Home Search Collections Journals About Contact us My IOPscience Performance measurements of a depth-encoding PET detector module based on positionsensitive avalanche photodiode read-out This article has been downloaded from IOPscience. Please scroll down to see the full text article. 24 Phys. Med. Biol. 49 4293 (http://iopscience.iop.org/31-9155/49/18/7) View the table of contents for this issue, or go to the journal homepage for more Download details: IP Address: 128.112.85.16 The article was downloaded on 11/12/21 at 18:33 Please note that terms and conditions apply.
INSTITUTE OF PHYSICS PUBLISHING Phys. Med. Biol. 49 (24) 4293 434 PHYSICS IN MEDICINE AND BIOLOGY PII: S31-9155(4)8165- Performance measurements of a depth-encoding PET detector module based on position-sensitive avalanche photodiode read-out P A Dokhale 1, R W Silverman 2, K S Shah 3, R Grazioso 3, R Farrell 3, J Glodo 3, M A McClish 3,GEntine 3, V-H Tran 1 and S R Cherry 1 1 Department of Biomedical Engineering, University of California-Davis, One Shields Avenue, Davis, CA 95616, USA 2 Department of Molecular and Medical Pharmacology, UCLA, Los Angeles, CA 995, USA 3 Radiation Monitoring Devices Inc., Watertown, MA 2172, USA Received 2 May 24 Published 3 September 24 Online at stacks.iop.org/pmb/49/4293 doi:1.188/31-9155/49/18/7 Abstract We are developing a high-resolution, high-efficiency positron emission tomography (PET) detector module with depth of interaction (DOI) capability based on a lutetium oxyorthosilicate (LSO) scintillator array coupled at both ends to position-sensitive avalanche photodiodes (PSAPDs). In this paper we present the DOI resolution, energy resolution and timing resolution results for complete detector modules. The detector module consists of a 7 7matrixof LSO scintillator crystals (1 1 2 mm 3 in dimension) coupled to 8 8mm 2 PSAPDs at both ends. Flood histograms were acquired and used to generate crystal look-up tables. The DOI resolution was measured for individual crystals within the array by using the ratio of the signal amplitudes from the two PSAPDs on an event-by-event basis. A measure of the total scintillation light produced was obtained by summing the signal amplitudes from the two PSAPDs. This summed signal was used to measure the energy resolution. The DOI resolution was measured to be 3 4 mm FWHM irrespective of the position of the crystal within the array, or the interaction location along the length of the crystal. The total light signal and energy resolution was almost independent of the depth of interaction. The measured energy resolution averaged 14% FWHM. The coincidence timing resolution measured using a pair of identical detector modules was 4.5 ns FWHM. These results are consistent with the design goals and the performance required of a compact, high-resolution and high-efficiency PET detector module for small animal and breast imaging applications. 31-9155/4/184293+12$3. 24 IOP Publishing Ltd Printed in the UK 4293
4294 PADokhaleet al 1. Introduction There has been considerable interest in recent years in developing high-resolution positron emission tomography (PET) systems for applications in small animal imaging and breast cancer imaging (e.g. Cherry et al 1997, Moses et al 1997, Knoess et al 23, Siedel et al 23, Ziegler et al 21, Surti et al 23, Huber and Moses 1999, Raylman et al 2, Doshi et al 21). A high-resolution, high-sensitivity PET detector based on discrete detector elements requires the use of long and narrow scintillation crystals. It is well known that the use of long narrow crystals leads to significant depth of interaction (DOI) or parallax errors, degrading the spatial resolution. In cylindrical geometry scanners, this leads to a radial spatial resolution component that degrades with increasing radial offset from the centre of the scanner. For polygonal geometry systems, there is a more uniform resolution degradation due to DOI effects across the transaxial field of view. For both geometries, axial resolution also is degraded for 3D acquisition, with the worst effects at the centre of the axial field of view. These effects can be reduced or eliminated by measuring the depth of interaction of the photon in the scintillator crystal (Moses et al 1991). A number of groups are designing detectors with DOI capability that can determine interaction locations in 3D in a scintillator array (Miyaoka et al 1998, Rogers et al 1996, MacDonald and Dahlbom 1998, Saoudi et al 1999, Moses and Derenzo 1994, Liu et al 21). The idea of extracting the DOI information by coupling both ends of the scintillator array to photodetectors is not new (Moses and Derenzo 1994, Shao et al 2, 22, Huber et al 21). However, in previous approaches at least one of the photodetectors had individual photosensors for each crystal in the array, leading to large numbers of read-out channels and associated electronics. In this work an avalanche photodiode (APD) design that provides intrinsic position sensing capability has been used, allowing a significant reduction in the number of read-out channels needed to decode the interaction location within the crystal array. We have previously presented design studies and preliminary results for a detector module consisting of a single LSO crystal with read-out at both ends by two 8 8 mm 2 position sensitive APDs (PSAPDs) (Shah et al 24). These studies showed encouraging results, however, they were carried out with only a single crystal, thus avoiding any deleterious effects due to light crosstalk and inter-crystal scattering. Therefore, an important next step in exploring the feasibility of this approach was to scale up the experiments to a full scintillator array and evaluate the performance of complete detector modules. In this paper we report results of the DOI resolution and energy resolution measurements made on a 49 element detector module made up from 1 1 2 mm 3 LSO scintillator elements. Timing resolution measurements were also made between two complete detector modules. 2. Materials and methods 2.1. Position sensitive avalanche photodiodes The construction and basic performance of the PSAPDs used in this work have been previously reported (Shah et al 22, 24). The device consists of a deep diffused, high gain APD with the front, optical entrance face followed by drift and multiplication (or space charge) regions. The back face of the APD consists of a resistive layer with four corner contacts (or anodes) that provide position resolution based on comparison of the signal measured at each corner anode. Thus, the device produces four position-related signals that vary in a continuous manner for events across the surface of the APD. As a result, a large imaging area can be decoded from just five APD outputs (the common cathode is used to gather timing information, and the four corner anodes generate position information see figure 1). For this work we
Performance measurements of depth-encoding PET detector module 4295 Front contact A B Drift region Resistive layer Surface charge region C D 4 Back contacts Front view (B+D) (A+C) X = (A+B+C+D) (A+B) (C+D) Y = (A+B+C+D) Back view Figure 1. A position sensitive silicon avalanche photodiode with four corner anode design. Table 1. Typical specifications for 8 8mm 2 PSAPDs. Characteristics Typical value Active area 8 8mm 2 Gain @ 175 V 1 Capacitance, C (pf) 45 pf Q. E. (4 7 nm) 6% Noise @ gain 1 2 electrons (FWHM) Rise time 1ns utilize 8 8mm 2 PSAPDs. Typical performance values for these devices are provided in table 1. 2.2. Detector module design The detector module was constructed by coupling a 7 7 element LSO scintillator array with two 8 8mm 2 PSAPDs at either ends. Each individual crystal in the array measured 1 1 2 mm 3, with a centre to centre spacing of 1.6 mm. The 1 1mm 2 surfaces of each crystal were mechanically polished while the other four surfaces were not given any surface treatment and were as-cut by the saw. We determined in preliminary experiments that this configuration was necessary to provide sufficient depth sensitivity (data not shown). The crystals were glued together with a multi-layer polymer reflector (3M Corp., St. Paul, MN) between each crystal. A thin layer of optical grease was used as a coupling material between the PSAPDs and the LSO array. Figure 2 shows a schematic diagram of the detector module. The thin cross section of the PSAPDs results in less than 7% attenuation for 511 kev photons passing through the front PSAPD prior to reaching the scintillator array. The ratio of the two PSAPD signals amplitudes were used to determine the DOI. The relatively high gain of these PSAPDs is important in accurately determining the signal ratio, especially at the ends of the crystal where the distant PSAPD receives a relatively small signal. The five contacts from each PSAPD were fed to electronics as described in the following section. Two complete detector modules were constructed for evaluation.
4296 PADokhaleet al Figure 2. Schematic of a PSAPD PET detector module. PSAPD 1 7 x 7 LSO array TRANSLATION TABLE 35 mm 35 mm PMT PSAPD 2 22 Na source 2 x 2 x 1 mm 3 LSO Figure 3. Detector and source geometry for the DOI measurements. 2.3. Energy resolution and DOI resolution To measure the DOI resolution, a holder was constructed that allowed precise alignment of the detector and the source (figure 3). A single LSO crystal (2 2 1 mm 3 ) was coupled to a1.5 diameter single-channel PMT for electronically collimating the interactions at different DOI positions. This second detector was mounted on a translation table for 3D positioning with respect to the PSAPD detector module. A.5 mm diameter 22 Na point source, which was located inside the source holder attached to the translation table, was placed between the LSO array and the collimating LSO crystal. The distance from the source to both detectors was 35 mm. By moving the translation table, interactions at different DOI locations in the LSO array were selected by acquiring coincidence events between the LSO array and the LSO crystal coupled to the PMT. A schematic of the signal processing electronics set-up is shown in figure 4. Amplification for each of the five PSAPD outputs was provided by an ac-coupled, single channel, charge
Performance measurements of depth-encoding PET detector module 4297 Pre-amp PSAPD 1 Na 22 PMT Pre-amp PSAPD 2 CFD Fan-in / Fan-out FFA CFD Stop Start TAC Shaping amplifier ADC Board Gate & Delay Figure 4. Schematic showing electronics and data acquisition set-up for DOI measurements. sensitive pre-amplifier (Cremat model CR-11, Watertown, MA). The amplified signals from the four corner anodes were then shaped and amplified again by a programmable 16-channel spectroscopy amplifier (CAEN model N568B, Viareggio, Italy) with a shaping time of 2 ns. The four amplified signals were simultaneously converted into 12-bit digital format by a 16-channel PCI-based ADC board (PCI-416L, Datel Inc., Mansfield, MA) upon receipt of the timing or coincidence trigger. The trigger was generated by the time coincidence between the sum of the cathode signal of two the PSAPDs and the PMT signal. The summed cathode signal was shaped and amplified by a fast filter amplifier (ORTEC model 579, Oak Ridge, TN) and then fed to a constant fraction discriminator (CFD) (Oxford Instruments Inc., Model TC 453, Oak Ridge, TN). The signal obtained from this CFD was used as a stop input to a time-to-amplitude converter/single channel analyser (TAC/SCA) (Oxford Instruments Inc., Model TC 863, Oak Ridge, TN). The PMT signal was amplified with a preamplifier and then fed to CFD. This CFD signal was used as a start input to the TAC/SCA. The TAC/SCA output voltage, proportional to the time difference in start and stop signals, was digitized on an event-by-event basis. The energy threshold was set just above the noise. The TAC/SCA unit was calibrated by adding two different delay times between the CFD and the stop input to TAC/SCA. The measured coincidence timing resolution between the PSAPD detector module and the second detector (a single LSO crystal coupled to a PMT) was 5.4 ns. The SCA time (LLD) and delta time (ULD) were adjusted so that the coincidence timing window was set to 12 ns. The data acquisition was triggered from the SCA output. The bias to both PSAPDs was set to approximately 175 V and adjusted so that the gain of each PSAPD was approximately 1. Data were collected at seven different DOI locations (2.5, 5, 7.5, 1, 12.5, 15 and 17.5 mm) along the 2 mm length of the crystal. From the source and the detector geometry we estimate that a band of approximately 2.5 mm in width was irradiated at each location. Because of the penetrating nature of 511 kev radiation, and practical constraints on the geometry of the experiment together with the need for reasonable counting statistics, it is difficult to achieve a much better source collimation than this.
4298 PADokhaleet al Pre amp Pre amp Fan-in / Fan-out Pre amp Pre amp Fan-in / Fan-out FFA FFA CFD CFD Start TAC Stop Delay Digitizer Figure 5. Schematic of the electronics set-up for coincidence timing resolution measurements. In a separate experiment, data were also collected in singles mode with a 22 Na source normally incident on each detector face. The list mode data from the two PSAPDs were used to obtain flood histograms. A MATLAB program was used to create crystal look-up tables for crystal identification for each module. The list mode data from the DOI experiments were then analysed by selecting events from specific crystals in the LSO array using the crystal look-up tables. The ratio of PSAPD1/(PSAPD1+PSAPD2) signals (as a measure of DOI) and the summed energy spectra (PSAPD1+PSAPD2) were computed on an event-by-event basis as a function of the DOI location for selected crystals. 2.4. Coincidence timing measurements For coincidence timing measurements, two identical detector modules were constructed. Figure 5 shows the schematic of the electronics for timing resolution measurements. The summed cathode signals from the two PSAPDs in each module were amplified and shaped by a fast filter amplifier (FFA) with a shaping time of 2 ns and then fed to the CFD. The CFD signal from the first module was used as a start input to a TAC/SCA and the CFD signal obtained from the second module was passed through a delay circuit and then used as the stop input to the TAC/SCA. The resulting voltage (proportional to the time difference in start and stop signals) was digitized on an event-by-event basis. The energy threshold was set to 15 kev. The TAC/SCA unit was calibrated by acquiring the output pulse amplitude with three different delay times given to the stop input. Data acquisition was controlled using Lab View 5.1 (National Instruments Corp., Austin, TX) software. 3. Results Figure 6 shows a flood histogram obtained from the 7 7 LSO array coupled at both ends to a8 8mm 2 PSAPD and exposed to a 22 Na source normally incident on the surface of the
Performance measurements of depth-encoding PET detector module 4299 Figure 6. Flood histogram of a 7 7 LSO array with crystals of 1 1 2 mm 3 coupled at both ends to a 8 8mm 2 PSAPD (only one of the two flood histograms is shown, the other is essentially identical) and exposed to a 22 Na gamma source. Irradiation is from the front surface through one of the PSAPDs. Signals from the three indicated crystals are presented in figure 8. PSAPD 1 2.5 mm 3 2 PSAPD 1 PSAPD 2 3 2 1 1 PSAPD 2 1 2 3 4 5 3 1 2 3 4 5 3 1 mm 2 1 2 1 1 2 3 4 5 1 2 3 45 5 3 3 2 2 1 1 17.5 mm 1 2 3 4 5 Pulse height 1 2 3 45 5 Pulse height Figure 7. Pulse height spectra from the two PSAPDs at either end of the LSO array for interaction depths of 2.5 mm, 1 mm and 17.5 mm.
43 PADokhaleet al 25 2 (a) 1.8 # Of Events 15 1 5 Ratio.6.4.2.2.3.4.5.6.7.8 5 1 15 2 # Of Events 25 2 15 1 (b) Ratio 1.8.6.4 5.2.2.3.4.5.6.7.8 5 1 15 2 # Of Events 25 2 15 1 5 (c) Ratio 1.8.6.4.2.2.3.4.5.6.7.8 PSAPD 1 (PSAPD1+PSAPD2) 5 1 15 2 Physical location (mm) Figure 8. Distribution of ratio signals on an event-by-event basis calculated from signals acquired at seven different DOI locations (2.5 mm apart) for edge (a), corner (b) and centre (c) crystals. Plots at right show ratio as a function of the physical location of the irradiation. array. All 49 crystals of the array are identified clearly. The pincushion distortion at the edge is expected from the simple four corner contact geometry of the PSAPD and does not present a problem for PET applications as long as the individual crystals can be identified in a look-up table. Signals were simultaneously acquired from both PSAPDs as the LSO array was irradiated from the side at different DOI locations. The DOI positions were defined with the origin at the end of the LSO crystal connected to PSAPD 1. Figure 7 shows typical energy spectra acquired at three different DOI positions 2.5, 1 and 17.5 mm for a LSO crystal that is at the centre of the array. The signal amplitude for this crystal measured by each PSAPD changes significantly with interaction depth. The distribution of the ratio of the signals, PSAPD1/(PSAPD1+PSAPD2), calculated on an event-by-event basis, for edge, corner and centre crystals is shown in figures 8(a), (b) and (c), respectively. Only photopeak events were considered to calculate the ratio distribution. The three crystals for which data are shown are indicated in figure 6.
Performance measurements of depth-encoding PET detector module 431 DOI Resolution (mm) DOI Resolution (mm) 6 4 2 5 1 15 2 6 4 2 (a) (b) Average resolution = 3.2 mm Interaction depth (mm) Average resolution = 3.1 mm 5 1 15 2 Interaction depth (mm) DOI Resolution (mm) 6 (c) Average resolution = 3.32 mm 4 2 5 1 15 2 Interaction depth (mm) Figure 9. DOI resolution as a function of interaction depth for edge (a), corner (b) and centre (c) crystals in the array. Each curve on the left-hand side of figure 8 corresponds to the distribution of ratio signals for interactions at one particular DOI position. The curves for all seven DOI locations are shown. The variation in the mean amplitude of the computed ratio shows the sensitivity of the ratio signal with respect to the DOI position. The width of the distributions is a measure of the uncertainty in the ratio calculation. The linearity of the ratio signal with the DOI position is shown in the plots on the right-hand side of figure 8. The DOI resolution can be estimated from the rate of change of the average ratio signal relative to the width of the distribution as described in (Moses et al 1991, Shao et al 22). Figure 9 shows the change in the DOI resolution with respect to depth for edge (a), corner (b) and centre (c) crystals. The DOI resolution was in the range 3 4 mm FWHM along the entire length of the crystal and for crystals in different locations within the array. The actual DOI resolution is probably significantly better than this given that a 2.5 mm wide band is irradiated at each position. The energy spectra obtained by summing the signals from the two PSAPDs (sum of all 8 anode signals) is shown in figure 1 as a function of the DOI location. The position of the photopeak varies by only about 5% as a function of interaction depth, demonstrating that
432 PADokhaleet al 17.5 mm 15 mm 12.5 mm 1 mm 7.5 mm 5 mm 2.5 mm 1 2 3 4 5 Pulse Height Figure 1. Energy spectra (sum of the eight anode signals from PSAPD 1 and PSAPD 2 signals for each event) for different DOI locations. The amplitudes have been offset to allow comparison of the photopeak position at each depth. 7 6 Timing Resolution = 4.5 ns (FWHM) 5 # of Counts 4 3 2 1 4 5 6 7 8 9 1 Time (ns) Figure 11. Coincidence timing resolution between two detector modules. total light collection is uniform along the length of the crystal. The energy resolution varied between 13% and 15% FWHM for different DOI positions. Figure 11 shows the coincidence timing spectrum obtained between two identical detector modules. The measured timing resolution was 4.5 ns FWHM. 4. Summary and conclusion DOI, energy and timing resolution were successfully measured with dual read-out PSAPD modules made up from 1 1 2 mm 3 LSO crystals. These results demonstrate that excellent (3 4 mm) and uniform DOI resolution can be achieved using this approach. It is likely that the DOI resolution is even better than this, given that a 2.5 mm wide band of the crystal was irradiated to make these measurements. In contrast to phoswich-style detectors, the DOI information provided by this detector is continuous in nature, and can be used in a PET imaging
Performance measurements of depth-encoding PET detector module 433 system to determine the appropriate line of response into which to bin a particular event. This may also allow finer sampling of the projection data (Virador et al 1998). Energy resolution varied from 13% to 15% FWHM along the length of the crystals, which is a significant improvement over fibre-optically coupled PET detectors developed by our group (Chatziioannou et al 21) using similar dimension LSO crystals. The total light signal is uniform along the length of the crystal, which means that energy thresholding can be applied in a depth-independent manner, an important consideration for PET scanner design. The deep crystal length of 2 mm ( 1.8 radiation lengths for 511 kev photons in LSO) used in these detectors, combined with the measured DOI resolution, should allow for a new generation of PET scanners for small animal or breast imaging that can simultaneously achieve a spatial resolution of around 1 mm with sensitivities that are significantly higher than currently available systems with that resolution. An additional benefit of the excellent DOI resolution is that the detector separation will be limited primarily by the dimensions of the object to be scanned; thus this approach offers opportunities to build compact systems with a relatively small number of detector modules. Acknowledgments This work was funded in part by NIH grant R21 CA96537 and by SBIR funding from NIBIB. The authors would also like to thank Concorde Microsystems Inc. for supplying the LSO arrays used in this work. References Chatziioannou A et al 21 Detector development for micropet: II. A 1 µl resolution PET scanner for small animal imaging Phys. Med. Biol. 46 2899 91 Cherry S R et al 1997 MicroPET: a high resolution PET scanner for imaging small animals IEEE Trans. Nucl. Sci. 44 1161 6 Doshi N K et al 21 MaxPET: a dedicated mammary and axillary region PET imaging system for breast cancer IEEE Trans. Nucl. Sci. 48 811 5 Huber J S et al 21 An LSO scintillator array for a PET detector module with depth of interaction measurement IEEE Trans. Nucl. Sci. 48 684 8 Huber J S and Moses W W 1999 Conceptual design of a high-sensitivity small animal PET camera with 4 π coverage IEEE Trans. Nucl. Sci. 46 498 52 Knoess C et al 23 Performance evaluation of the micropet R4 PET scanner for rodents Eur. J. Nucl. Med. Mol. Imaging 3 737 47 Liu H et al 21 Development of a depth of interaction detector for γ -rays Nucl. Instrum. Methods A 459 182 9 MacDonald L R and Dahlbom M 1998 Depth of interaction for PET using segmented crystals IEEE Trans. Nucl. Sci. 45 2144 8 Miyaoka R S et al 1998 Design of a depth of interaction (DOI) PET detector module IEEE Trans. Nucl. Sci. 45 169 73 Moses W W et al 1991 A new algorithm for using depth-of-interaction measurement information in PET data acquisition J. Nucl. Med. 32 995 Moses W W et al 1997 Design of a high-resolution, high-sensitivity PET camera for human brains and small animals IEEE Trans. Nucl. Sci. 44 1487 91 Moses W W and Derenzo S E 1994 Design studies for a PET detector module using a PIN photodiode to measure depth of interaction IEEE Trans. Nucl. Sci. 41 1441 5 Raylman R R et al 2 The potential role of positron emission mammography for detection of breast cancer. A phantom study Med. Phys. 27 1943 54 Rogers JG et al1996 A practical block detector for a depth-encoding PET camera IEEE Trans. Nucl. Sci. 43 324 8 Saoudi A et al 1999 Investigation of depth-of-interaction by pulse shape discrimination in multicrystal detectors read out by avalanche photodiodes IEEE Trans. Nucl. Sci. 46 462 7
434 PADokhaleet al Seidel J et al 23 Resolution uniformity and sensitivity of the NIH ATLAS small animal PET scanner: comparison to simulated LSO scanners without depth-of-interaction capability IEEE Trans. Nucl. Sci. 5 1347 5 Shah K S et al 22 Position-sensitive avalanche photodiodes for gamma-ray imaging IEEE Trans. Nucl. Sci. 49 1687 92 Shah K S et al 24 Position sensitive APDs for small animal PET imaging IEEE Trans. Nucl. Sci. 51 91 5 Shao Y et al 2 Design studies of a high resolution PET detector using APD arrays IEEE Trans. Nucl. Sci. 47 151 7 Shao Y et al 22 Dual APD array readout of LSO crystals: optimization of crystal surface treatment IEEE Trans. Nucl. Sci. 49 649 54 Surti S et al 23 Design evaluation of a-pet: a high sensitivity animal PET camera IEEE Trans. Nucl. Sci. 5 1357 66 Virador P R G et al 1998 Reconstruction in PET cameras with irregular sampling and depth of interaction capability IEEE Trans. Nucl. Sci. 45 1225 3 Ziegler S I et al 21 A prototype high-resolution animal positron tomography with avalanche photodiode arrays and LSO crystals Eur. J. Nucl. Med. 28 136 43