1. ENVISION Activities Overview of the IRIS group John Barrio, Jorge Cabello, Pablo García, John Gillam, Gabriela Llosá, Paola Solevi, Carlos Lacasta, Josep Oliver, Magdalena Rafecas, Carles Solaz, Vera Stankova, Irene Torres Espallardo, Marco Trovato for the IRIS group Instituto de Fisica Corpuscular, IFIC (CSIC- UVEG), Valencia, Spain http://ific.uv.es/iris

2. PCI Workshop 23.05.13 I. Torres ENVISION Project TOF in-beam PET (crystal- and RPC-based) to reconstruct the positron distribution generated by the hadron beam to image the „prompt gamma“ generated in the first nanoseconds. European NoVel Imaging Systems for ION therapy on-line dose monitoring by developing novel imaging modalities related to dose deposition that allow to assess the treated volume during hadrontherapy Scintillators crystals (LaBr3) to build a Compton Camera Hadron beam γ γ γ TOF Patient Tumor

3. PCI Workshop 23.05.13 I. Torres WPs in ENVISION IRIS group is involved in: WP2: TOF in-beam PET (crystal- and RPC-based). WP3: detection of prompt radiation (in-beam SPAT: gamma, light charged particles and neutrons), overcoming the inherent sensitivity of ib-PET to metabolism.

4. PCI Workshop 23.05.13 I. Torres In-beam, in-room PET and Compton Camera for HT monitoring - Wash-out processes - Low positron yield, short half- life positron emitters - Partial ring geometry (*) (*)only for in-beam Time Of Flight (TOF ) Nowadays, PET is the only feasible and clinically used method. Challenges: Novel method of in-vivo dosimetry based on the detection of the prompt- gamma produced after the therapeutic beam  not affected by the wash-out processes Detection technique: Compton Camera

5. PCI Workshop 23.05.13 I. Torres PET for HT monitoring Crystal-based vs. Resistive-Plate-Chamber Gas detector Advantages: γ e-e- HV E Position readout Excellent timing resolution Inexpensive construction in large areas. Depth of interacion information (layered structure) Scintillating crystals and photon sensor. Standard technology for clinical PET. γ Scintillating crystal Photodetectors

6. PCI Workshop 23.05.13 I. Torres Simulations of the hadron beam with Geant4 Output: -Positron distribution produced within the PMMA phantom. -Prompt-gamma generated in the phantom. Proton beam (5∙10 7 ) E (MeV)R (cm) 15614.6 15814.9 16015.2 16215.5 16415.8 e + 30 minPrompt-γ 2,2∙10 6 1,55∙10 7 2,2∙10 6 1,58∙10 7 2,3∙10 6 1,61∙10 7 2,3∙10 6 1,64∙10 7 2,4∙10 6 1,66∙10 7 Carbon-ions beam (4∙10 6 ) E (MeV/A)R (cm)e + 30 minPrompt-γ 29814.68,6∙10 5 1,17∙10 7 301.514.98,7∙10 5 1,18∙10 7 305.215.28,8∙10 5 1,22∙10 7 308.915.59,0∙10 5 1,31∙10 7 312.515.89,1∙10 5 1,35∙10 7 Ideal hadron beam (no energy, no position uncertainty) Physics packages: QGSP_BIC_HP for hadronic model and standard for electromagetic model (L= 20 cm, D = 15 cm)

7. PCI Workshop 23.05.13 I. Torres Simulations of the detectors using GATE CRYSTAL-BASED PET (Gemini) 28 Heads 20 Heads16 Heads RPC PET Complete and 14-Heads RPC system 80 cm 20/60 modules in a radial sector TOF FWHM: 50, 100, 200 ps Spatial Resolution: σ x = 3.9 mm (module); σ y = 2 mm; σ z = 4 mm (Data provided by D.Watts & F.Sauli) Geometry courtesy of F. Diblen, UGhent Crystal: 4x4x22 mm 3, LYSO TOF resolution: 200, 400, 600 ps FWHM Axial FOV: 18 cm Diameter: 90 cm Energy window: 440-665 keV Stack composition 5 glass plates with 120x300x3.2 mm 3 8 mm pitch 4 gas volumes filled with Freon Axial FOV: 30 cm Diameter: 80 cm No energy available

8. PCI Workshop 23.05.13 I. Torres Image Reconstruction TOF-MLEM: Maximum-Likelihood Expectation-Maximization modified to include TOF information Ray tracing based on Siddon algorithm TOF: Gaussian distribution whose FWHM equal to the time resolution and mean equal to the expected position of the emission using δT. Image Quality Evaluation R50: z coordinate where the distal part of the activity profile drops at 50% of the maximum. E98: z coordinate where the integral of the activity curve reaches 98% of the total area R50 50% of I max E98 ROI: 20x20x200 mm centered on the beam

9. PCI Workshop 23.05.13 I. Torres Results for proton beams (I) CRYSTAL-PETRPC-PET

10. PCI Workshop 23.05.13 I. Torres Results for proton beams (II) Crystal-based (complete and partial ring system)

11. PCI Workshop 23.05.13 I. Torres Results for proton beams (III) RPC-based, complete system

12. PCI Workshop 23.05.13 I. Torres Results for proton beams (IV) RPC-based, partial ring

13. PCI Workshop 23.05.13 I. Torres Results for proton beams (V) R50 ΔR ref (mm) Full RingPartial Ring E (MeV) Crystal-based RPC-based

14. PCI Workshop 23.05.13 I. Torres Results for proton beams (VI) E98 ΔR ref (mm) Full RingPartial Ring E (MeV) Crystal-based RPC-based

15. PCI Workshop 23.05.13 I. Torres Results for carbon beams (I) CRYSTAL-PETRPC-PET

16. PCI Workshop 23.05.13 I. Torres Prompt-γ for HT monitoring The γ must have a trajectory along a cone surface, described by the axis β and the half angle θ The energy of the incident gamma E 0, and the location of the second event must be measured The resultant cone-surface has width implied by measurement uncertainty (and Doppler broadening) γ, E 0 β θ γ'γ' E 1, x 1 E 2, x 2 Event 1 Event 2 Estimated (it might be assumed that the second interaction is photoelectric), or Measured (the measurement of three or more photon interactions) Initial energy, E 0 Detectors Novel method of in-vivo dosimetry based on the detection of the prompt-gamma produced after the therapeutic beam  not affected by the wash-out processes Detection technique: Compton Camera

17. PCI Workshop 23.05.13 I. Torres Compton Camera Image Reconstruction MLEM algorithm adapted to use of 3-interaction events (measured initial energy) or 2-interaction events (estimating the initial energy, without assuming absorption). The 2-interaction case has higher efficiency. LaBr3 continuous crystals, with high Compton probability, high sensitivity and no absorber required. Scintillators are coupled to SiPM Gamma energies up to 20 MeV and above. 12 C at 4.4 MeV 16 O at 6.1 MeV

18. PCI Workshop 23.05.13 I. Torres CC - First Prototype 1 st detector 2 nd detector Preliminary images reconstructed from a Na 22 source at 10 cm at 2 different locations using Compton-SOPL First prototype developed with 2 layers ~ march 2012

19. PCI Workshop 23.05.13 I. Torres Results for proton beams 2-interactions3-interactions 2-interactions 3-interactions 156 158 160 162 164 COMPTON CAMERA (work-in-progress) Ep

20. PCI Workshop 23.05.13 I. Torres Results for carbon beams 2-interactions 3-interactions 2-interactions 3-interactions 298 302 305 309 313 Ec COMPTON CAMERA (work-in-progress)

21. PCI Workshop 23.05.13 I. Torres Effect of protons, neutrons and multiple scattering on the data Examples Clustering Neutron Proton

22. PCI Workshop 23.05.13 I. Torres Effect of neutrons and random events 2Cγ: Scoring of gamma track interactions with two or more than two interations, without neutrons 1Cγ: Scoring of gamma track interactions with one or more than one interations, without neutrons 2Cγ+n: Scoring of gamma and neutrons tracks interactions with two or more than two interations 1Cγ+n: Scoring of gamma and neutrons tracks interactions with one or more than one interations (10 8 proton; 3-interactions reconstruction)

23. PCI Workshop 23.05.13 I. Torres Reconstructed images for proton and carbon beams CRYSTAL-PET RPC-PET 2-interactions3-interactions Compton Camera 2-interactions3-interactions PROTONS (Ep = 160 MeV) CARBON IONS (Ec = 305∙A MeV) In order to proper investigate the feasibility of on-line monitoring hadrontherapy, it is necessary to build sources as close as possible to the clinical situation. This means: Including beam time structure in the simulations. Creating a Spread Out Bragg Peak (SOBP), to cover the lesion along the beam axis But we need more realistic sources…

24. PCI Workshop 23.05.13 I. Torres Beam Time Structure By default, the time stamp assigned to a process/particle is to respect its primary particle (proton/carbon ion from beam). It is possible to set a global time using the particleGun, an expression that emulates the time and the eventID. Hadron beams (despite coming from cyclotron or synchrotron) have a time structure which is the product of two time structures: Long time structure linked to the acceleration time and extraction scheme Short time structure linked to the HF for all accelerators. Example IBA Isochronus cyclotron: continuous extraction  HF runs at 100 MHz, there are pulses every 10 ns of 2 ns duration Pencil beam scanning mode: minimum time per spot (1 ms), maximum current (2 nA at 100 MeV, 4 nA at 230 MeV for protons). From PSI 1 spot every 30 ms.

25. PCI Workshop 23.05.13 I. Torres Time Structure: Pencil Beam Scanning (Secondary Particles) Proton beam (1.2∙10 7 particles/spot; 4∙10 8 part./s) Carbon ion beam (2∙10 6 particles/spot; 7∙10 7 part./s)

26. PCI Workshop 23.05.13 I. Torres Time Structure: Short component (Secondary Particles) Proton beam (1.2∙10 7 particles/spot; 4∙10 8 part./s) Carbon ion beam (2∙10 6 particles/spot; 7∙10 7 part./s)

27. PCI Workshop 23.05.13 I. Torres Spread Out Bragg Peak (I) A monoenergetic proton beam is not suitable for treatment  longitudinally narrow Bragg Peak (BP) Need to spread out the BP to treat the target along the beam: SOBP With these spots (1.2∙10 7 part.) and depending on the energy  absorbed dose is 0.2 Gy/spot for protons: Which intensity do we need?

28. PCI Workshop 23.05.13 I. Torres Spread Out Bragg Peak (II) Several authors have studied the production of a SOBP  focused on the analytical expresions: Bortfeld, 1996 Jette, 2010 How can SOBP be created?? Weighting monoenergetic proton beams

29. PCI Workshop 23.05.13 I. Torres Spread Out Bragg Peak (III) But all these works are focused on passive scanning and there are some approximations that are not suitable to the active scanning beam that we plan to simulate: Bortfeld´s weights are based on the Bragg-Kleeman rule: R = αE p p-beam Pencil Beam Broad Beam Integration volume

30. PCI Workshop 23.05.13 I. Torres Spread Out Bragg Peak (IV) How to proceed in our case: Number of protons α Dose at Maximum Initially the weighting factors were calculated considering that the tails of the dose distribution are negligible were considered  But depending on the given dose this approximation is not valid  count all the contributions: dose equation system When using this method, there are some details to take into account. Select: Maximum Energy Width of the SOBP Distance between energies (~number of beams) Dose integration volume Perform simulations with the same number of protons /carbon ions Calculate the weights from the dose contribution of each simulated beam using the above dose equations

31. PCI Workshop 23.05.13 I. Torres Spread Out Bragg Peak (V) Monoenergetic beam weights depend on dose integration volume, beam sampling  simulations calculated dose, number of protons Difficulties: SetIntensity command GATE Other parameters:

32. PCI Workshop 23.05.13 I. Torres Thanks for your attention!!! We are ready for a 3D dose distribution for XCAT-Phantom

33. PCI Workshop 23.05.13 I. Torres Conclusions Crystal-based PET with TOF information is able to distinguish range differences below 5 mm, even in the case of a partial ring geometry, for proton and carbon-ion beams. RPC-based PET needs to improve its sensitivity to detect range shifts below 5 mm for proton beams. Similar results as crystal-PET are observed for carbon beams and complete system, further studies are necessary to assess the behaviour of the partial ring. In Compton Imaging, 2-interaction events reconstruction shows better results than 3- interaction events (there are more events in the 2- interaction case). Ongoing work in improving the reconstruction algorithms.

34. PCI Workshop 23.05.13 I. Torres TOF RPC-based PET RPC stands for Resistive Plate Chamber Gas detector Advantages: Excellent timing resolution Inexpensive construction in large areas. Depth of interacion information (layered structure) Basics: Incoming photon ionizes the gas High voltage creates an avalanche of electrons Several layers gives good detection efficiency Thin layers gives better timing properties (M. Couceiro, Jornadas DFM, 2010) Sensitive Area (precise small gap 300 μm)

35. PCI Workshop 23.05.13 I. Torres Simulations of the hadron beam with Geant4 Output: -Positron distribution produced within the PMMA phantom. -Prompt-gamma generated in the phantom. Proton beam (5∙10 7 ) E (MeV)R (cm) 15614.6 15814.9 16015.2 16215.5 16415.8 e + 30 minPrompt-γ 221143815586824 22490371587544 228815216141888 232415616393624 235768516676552 Carbon-ions beam (4∙10 6 ) E (MeV/A)R (cm)e + 30 minPrompt-γ 29814.686140711684984 301.514.987026011853864 305.215.288217612254128 308.915.589662313067652 312.515.890527113472932 Ideal hadron beam (no energy, no position uncertainty) Physics packages: QGSP_BIC_HP for hadronic model and standard for electromagetic model (L= 20 cm, D = 15 cm)

36. PCI Workshop 23.05.13 I. Torres Results for carbon beams (I) CRYSTAL-PETRPC-PET

37. PCI Workshop 23.05.13 I. Torres G4 Simulations & Phys. Mod.