1. @MGH, Boston, October 19, 2007 PET protoytpe development and experimental results for "in vivo" hadrontherapy dosimetry Alberto Del Guerra Professor of Medical Physics Head and Director Specialty School in Medical Physics Head, Functional Imaging and Instrumentation Group Department of Physics “E.Fermi”, University of PisaFunctional Imaging and Instrumentation Group and INFN, Pisa (Italy) Department of Physics “E.Fermi” University of Pisa INFN - Pisa Functional imaging and Instrumentation Group – Univ. Pisa

2. Proton induced activity and dose delivery Therapeutic proton beams produce in biological tissue short-lived β + -emitters (mainly 11 C and 15 O) by means of target nuclei fragmentations; Fragmentation of the target: – 16 O (p,n) 15 O 12 C (p,n) 11 C τ 15-O =121.8 s τ 11-C =1222.8 s Nuclear cross sections fall off at low energy just few millimeters before Bragg peak [15-20 MeV threshold ]; By finding the distribution of positron annihilation points it would be possible to extract in vivo information about dose localization (unfolding). W ARNING : V ERY L OW T RACER C ONCENTRATIONS!

3. How to gather information on proton range from reconstructed activity? FILTERING APPROACH TO UNFOLDING Parodi & Bortfeld * proposed a fast method based on the description of the PET image as a convolution of the dose distribution with a filter function, also showing the uniqueness of the filter for proton energy ranging from 70 to 220 MeV. Reconstructing expected PET signal from local planned dose distribution and comparing it with measured one it is possible to verify if dose has been delivered correctly. AN INVERSION OF THE FILTER WOULD ALLOW DIRECT DOSE INFORMATION We have studied the possibility to adopt the same approach in the energy range of interest for CATANA (40–70 MeV), as a starting point for the inverse process. K Parodi and T.Bortfeld. A filtering approach based on Gaussian-power law convolutions for local PET verification of proton radiotherapy (Phys. Med. Biol. 51 (2006) 1991 - 2009).

4. Theoretical research of filter functions Predicted 3D depth dose and activity profiles were obtained by means of a code based on a semi-analytical model of the Ziegler-Vavilov theory: –Fokker-Planck scheme for evolving energy distribution is combined with nuclear cross sections interpolated from EXFOR. Monoenergetic Bragg peaks and activity profiles have been simulated for several initial energies; Two isotope distribution filters have been computed and weighted according to PET acquisition time in order to give the total activity filter. T HE U NIQUENESS OF THE F ILTER F UNCTIONS (F and F -1 ) E NSURES THAT THE M ETHOD C AN B E A PPLIED TO E XTENDED D OSE D ISTRIBUTIONS

5. Reconstructed activity profile in comparison with the filtered dose (SOBP of 20 mm modulation); T Irr : [0, 173] s; T Scan : [173, 2000] s; Transverse domain of integration: 19 mm Ø; Reconstructed activity profile in comparison with the filtered dose (SOBP of 12 mm modulation); T Irr : [0, 189] s; T Scan : [189, 2000] s; Transverse domain of integration: 19 mm Ø; Testing the theoretical filters on experimental data Reconstructed activity profile in comparison with the filtered dose (monoenergetic Bragg curve); T Irr : [0, 138] s; T Scan : [138, 1938] s; Transverse domain of integration: 19 mm Ø;

6. Separation of Isotope Contributions Individual main isotope components ( 11 C and 15 O) has been separated by time analysis on the true count rate from measured list mode file; It has been thus possible to quantify quite accurately the relative abundance and the spatial distribution of each isotope specie produced at the end of irradiation; This way, in spite of the low statistics, we obtained at once both a C ROSS S ECTION V ALIDATION and a more cogent T EST FOR F ILTERING A PPROACH suitability. 12 mm SOBP isotope productionFE isotope production

7. β + - emitter fragments with Protons and with Carbon ions: –Proton as a projectile: Fragmentation of the target: – 16 O (p,n) 15 O 12 C (p,n) 11 C τ 15-O =121.8 s τ 11-C =1222.8 s [15-20 MeV threshold for p-induced nuclear reactions that cause poor spatial correlation between β + -activity and dose depth profile] –Carbon as a projectile; Fragmentation of the target: –X ( 12 C, 11 C+n) X X ( 12 C, 10 C+2n) X τ 11-C =1222.8 s τ 10-C =19.3 s Fragmentation of the projectile: – 16 O ( 12 C, X) 15 O+n 12 C ( 12 C, X) 11 C +n τ 15-O =121.8 s τ 11-C =1222.8 s Hadron-driven PET with Ions

8. The PET-tomograph prototype Two planar heads, each with an active area of 45 mm x 45 mm Distance between the heads: 5+5 cm, 7+7 cm, 10+10 cm (for phantoms study) and 5+15 cm (for eye monitoring) LYSO crystal matrix, 21 x 21 pixels 2.152 mm x 2.152 mm each (Hilger Crystals) Crystal thickness: 18 mm 64-anode PMT (Hamamatsu) “multiplexed” read-out electronics: 64-inputs/4 outputs

9. A cylindrical PMMA phantom ( 7cm diameter, 7cm length) was irradiated with 3 mono- energetic 12 C beams ( 108.53 112.60 116.57 AMeV). A square section beam of 28mm in side was adopted and a total dose of 60Gy was delivered for each energy. The acquisition time was set at ~30 minutes for both PET systems. The experimental configuration of the two PET systems is reported on the left.

10. The linear profiles from the central slice of the Dopet reconstructed activity at 3 12 C energies are reported below on the left. On the right, for comparison, the activities acquired, at 116.57 AMeV, with Dopet( ) and Bastei (---).

11. Y B1 SCD + 2D chain Scheme + sum SCD + 2D chain Scheme + sum Y B2 Y B3 Y B4 Y A1 Y A2 Y A3 Y A4 X B1 X B2 X B3 X B4 X A1 X A2 X A3 X A4 X A2 X A1 X A4 X A3  X B2 X B1 X B4 X B3  Y A2 Y A1 Y A4 Y A3  Y B2 Y B1 Y B4 Y B3  XAXA XBXB YAYA YBYB Resistors in the 2D chain are weighted according to the formula: R NA = R 0 /N; R NB =R 0 /(16-N+1) Where N ranges between 1 and 16

12. STATUS OF MAROC- SPIROC measurement Michel Bouchel, Stéphane Callier, Frédéric Dulucq, Julien Fleury, Gisèle Martin-Chassard, Christophe de La Taille, Ludovic Raux

13. MAROC : 64 ch MAPMT chip for ATLAS lumi Complete front-end chip for 64 channels multi-anode photomultipliers –Auto-trigger on 1/3 p.e. at 10 MHz, 12 bit charge output –SiGe 0.35 µm, 12 mm2, Pd = 350mW PMF Hold signal Photomultiplier 64 channels Photons Variabl e Gain Preamp. Variable Slow Shaper 20-100 ns S&H Bipolar Fast Shaper Unipolar Fast Shaper Gain correction 64*6bits 3 discri thresholds (3*12 bits) Multiplexed Analog charge output LUCID S&H 3 DACs 12 bits 80 MHz encoder 64 Wilkinson 12 bit ADC 64 trigger outputs (to FPGA) Multiplexed Digital charge output 64 inputs

14. Active board pictures MAROC sideLattice side 64 ch PMTMAROC2 chip bounded at CERN