Tumor Tracking Current & Future Developments. Josh Evans, Ph.D. Virginia Commonwealth University Richmond, VA

Transcription

1 Tumor Tracking Current & Future Developments Josh Evans, Ph.D. Virginia Commonwealth University Richmond, VA

2 None. Disclosures

3 To understand: Learning Objectives the physiological characteristics of tumor motion in different treatment sites. the basics of existing technology for tumor tracking and their limitations. the background and concept of future technologies for tumor tracking.

4 Outline Tumor motion characteristics Lung, prostate, brain Requirements for successful tumor tracking Identify, anticipate, reposition, adapt dosimetry Current and Future Systems CyberKnife, Robotic couch, Vero, integrated MRI-Treatment

5 What organs move? Breathing Kidneys Lungs Pancreas Esophagus Liver Breast Peristalsis Gyn GU

6 Lung motion Periodic, but can be irregular Hysteresis Location of tumor in lung can be important P. J. Keall, G. S. Mageras, J. M. Balter, R. S. Emery, K. M. Forster, S. B. Jiang et al., The management of respiratory motion in radiation oncology report of AAPM Task Group 76, Med Phys 33 (10), (2006).

7 Breathing motion Surface monitoring illustrates baseline breathing amplitude changes over time P. J. Keall, G. S. Mageras, J. M. Balter, R. S. Emery, K. M. Forster, S. B. Jiang et al., The management of respiratory motion in radiation oncology report of AAPM Task Group 76, Med Phys 33 (10), (2006).

8 Prostate Calypso EM motion traces (9 10 mins) Prostate motion can be irregular and unpredictable P. Kupelian, T. Willoughby, A. Mahadevan, T. Djemil, G. Weinstein, S. Jani et al., Multi-institutional clinical experience with the Calypso System in localization and continuous, real-time monitoring of the prostate gland during external radiotherapy, Int J Radiat Oncol Biol Phys 67 (4), (2007).

9 Brain CyberKnife tracking experience of brain SRS patient Baseline drift of skull postion over 30 min Emphasizes need for treatment efficiency and intra-fraction position monintoring 8 M. J. Murphy, Tracking moving organs in real time, Semin Radiat Oncol 14 (1), (2004).

10 Motion management approaches Motion-encompassing methods Respiratory-gating Breath-hold Forced shallow breathing (compression) Respiration-synchronized techniques P. J. Keall, G. S. Mageras, J. M. Balter, R. S. Emery, K. M. Forster, S. B. Jiang et al., The management of respiratory motion in radiation oncology report of AAPM Task Group 76, Med Phys 33 (10), (2006).

11 Outline Tumor motion characteristics Lung, prostate, brain Requirements for successful tumor tracking Identify, anticipate, reposition, adapt dosimetry Current and Future Systems CyberKnife, Robotic couch, Vero, integrated MRI-Treatment

12 Successful tumor tracking 1. Identify tumor position in real time 2. Anticipate tumor motion (compensate for system lag) 3. Reposition beam in real time 4. Adapt dosimetry for changing target and critical structure locations.

13 Successful tumor tracking 1. How am I going to locate the target? 2. How far ahead do I need to know where the target is? 3. How am I re-aligning the beam & target? 4. Could this affect my target coverage or OAR sparing?

14 Successful tumor tracking 1. How am I going to locate the target? What are you locating? Visualize target directly Implanted fiducials Target position surrogates Imaging modality Ionizing: 2d, 3d, x-ray Non-ionizing: EM transponders, MRI, ultrasound Imaging frequency If ionizing; cumulative dose considerations

15 Calypso 1. How am I going to locate the target? EM transponder fiducial tracking FDA approved for prostate; under consideration for lung tracking Urology Nevada

16 Successful tumor tracking 2. How far ahead do I need to know where the target is? What is the total system lag time? Image acquisition, data processing, target/beam realignment Predictive motion models Residual uncertainty: add to target margin

17 Motion prediction 2. How far ahead do I need to know where the target is? System latency Prediction algorithm accuracy Most work done for breathing motion Krauss compare 4 motion prediction models for breathing motion RMSE prediction errors ~ 1 mm (0.2 s latency) to 2 mm (0.6 s latency) TG-76 recommends total system latency < 0.5 sec A. Krauss, S. Nill, and U. Oelfke, The comparative performance of four respiratory motion predictors for real-time tumour tracking, Phys Med Biol 56 (16), (2011).

18 Successful tumor tracking 3. How am I re-aligning the beam and target? Move the beam MLC tracking Gimbaled linac head Robot-mounted linac Move the patient Re-position couch

19 Dynamic MLC tracking 3. How am I re-aligning the beam & target? Move the MLC to align with the moving target Sawant 2008 Good tracking accuracy < 1 mm for in plane motion Lower efficiency for high freqency motion perpendicular to leaf direction. A. Sawant, et al., Management of three-dimensional intrafraction motion through real-time DMLC tracking, Med Phys 35 (5), (2008).

20 Successful tumor tracking 4. Could this affect my target coverage or OAR sparing? Will I miss my target? Am I moving my beam in to normal tissues? Are target and normal tissues moving rigidly or deformably? What s the dosimetric effect of not tracking? How much improvement with tracking?

21 4. Could this affect my target coverage or OAR sparing? Time-resolved dosimetry Poulsen 2012 TPS calculation of timeresolved dosimetry Validated in dynamic thorax phantom with film dosimetry (simple 1D motion) Limitations: Rigid shift of whole patient: no rotations, no deformations P. R. Poulsen, et al., A method of dose reconstruction for moving targets compatible with dynamic treatments, Med Phys 39 (10), (2012).

22 Dynamic Prostate Tracking Keall 2014 recent technical note on dynamic tracking for prostate patient 1. Calypso EM transponder fiducials used for target tracking 2. No prediction used since prostate motion typically slow and non-periodic 3. Dynamic MLC repositioning 4. Dose perturbation Fractional rectum V60Gy: + 5% (with tracking) + 30% (no tracking) P. J. Keall, E. Colvill, R. O'Brien, J. A. Ng, P. R. Poulsen, T. Eade et al., The first clinical implementation of electromagnetic transponder-guided MLC tracking, Med Phys 41 (2), (2014).

23 Very active research topic Relevant articles in June 2014 issue of Medical Physics: S. Yip, J. Rottman, and R. Berbeco, The impact of cine EPID image acquisition frame rate on markerless soft-tissue tracking, Medical Physics 41, (2014). Y. Ge, R. O'Brien, C. Shieh, J. T. Booth, and P. Keall, Toward the development of intrafraction tumor deformation tracking using a dynamic multi-leaf collimator, Medical Physics 41, (2014). B. Nelms, D. Opp, G. Zhang, E. Moros, and V. Feygelman, Motion as perturbation II. Development of the method for dosimetric analysis of motion effects with fixed-gantry IMRT, Medical Physics 41, (2014).

24 Outline Tumor motion characteristics Lung, prostate, brain Requirements for successful tumor tracking Identify, anticipate, reposition, adapt dosimetry Current and Future Systems CyberKnife, Robotic couch, Vero, integrated MRI-Treatment

25 CyberKnife First clinical system designed to detect and compensate for intrafx motion. Hybrid tumor monitoring Surface imaging for continuous breathing monitoring Fiducial based tracking of tumor on stereoscopic x-ray system Build correspondence model Move entire linac mounted on robot. Image from Wikipedia

26 Cyberknife lung tumor tracking Seppenwoolde 2007 Simulation to test the CyberKnife correspondence model 8 patients with external and internal lung tumor tracking data 95 th percentile of 3D errors ~ 2-4 mm Good discussion on model update frequency Y. Seppenwoolde, et al. Accuracy of tumor motion compensation algorithm from a robotic respiratory tracking system: a simulation study, Med Phys 34 (7), (2007).

27 Robotic couch re-positioning 3. How am I moving my beam? Move the patient to align with the statically positioned beam Translate and rotate around a virtual pivot point set to isocenter Images courtesy of Civco

28 3. How am I moving my beam? Move the patient to align with the statically positioned beam McNamara patient breathing traces: 2 10 mins each Program phantom: record motion with IR cameras Average over 3 breathing cycles to estimate baseline drift Lower frequency couch correction scheme: 10 sec delivery bins Overall mean RMSD from 4.9 mm (no correction) to 1.7 mm (couch correction) J. E. McNamara, et al., Toward correcting drift in target position during radiotherapy via computercontrolled couch adjustments on a programmable Linac, Med Phys 40 (5), (2013).

29 Vero Ring gantry for stability MV EPID imaging Orthogonal kv imaging Gimbaled MV linac with MLC Improve isocenter accuracy Tumor tracking Images courtesy of Vero

30 Vero tumor tracking 3. How am I moving my beam? Use gimbaled linac Depuydt 2011 Assess gimbals tracking with moving beam s light field and digital camera 2d Lego robot reproduce lung tumor trajectories System lag = 47.7 ms for both directions (total ~ 200 ms for real-time flouroscopic tracking) 2d tumor tracking errors: 0.54 mm ± 0.21mm T. Depuydt, et al., Geometric accuracy of a novel gimbals based radiation therapy tumor tracking system, Radiother Oncol 98 (3), (2011).

31 Depuydt 2013 Vero tumor tracking Simulate dynamic tumor tracking process to assess workflow with Vero (5 patients) Continuous surface imaging and fiducial based x-ray imaging Visicoil placed in tumor for tracking 3.2 mins to build correspondence model Average of ~ 9 mins from entering room to first beam on. Tracking errors of ~ 0.5 mm to 1.0 mm Image dose: average = 1.8 mgy/min for 1 Hz x-ray monitoring T. Depuydt, K. Poels, D. Verellen, B. Engels, C. Collen, C. Haverbeke et al., Initial assessment of tumor tracking with a gimbaled linac system in clinical circumstances: a patient simulation study, Radiother Oncol 106 (2), (2013).

32 Integrated MRI-MV treatment MRI Vs. x-ray: machines Improved soft tissue contrast No need to implant fiducials Non-ionizing No imaging frequency / imaging dose tradeoff Integrating MRI scanner with MV treatment machine is technically challenging Magnet strength: Tesla MV source: Co-60 Vs. Linac MRI-MV orientation: in-line, perpendicular, rotating

33 MRI-MV configurations Orientation of magnetic field and treatment beam has important implications Engineering challenges Linac output Patient dosimetry C. Kirkby, et al. Lung dosimetry in a linac-mri radiotherapy unit with a longitudinal magnetic field, Med Phys 37 (9), (2010). D. E. Constantin, et al. A study of the effect of in-line and perpendicular magnetic fields on beam characteristics of electron guns in medical linear accelerators, Med Phys 38 (7), (2011).

34 First clinically available: ViewRay Fixed cylindrical configuration 0.35 T super-conducting magnet 3x Co-60 sources Images courtesy of ViewRay

35 ViewRay imaging Can image either 1 4 frames / second OR 3 2 frames / second Images courtesy of ViewRay

36 Successful tumor tracking 1. Identify tumor position 2. Anticipate tumor motion (system lag) 3. Reposition beam 4. Adapt dosimetry

37 What was the first clinical system to employ realtime motion correction? 94% 1. CyberKnife 2. GammaKnife 3. Calypso 4. ViewRay 5. Vero 2% 3% 0% 0%

38 What was the first clinical system to employ realtime motion correction? 1. CyberKnife 2. GammaKnife 3. Calypso 4. ViewRay 5. Vero M. J. Murphy, Tracking moving organs in real time, Semin Radiat Oncol 14 (1), (2004). 9 A. Schweikard, G. Glosser, M. Bodduluri, M. J. Murphy, and J. R. Adler, Robotic motion compensation for respiratory movement during radiosurgery, Comput Aided Surg 5 (4), (2000). ANSWER = 1: ref = Murphy 2004 seminars in RO & Schwiekard 2000

39 The Vero system uses to compensate for tumor motion and has been shown in phantom studies to achieve a mean tracking accuracy of ~ mm: 5% 1. Dynamic couch movement / ~ 2 mm 2. Gimbaled linac motion / ~ 0.5 mm 3. Dynamic MLC motion / ~ 0.8 mm 4. Gimbaled linac motion / ~ 3 mm 5. Dynamic MLC motion / ~ 1 mm 77% 0% 14% 5%

40 The Vero system uses to compensate for tumor motion and has been shown in phantom studies to achieve a mean tracking accuracy of ~ mm: 1. Dynamic couch movement / ~ 2 mm 2. Gimbaled linac motion / ~ 0.5 mm 3. Dynamic MLC motion / ~ 0.8 mm 4. Gimbaled linac motion / ~ 3 mm 5. Dynamic MLC motion / ~ 1 mm ANSWER = 2: ref = Depuydt 2011 T. Depuydt, D. Verellen, O. Haas, T. Gevaert, N. Linthout, M. Duchateau et al., Geometric accuracy of a novel gimbals based radiation therapy tumor tracking system, Radiother Oncol 98 (3), (2011).

41 The ViewRay system has a configuration and can image a single plane up to times / second 10% 1. Parallel / 2 times per sec 2. Perpendicular / 8 times per sec 3. Oblique / 2 times per sec 4. Parallel / 4 times per sec 5. Perpendicular / 4 times per sec 11% 5% 18% 56%

42 The ViewRay system has a configuration and can image a single plane up to times / second 1. Parallel / 2 times per sec 2. Perpendicular / 8 times per sec 3. Oblique / 2 times per sec 4. Parallel / 4 times per sec 5. Perpendicular / 4 times per sec ANSWER = 5: ref = Dempsey 2005 J. Dempsey, D. Benoit, J. Fitzsimmons, A. Haghighat, J. Li, D. Low et al., A device for realtime 3D image-guided IMRT, Int J Radiat Oncol Biol Phys 63 (2), S202 (2005).

43 To minimize the impact of breathing irregularity on lung target prediction algorithms, TG-76 recommends that overall system lag for real-time tumor tracking should not exceed : 9% ms 88% ms 1% 2% 2% ms ms ms

44 To minimize the impact of breathing irregularity on lung target prediction algorithms, TG-76 recommends that overall system lag for real-time tumor tracking should not exceed : ms ms ms ms ms P. J. Keall, G. S. Mageras, J. M. Balter, R. S. Emery, K. M. Forster, S. B. Jiang et al., The management of respiratory motion in radiation oncology report of AAPM Task Group 76, Med Phys 33 (10), (2006). ANSWER = 2: ref = Keall 2006 (TG-76)

45 Tumor Tracking Current & Future Developments Josh Evans, Ph.D. Virginia Commonwealth University Richmond, VA