Fast readout of GEM detectors for medical imaging

1. FRONTIER DETECTORS FOR FRONTIER PHYSICS 12th Pisa Meeting on Advanced Detectors May 20 - 26, 2012 La Biodola, Isola d'Elba (Italy) Fast readout of GEM detectors for medical imaging M. Bucciantonio, U. Amaldi, R. Kieffer, N. Malakhov,F. Sauli, D. Watts

2. Hadrontherapy(I) X-ray gamma neutron Relative dose [%] proton carbon Depth in water [cm] Loco-regionaltreatmentsfail for 18% of allcancerpatients: notremovable by surgery, proximity to vitalorgans… More targeted and effectivecancertreatments: protons and Carbon ions 10 millionhabitants (20000pts/ year) Hoffmann, PPARC 2005 - lower tissue damage - biological effect ~10% > Xray - less energy deposited in healthy tissues - better conformity in dose distribution 12% of Xraypatients (2400pts/year) Hoffmann, PPARC 2005 Higher tumor control probability Martina Bucciantonio - TERA Foundation

3. Hadrontherapy(II) • Delivery system • Beam modelling • CT units and range • Instrumentation positioning • Patient positioning • Target delineation • Organ motion • Patient anatomy, motion, repositioning Dose delivery uncertainties Precise positioning of the Braggpeak Spatial uncertainties Quality Assurance plays a key role for Treatment planning Patient positioning Treatment planning verification Therapy modifications Medical Imaging devices R&D Treatment risk: severe underdosage of the tumor or to an overdosage outside the target volume due to any uncertainty in the delivered dose distribution Martina Bucciantonio - TERA Foundation

4. AQUA R&Ds at TERA Foundation GEM based detectors Envision WP3 TOF CNAO, PSI, MGH, AGH Un. Envision WP2 Martina Bucciantonio - TERA Foundation

5. Proton Range Radiography (I) Principle Energy loss proportional to the integrated relative electron density of the target Range by direct measurement of the electron density which resolution is for L= 25 cm and N~100counts/pixel ,p parameters of the material residual range (~0.6%) straggling (~1.3%) beam momentum (~0.4%) (suitable for a good medical imaging system) • How • mono-energetic beam of energy above total absorption in the target • correlated measurement of track position and residual energy • 2D integrated density image Martina Bucciantonio - TERA Foundation

6. Proton Range Radiography (II) Purposes • Optimal patient positioning (low dose radiography) • Treatment planning verification • Real size proton radiography (first step toward a complete protonCT) Realization • First Proton Range Radiograpy prototype – PRR10 (2010) CT Scan PRR Simulation G. Chen, MGH Boston 10cm w.e. 10cm 10cm Range/Energy loss 10 cm water equivalent Tracker 10x10cm2 active area Martina Bucciantonio - TERA Foundation

7. Proton Range Radiography (III) PRR10 Test beam PSI CNAO Lung (.20) Trabecular bone (1.16) Breast 50/50 (.99) U. Amaldi et al, Nucl. Instr. and Meth. A629(2011)337 1 mm Ø Major limitations: - slow readout ~ 10 kHz - small size radiography (10x10 cm2) For 1x1 mm2 pixels and an image size of 30x30 cm2 (105pixels) ∼107 proton tracks to be recorded (achievable in 10 seconds with 1 MHz readout rate) Present R&D Larger area (30x30 cm2) 48 scintillators (~ 15 cm tissue equivalent) Faster readout electronics~ 1 MHz PRR30 in construction Martina Bucciantonio - TERA Foundation

8. PRR30 In collaboration with:AGH University, PSI • GEMs: Tracker • - 30x30 cm2 Gas Electron Multiplier (GEM) • - Gas: Ar-CO2 (70:30) • - HV: 3.9-4.2 kV • 2D strip readout • High rate (~1 MHz) • Radiation resistant Scintillators stack: Range/Energy loss • - Plastic scintillators (3mm): • - 2 modules in coincidence for trigger • - 48 modules (~15 cm water equivalent) • - density ~1 (almost tissue equivalent) • - Wavelength shifter fibers • Silicon Photomultipliers Readout • High intrinsic rate capability (>1 MHz) • 30 MeV < RESIDUAL RANGE< 190 MeV Martina Bucciantonio - TERA Foundation

9. GEM readout • Readout board 256 strips (x,y directions) at 800 μm pitch • 50 μm wide (top layer) • 340 μm (bottom layer) • Complete decoupling of the amplification stage and the readout electrode (charge collector) • projective two-dimensional analog readout almost equal charge sharing Electronics requirements for Proton Radiography Input signals - short current pulses with duration time of 40 ns Expected particle flux on GEM 30x30cm2 106 s−1 Input charge from 0 to 500 fC (50 fC most probable) Minimum discrimination threshold 6 fC input equivalent Self-triggering Required time resolution better than 100 ns p-p Novel dedicated ASIC for GEM chambers GEMROC Hybrid Front End board developed by AGH Cracow University in collaboration with TERA Martina Bucciantonio - TERA Foundation

10. GEM 30x30 readout 12 GEMROC front-end boards 6 GR_DAQ GR_DAQ SDAQ Imb LabView control 1 Master unit Imb-SDAQ Final aim: ~ 1 MHz DATA THROUGHPUT USB module Martina Bucciantonio - TERA Foundation

11. GEMROC ASIC working principle 32-channel ASIC : readout time, space and amplitude Random rate ≳1MHz Each channel is split into Energy channel for analog pulse-height  pileups avoided Timing channel for self-triggering  low noise and thresholds Analogpulse 32MHz Multiplexed Analog Output Buffer Slow Shaper 100ns Peak Detect & Hold Analog FIFO LVDS IN<31:0> Preamplifier (variable gain) Digital data 125MHz Digital Output Buffer Digital FIFO Comparator with TWC Time Stamp Latch Fast Shaper 60ns 8 lines bus LVDS x 32 Mask register Calib Time Stamp Generator Token Manager I2C Control Logic Continuous readout of the memory (FIFOs) Differences between channels adjusted by individually (TrimDACs) Derandomization of data and zero suppression in the token-based readout Self-triggering mode - readout initiated by the input signals Internal testability functions Switchable gain and signal polarity selection Martina Bucciantonio - TERA Foundation

12. GEMROC Front End 128 wires connected in pairs -> 64 channels GEMROC design: Novel design ASIC (0.35m CMOS) tuned for 2-D readout of GEM used in the PRR diode protection 2 GEMROC ASIC (32 channels/each) Power+I2C High frequency connector ->DAQ board ASIC2 ASIC1 Analog 1LVDS line Analog 1LVDS line Digital 8LVDS lines Digital 8LVDS lines Martina Bucciantonio - TERA Foundation

13. GR_DAQ Altera Cyclone III • provides power, biases and thresholds • for 4 GEMROC ASICs • - serialization and digitalization of the GEMROCs analog output buffer through the pipeline 12-bit ADC , serial LVDS output (1 ADC per 2 GEMROC-FE board) • Linear Technologies LTC2265-12 to Master board (Imb-SDAQ) ASIC1 Digit FPGA ASIC3 Digit ASIC2 Digit ASIC4 Digit 32 MHz ASIC1 Analog ASIC3 Analog GEMROC-FE boards connected on the bottom side ASIC2 Analog ASIC4 Analog 125 MHz 32 MHz Martina Bucciantonio - TERA Foundation

14. Single GR_DAQ Firmware 8 parallel lines/6clocks serial data sending - De-GrayingASICs’ time stamps and channel IDs - Reading digitized analogue amplitudes - Adding course time stamp to let it run without overflow - Adding GR_DAQ IDs - Combining above to create 48bit frame of raw hit - Queuing data into FIFO’s for each ASIC - Serial encoding the data - Parallelizing in 8 lines - Serial sending to the SDAQ End transfer Serial encoder Data ready Next data Manager FIFO FIFO FIFO FIFO 256x42 bits Merger Merger Merger Merger ADC 2 ADC 2 ASIC 3 ASIC 4 ASIC 3 ASIC 4 Martina Bucciantonio - TERA Foundation

15. SDAQ • generic DAQ , flexible for different read-out applications • ALTERA Cyclone III FPGA • programmed by and communicate with • PC via a QuickUSB Module • (Bitwise Systems) • embedded IC interface • transfer data rate up to 48 Mbytes/s • data from the 6 GR_DAQs • Voltage, temperature controls • Slow controls Firmware for 1 GR_DAQ FIFO x6 Manager master SDAQ FIFO 256 x 44 bits Parallel Decoder 8 parallel lines/6clocks serial data sending (encoder) Martina Bucciantonio - TERA Foundation

16. SDAQ Firmware to PC for 6 GR_DAQ QuickUSB module 16 lines fast bus FIFO 16384 x48bits Manager master SDAQ SDAQ FPGA Martina Bucciantonio - TERA Foundation

17. DATA format GR_DAQ FPGA data_in at 32MHz 6 clock latency because of ADC at 125MHz Channel id 19……15 overflow 13 Time Stamp 11…………….….0 Data const 14 pileup 12 ADC data 11……………0 GR_DAQ FPGA data_out on/off 47 GR id 46…44 ASICid 43,42 ADC data 41……………30 Time Stamp long 29……….…20 Channel id 19……15 overflow 13 Time Stamp 11…………….….0 Data const 14 pileup 12 For 3 channels/cluster in both XY @ 30MB/sec 1.6 MHz throughput SDAQ FPGA 6.25 MHz 8 clocks at 50MHz to data transfer QuickUSB Module up to 48Mbytes/sec Martina Bucciantonio - TERA Foundation

18. Preliminary DAQ timing performance Data packet transferred at 1.3 MSample/sec LVDS DATA (GR_DAQ_1) J1_clock_in (from GR_DAQ_1) 8 parallel lines transfer mode  6 GR_DAQ data out at 1.3 MSample/sec Martina Bucciantonio - TERA Foundation

19. DAQ improvements Used Resources GR_DAQ FPGA : 8% total logic element, 5% combinational functions, 6% dedicated logic registers, 50% PLL, 63% total memory bits SDAQ FPGA : 6% total logic element, 4% combinational functions, 4% dedicated logic registers, 25% PLL, 79% total memory bits Where do better SDAQ Cluster online QuickUSB or USB 3.0 Data transfer speed x10 Martina Bucciantonio - TERA Foundation

20. Conclusions - GEMROC ASIC novel design suitable for GEM detectors sparse readout: records only the channels with a trigger and their neighbours (increasing the readout speed) - parallel processing of 4 GEMROC ASICs /1 GR_DAQ - reading out digitized analogue amplitude using 12bits ADCs - time stamp, channel IDs properly delayed and merged with the analog data - complete GEM readout – for Parallel GR_DAQs data taking - DAQ speed ~ 1 MHz DATA THROUGHPUT Martina Bucciantonio - TERA Foundation

21. Backup Martina Bucciantonio - TERA Foundation

22. Hadrontherapy : secondary particles GEANT 4 simulation 12C 300 MeV/u → H2O p • electromagnetic interactions • Dose deposition • Nuclear fragmentation • - High Probability Influence on dose deposition • - Secondary particles • - Radioactive isotopes (b+) • - g, n, p, fragments secondaries for the online treatment control n g dose Depth (cm) • QA techniques: • TOF-PET • Prompt gamma imaging • Interaction Vertex Imaging (IVI) Martina Bucciantonio - TERA Foundation

23. INTERACTION VERTEX IMAGING (ENVISION WP3) Dose control by secondary particles emission GEMs position sensitive detector for high momentum outgoing proton tracks reconstruction Tracks intersected with the known direction of the beam  • interaction vertex density distribution in the target • energy loss profile and penetration depth of the beam (indirectly) Simulation results (12C ion beam) Solevi, PhD Milano Bicocca (2007) Martina Bucciantonio - TERA Foundation

24. GEM • High-gain position-sensitive proportionalchamber • Dense hole pattern on thin, metal-clad polymer foil • Electrons • released in the overlaying gas layer • sucked into the holes • multiplied in the high electric field (50-70 kV/cm) • passed in the lower region • 3 GEM foils in cascadeusedto increasethe gain Martina Bucciantonio - TERA Foundation

25. GEM electrical parameters Martina Bucciantonio - TERA Foundation

26. GEMROC Internal Calibrations S-curves after trimming S-curves before trimming (ASICs 8-9) (ASICs 8-9) - Trimming to get the same threshold level for all the ASICs’ channels - 5-bit trimming DAC implemented in each channel - The channel-to-channel threshold spread reduced to 1.3 LSB (1 ThLSB0.1 fC) Analog calibration pulse Martina Bucciantonio - TERA Foundation

27. S-curve(ASICs 8-9-10-11) - on GEM ADC calibration Calibration pulse: 10fC Step 10, Th 25 S-curve (trimming) Calibration pulse: 10fC High Gain Martina Bucciantonio - TERA Foundation

28. Electronic noise (ASICs 8-9-10-11) – on GEM Martina Bucciantonio - TERA Foundation

29. Setup • GEM 30x30 cm2 • Strips connected in pairs: 800 μm pitch •  lesssignal/noise because of the large capacitance, but • compensated by larger signals (shared between several strips) • Ar/CO270/30 gas mixture • CAEN NIM HV module (N470) • 2 GEMROC Front-End boards • DAQ system • Control software for data taking • 55Fe source DAQ system GEMROC front-ends ASICs 10-11 ASICs 8-9 GEM 30x30 cm2 Bottom Plane (strip width 340 μm ) Martina Bucciantonio - TERA Foundation

30. GEM 30x30 preliminary tests with GEMROC offline analysis HV OFF HV ON =4000V (No 55Fe source) • Noise spectra: • threshold 50LSB • 55Fe spectra: • Same threshold, same ASICs 55Fe spectra HV 3980 Channels over threshold distribution 1 channel 2 channels 3 channels 4 channels external electromagnetic noise  5 channels

31. Software interface • LabView programs for: • configure GEMROCs • I2C control Martina Bucciantonio - TERA Foundation

32. Software interface • LabView programs for: • threshold scan • ASICs row data analysis • GEMROCs calibration • ….and so on

33. GEM data analysis • LabView programs for: • Data analysis • (cluster, fake eraser…) online analysis