and specificity analysis Plan TP = dose violation aapm ss/mageras Plan TN = no violation (Zhang 2011) 35 STUDIES OF COMPARATIVE MOTION MANAGEMENT STRATEGIES aapm ss/mageras 36 Cumulative lung dose for different motion management strategies • NSCLC, T1N0‐T2N3, n=18 (Hugo 2009) • Compare – Breath hold at end inspiration,(BH) – target tracking with aperture,(TT) – mid‐ventilation aperture (MVA) ΔMLCTT 4%/5mm • RCCT, DIR & dose accumulation • Obs. differences in MLD ≤1Gy for excursions ≤13mm 13mm excursion aapm ss/mageras ΔMLCBH 5%/5mm 37 Target tracking vs RC inverse planning (Zhang 2008) • Lung ca, n=4, 10‐30mm excursion • Compare – Target tracking: static IMRT plan each RCCT – RC inverse plan: optimized using motion PDFs/cum. dose 29mm excursion, cumulative dose 70 Gy 68 66 49 Target tracking 4D inverse planning aapm ss/mageras 38 Summary Respiratory targets: • Motion affects imaging, radiation delivery • RCCT + motion encompassing ITV widely used • Uncertainties: – Irregular breathing introduces artifacts in RCCT, affects tumor size/shape, hence ITV – Max intensity projection may underestimate lung tumor motion extent – RCCT represents single sample but breathing is variable motion requires verification at tx aapm ss/mageras 39 Summary Respiratory targets, cont’d: • Cumulative dose: – CTV cumulative dose similar to static ITV dose for early stage lung, less clear for advanced stage – Small effect on normal tissue dose, relative to static case – Cum. dose calc cumbersome at present • When motion extent< ~15mm, motion mitigation in TP is effective, provided mean target position is controlled aapm ss/mageras 40 Summary Prostate: • Motion also affects imaging, dose delivery • Real‐time motion tracking (EM, radiographic) means of assessing effect on TP, motion mitigation strategies – Studies have used population motion data – Pt‐specific motion data feasible at simulation aapm ss/mageras 41 Analyze dose end points vs. motion CTV D95% SI mean SI max AP mean AP max LAT mean LAT max aapm ss/mageras ? CTV Dmin 6.5Gyx5 Rectum wall Dmax 7Gyx5 Rectum wall D1cc 7.5Gyx5 Bladder wall Dmax 8Gyx5 Bladder wall D1cc 8.5Gyx5 Urethra Dmax (Zhang 2011) 42 Comparisons of two intrafraction motion management strategies (Su et al, 2010) • Study design • 17 patients • Setup and intrafraction monitoring using Calypso • Intervention strategy I • Interrupt Tx whenever motion > action threshold=3mm • Margin required: LR=1.1mm, SI=1.8mm, and AP=2.3mm • Intervention strategy II • Automatically reposition pt at time intervals of 2 min • Margin required: LR=0.5mm, SI=1mm, and AP=1.5mm •Strategy II substantially increases #interruptions, and treatment time (factor of 8) aapm ss/mageras 43