1. RatCAP: A Small, Head Mounted PET Tomograph for Imaging the Brain of an Awake Rat Craig Woody Brookhaven National Lab

2. C.Woody, RatCAP, 6/20/042 The Collaboration Part of a larger project for Imaging The Awake Animal also involving motion tracking in both PET and MRI BNL

3. C.Woody, RatCAP, 6/20/043 The People BNL Physics: C. Woody, S. Stoll, M. Purschke, W.Lenz, M. Lenz, S.Boose Chemistry: D. Schlyer, A. Villanueva, J. Fowler Medical: P. Vaska, N. Volkow* (now at NIH) Instrumentation: P. O’Connor, V. Radeka, J.-F. Pratte*, S. Junnakar, G. DeGeronimo, B.Yu, A.Kandasamy, K.Wolniewicz, S. Rescia Stony Brook S. Shokouhi, A. Kriplani, S.Krishnamoorthy, S.Southekal (BME students) I. Rampil, C.Page, A.Bell (Dept. of Anesthesiology) Sherbrooke* R.Lecomte, R.Fontaine, S. Robert

4. C.Woody, RatCAP, 6/20/044 Positron Emission Tomography g e+e+ g e+e+ e-e- Radioisotope PET Scanner PET detects coincident 511 keV gamma rays Isotopes emit positrons with energies of a few hundred keV range ~ few mm

5. C.Woody, RatCAP, 6/20/045 Positron Emitting Radiotracers Organic molecules are labeled with positron emitting isotopes and used as tracers Tracers can be used to study neurotransmitter activity in the brain

6. C.Woody, RatCAP, 6/20/046 The Problem Anesthesia can greatly depress brain functions and affect the neurochemistry that one is trying to study Cannot study animal behavior while under anesthesia One wants to use PET to study the neurophysiological activity and behavior in laboratory animals in order to understand and treat these effects in humans. However, animals currently need to be anesthetized during PET imaging.

7. C.Woody, RatCAP, 6/20/047 Neurotransmitter Activity in the Brain Drugs like cocaine can block the re-uptake sites for neurotransmitters like dopamine which upsets the normal equilibrium and can cause effects of addiction DA MAO A DA signal DA 11 C-Cocaine

8. C.Woody, RatCAP, 6/20/048 Dopamine Studies using PET J.Fowler (BNL)

9. C.Woody, RatCAP, 6/20/049 Effects of Anesthesia on the Brain Blood flow and metabolism depend on type and dose of anesthetic How is global vs regional cerebral blood flow (CBF) modulated by anesthetics Cerebrovascular reactivity to CO 2 is perturbed Suppression of neural activity Effects are different for different animals and species M.Pomper, Johns Hopkins

10. C.Woody, RatCAP, 6/20/0410 Effects of Anesthesia in Primates The effect of anesthesia on the uptake of β-CFT on dopamine transporters in the monkey brain. From the 9th International Conference: Peace Through Mind/Brain Science, Hamamatsu, Japan, Jan. 30-31, 2002 Response of regional CBF to levodopa in baboons T.Hershey et.al., Exp. Neurol 2000; 166:342

11. C.Woody, RatCAP, 6/20/0411 Effects of Anesthesia in Humans Reduction in glucose metabolism with isoflurane D.B. Stout et al, UCLA 1998 Anesthetic and dosage effect on regional CBF K.Kaisti, Anesthesiology 2002; 96:1358

12. C.Woody, RatCAP, 6/20/0412 RatCAP: Rat Concious Animal PET Mockup of the portable ring on the head of a rat A septa-less, full-ring tomograph with a diameter of 4 cm and an axial extent of 2 cm, suspended by a tether, which will allow nearly free movement of the awake animal Harderian gland (at eye position) microPET images RatCAP mounting position superimposed on Concorde R4 MicroPET image of a rat brain

13. C.Woody, RatCAP, 6/20/0413 Design Requirements The tomograph ring must be light enough to be supported by the rat and allow reasonable freedom of movement Light weight detectors (~ 150 g total weight) Light weight electronics with low power dissipation  New custom ASIC High data rates and large singles background Small field of view and large parallax effects Limited sampling due to space and weight requirements Must be rugged enough withstand activity of the rat

14. C.Woody, RatCAP, 6/20/0414 Freedom of Movement Similar support structures are used in microdialysis experiments “Ratturn Bowl” Tether support system allows reasonable freedom of movement

15. C.Woody, RatCAP, 6/20/0415 RatCAP Support System Gimbal ring allows head movement Weight is completely counterbalanced (animal only feels inertia)

16. C.Woody, RatCAP, 6/20/0416 Tests with the Mockup on Live Animals Inner support ring Support ring attached to head using only bone cement Complete ring assembly Rat is able to move freely about with entire ring assembly mounted to head

17. C.Woody, RatCAP, 6/20/0417 Tomograph Ring Ring containing 12 block detectors of 2x2 mm 2 x 5 mm deep LSO crystals with APDs and integrated readout electronics Readout chip APD LSO Socket LSOAPD

18. C.Woody, RatCAP, 6/20/0418 Sensitivity Comparison of sensitivity with other small animal PET scanners BNL RatCAP 4 cm Ø x 2 cm ~ 150 cps/  Ci (0.41%) Concord m PET R4 15 cm Ø x 8 cm (1.86%) UCLA m PET 17 cm Ø x 2 cm Intrinsic detector sensitivities are essentially the same Coincidence sensitivities are proportional to axial acceptance

19. C.Woody, RatCAP, 6/20/0419 Parallax Effects Due to the small diameter of the ring, the parallax error is a major factor in determining the spatial resolution 2 cm

20. C.Woody, RatCAP, 6/20/0420 Spatial Resolution and Field of View Resolution at the edge of the rat brain Single layer - 10 mm ~ 2.5 mm Double layer - 5 mm ~ 2.0 mm First prototype will be a single layer of 5 mm crystals Will sacrifice sensitivity for improved spatial resolution SimSet Simulation

21. C.Woody, RatCAP, 6/20/0421 LSO Crystal Arrays Studying different types of crystal arrays to optimize light output and energy resolution CTI white powder reflector Proteus unbonded 3M reflector Proteus single crystals

22. C.Woody, RatCAP, 6/20/0422 Optimization of Light Collection unwrapped air unwrapped cookie wrapped cookie wrapped air Dashed = Measured Solid = Opticad OPTICAD Ray Tracing Program 2x2 mm 2 single crystals wrapped in 3M reflector and coupled with a silicone cookie to a calibrated PMT N pe /MeV = N  /MeV x  x QE 2400 = 25,000 x 0.4 x 0.24

23. C.Woody, RatCAP, 6/20/0423 Comparison of Light Output and Energy Resolution of Different Crystals Prototype LSO arrays measured with Hamamatsu 4x8 APD Conclusion: Crystals wrapped with 3M reflector and air gap give the highest light output and best energy resolution

24. C.Woody, RatCAP, 6/20/0424 Production of Crystal Arrays Cutting, polishing and assembly process Average light yield ~ 5400 p.e./MeV (~ 22% higher than prototype) All final crystal arrays have been received and tested Arrays were produced by Proteus, Inc (Chagrin Falls, OH)

25. C.Woody, RatCAP, 6/20/0425 Measurement of Light Output and Energy Resolution of Crystals Arrays Energy resolution ~ 18% for 511 keV  ’s (measured with 3rd prototype ASIC)  ~ 13% Total spread ~ ± 40% Spread within a crystal array ~ ± 10%

26. C.Woody, RatCAP, 6/20/0426 Characterization of Avalanche Photodiodes 4x8 array 1.6 x 1.6 mm 2 active pixel area Hamamatsu S8550 N pe ~ 2700 Typical Gain ~ 50  ~ 135K signal e’s Common voltage for each 16 channels Dark current < 40 nA (~ 1.2 nA/ch) Expected noise in final ASIC ~ 1100 e’s (C T ~ 10 pF) All final APD arrays have been received and tested

27. C.Woody, RatCAP, 6/20/0427 APD Gain and Quantum Efficiency Variation Measured with N 2 laser + optical fiber Gain + Quantum Efficiency variation Total spread ~ ± 20 % Channel to channel differences dominated by quantum efficiency variation

28. C.Woody, RatCAP, 6/20/0428 Electronics & Readout System  Tomograph ring with detectors and front end readout electronics are mounted to the head of the rat  Tether carries signals, high and low voltage power to a Control Module Data collection module Control Module  CM communicates with a VME based Data Collection Module that adds time stamp information and reads out the data

29. C.Woody, RatCAP, 6/20/0429 VME Module  Provides primary communication with tomograph ring  Receives singles data in list mode  Adds time stamp  Can do coincidence processing  Provides serial communication for downloading DAC values for setting thresholds FPGA

30. C.Woody, RatCAP, 6/20/0430 Flexible Readout Circuit  Used to mount crystals, APDs and readout electronincs  Contains individual data transfer line for each block  Distributes high voltage for APDs, low voltage for ASICs, serial lines for setting DACs Control card contains drivers to communicate with VME module

31. C.Woody, RatCAP, 6/20/0431 Custom Front End Readout Chip Custom ASIC in 0.18 μm CMOS 4 mW per channel 32 channels  125 mW per chip ~ 1.5 W total on ring Final size: 4.2 x 1.7 mm Prototype 3 J.-F. Pratte

32. C.Woody, RatCAP, 6/20/0432 Custom Front End Readout Chip Discriminator pulse is encoded to give 5 bit address Leading edge of encoded serial pulse train gives time information 100 MHz clock Maximum 100 KHz/block  0.7% deadtime CSP gain 3.3 mV/fC Overall gain 15 mV/fC Digital Output (P.O’Connor) Preamp/Shaper/ZCD

33. C.Woody, RatCAP, 6/20/0433 ASIC Performance Slope = 15.2 mV/fC ENC = 902  29 e’s rms LinearityEnergy Resolution Peaking Time Measurements C in MeasurementSimulation 0 pF73.2 ± 1.4 ns74.0 ns 12 pF80.4 ± 1.8 ns80.0 ns FWHM = 18%

34. C.Woody, RatCAP, 6/20/0434 ASIC Performance Noise ENC min = 1117 @ 100 ns Cin tot = 20 pF Timing (ZCD) Noise is dominated by photostatistics contribution Expect ~ 2.4 ns rms total ENC electrons (RMS)

35. C.Woody, RatCAP, 6/20/0435 ASIC Performance - Timing Resolution 3.54 ns FWHM6.7 ns FWHM (2.4 ns rms) LSO+APD vs BaF2+PMT Electronic Timing Resolution at 511 keV equivalent energy Coupled to LSO/APD with 511 keV  ’s timed against a BaF 2 scintillator w/PMT Same with CFD Preamp/Shaper + ZCD

36. C.Woody, RatCAP, 6/20/0436 Rotatable Stage for Obtaining Partial Tomographic Data Source phantoms mounted on rotatable stage to simulate full tomographic ring

37. C.Woody, RatCAP, 6/20/0437 Point Source Resolution 1 mm dia 68 Ge point source at R=1.6 mm in rotation stage profile axis (mm) 246 0 im ag e int en sit y (ar b) 0 2 4 FWHM = 2.1 mm Intrinsic spatial resolution measured with 22 Na point source (< 1 mm dia) Average peak FWHM = 1.28 mm Concorde P4 MicroPET = 1.75 mm UCLA MicroPET = 1.58 mm

38. C.Woody, RatCAP, 6/20/0438 Source Images Single 1 mm dia. 68 Ge point source 1.6 mm off axis 2 mm Multiple 68 Ge point sources (~ 2mm) spaced ~ 4 mm 4 mm

39. C.Woody, RatCAP, 6/20/0439 Image Reconstruction P(  r) r  y x LOR  + Source Sinogram of all projections r  Each crystal pair defines a line of response Lines of response give projections Each line of response is specified by (r,  Data are binned into sinograms which serve as the input for image reconstruction

40. C.Woody, RatCAP, 6/20/0440 Modeled Reconstructed Images Monte Carlo simulations using SimSet Reconstruction using Filtered Back Projection w/ramp filter Fully sampled image of four phantom sources Image reconstructed using incomplete data set with interpolation Image reconstructed using incomplete data set Test Phantom S.Shokouhi

41. C.Woody, RatCAP, 6/20/0441 Iterative Image Reconstruction Maximum Likelihood Expectation Maximization (MLEM method) y = A x System Matrix Sinogram data Image voxels x is a vector of all image voxels y is a vector of all projection elements corresponding to x A is the System Matrix which converts the image voxel data to its projections A ij is just the probability that a decay in image voxel x j produces a count in projection y i System Matrix can include many physics effects in the detector and object Scatter Attenuation Detector geometry Detector efficiency Positron range Noise Shepp & Vardi, IEEE Trans. Med. Imaging (1982) 113-122 Uniform distribution used to generate SM First guess for real image reconstruction Forward projection of Monte Carlo data Real data

42. C.Woody, RatCAP, 6/20/0442 Ratcap Reconstruction using MLEM Sinograms Full 3D sinograms (one sinogram per crystal pair) 48 crystals/ring x 8 rings  47 R bins x 24  bins (interleaved) x 15 segments System Matrix 921x15x72192 = 10 9 elements (but very sparse)  2GB uncompressed data Generate ~ 10 7 photons per 1 mm 3 voxel Detector Geometry using SimSET Unique features of the RatCAP: Small but nearly completely filled field of view Large parallax error (no DOI information) Finite pixel size (2x2 mm 2 ) at small radius Incomplete azimuthal sampling Univ. of Washington PET Simulation software Will move to more detailed simulation using GATE (new PET simulation software based on GEANT4)

43. C.Woody, RatCAP, 6/20/0443 First results from MLEM applied to RatCAP Point source reconstruction Single 1 mm 3 voxel Next step... Derenzo Phantom Need to study reconstruction of more complicated objects Tradeoff of resolution vs noise

44. C.Woody, RatCAP, 6/20/0444 Other Applications A wide range of quantitative PET studies using tracer kinetic modeling demand accurately measured radiotracer concentration in arterial blood as a function of time after injection (Arterial Input Function). The common method of measuring the input function is the invasive withdrawal of blood from a wrist artery. However, because of its health risks for both patients and hospital personnel, it is not compatible with clinical studies. A small ring tomograph similar to the RatCAP can be used to image the artery and measure the input function Non-Invasive Wrist Monitor

45. C.Woody, RatCAP, 6/20/0445 Wrist Monitor for Measuring the Arterial Input Function For human studies, the input function is taken from the radial artery in the wrist. Activity in the surrounding veins produce a significant background which can be rejected using the good spatial resolution of the wrist monitor. Planar image of a 1 mm diameter 68 Ge line source A.Villaneuva Wrist Phantom

46. C.Woody, RatCAP, 6/20/0446 Simulated Input Function Measurement Measure activity as an aqueous solution of a 11 C - labeled radioisotope passes between two block detectors w/scatterw/o scatter Expected input function resolution using Wrist Monitor

47. C.Woody, RatCAP, 6/20/0447 Beta Scintillation Microprobe Radioisotopes used in PET emit positrons with energies of a few hundred keV which have a range of several mm in blood or tissue. This range is comparable to the spatial resolution obtained in most PET cameras. These positrons can be detected directly using plastic or crystal scintillators. With crystal scintillators which have high density, the positrons will range out, depositing all of their energy in the crystal and producing a large signal. Small scintillation probes can be used to directly measure the radiotracer concentration in the blood or tissue. C.L.Woody et.al., IEEE Trans. Nucl. Sci NS-49 (2002) 2208-2212.

48. C.Woody, RatCAP, 6/20/0448 Comparison of LSO vs Plastic Scintillator Response of LSO and plastic scintillation probes to betas ( 32 P) and gamma rays ( 137 Cs). Range and energy loss of positrons in LSO and plastic scintillator

49. C.Woody, RatCAP, 6/20/0449 Sensitivity and Position Resolution Material LSO LSO Plastic Size 0.3 x 0.5 mm 0.5 x 1.5 mm 1.0x1.0 mm Volume 0.035 mm 3 0.295 mm 3 0.795 mm 3 Sensitivity 10.1 20.7 51.2 (Hz/mCi/cc) Region of sensitivity around the probe can be selected by adjusting the readout threshold to improve spatial resolution

50. C.Woody, RatCAP, 6/20/0450 Probe Construction LSO microprobe consisting of 0.5 mm diameter x 1.5 mm long LSO crystal wrapped with several layers of white reflecting teflon and covered with polyester shrink tubing.

51. C.Woody, RatCAP, 6/20/0451 Rat Brain Studies with Microprobe Uptake of 11 C-methylphenidate in the nucleus accumbens region of a rat brain with LSO probe nucleus accumbens ~ 2 mm

52. C.Woody, RatCAP, 6/20/0452 Input Function Measured with Microprobe LSO microprobe (0.3 mm dia. x 0.5 mm) inserted inside an 18 gauge syringe needle for blood flow study. Activity of 11 C-tyrosine measured in baboon blood flow during a PET scan using a syringe mounted LSO microprobe. Input function

53. C.Woody, RatCAP, 6/20/0453 Summary The ability to carry out PET imaging studies in live, awake animals would provide valuable new information on the neurophysiological behavior in both animals and humans The RatCAP will provide a new tool for carrying out these studies in laboratory rats. Preliminary studies have shown that the design parameters of the RatCAP have been met and we are proceeding to complete the design and construct the actual detector. A small imaging tomograph such as this may have other useful applications in nuclear medicine

54. C.Woody, RatCAP, 6/20/0454 Backup Slides

55. C.Woody, RatCAP, 6/20/0455 Effects of Anesthesia The effect of anesthesia on the uptake of β-CFT on dopamine transporters in the monkey brain. H.Hideo et.al., Synapse 42 (2001) 273-280 Reduction in glucose metabolism with isoflurane in humans Similar effects are seen in the rat D.B. Stout et al, UCLA 1998

56. C.Woody, RatCAP, 6/20/0456 Electronics & Readout System  No analog information  Pixel address is encoded and data is transmitted by individual serial line for each block to the Data Collection Module  Single ZCD per channel generates time information  DCM adds time stamp to each event and does coincidence timing or list mode processing of singles data Singles rates up to 100 kHz/block 1 byte/block  1.2 MB/sec off ring DCM produces 64 bits (43 bit time stamp)  ~ 10 MB/sec to PC Block 1 Block 2 Block … Block 12 DCM CM PC VME Time Stamp 500 MHz HV LV DAC Serial data 100 MHz x12 8 bits