A temperature-dependent gain control system for improving the stability of Si-PM-based PET systems

A temperature-dependent gain control system for improving the stability of Si-PM-based PET

systems

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2011 Phys. Med. Biol. 56 2873

(http://iopscience.iop.org/0031-9155/56/9/015)

Download details:

IP Address: 193.140.216.203

The article was downloaded on 07/06/2011 at 11:41

Please note that terms and conditions apply.

View the table of contents for this issue, or go to the journal homepage for more

Home Search Collections Journals About Contact us My IOPscience

IOP PUBLISHING PHYSICS IN MEDICINE AND BIOLOGY

Phys. Med. Biol. 56 (2011) 2873–2882 doi:10.1088/0031-9155/56/9/015

A temperature-dependent gain control system forimproving the stability of Si-PM-based PET systems

Seiichi Yamamoto1,4, Junkichi Satomi1, Tadashi Watabe2,Hiroshi Watabe3, Yasukazu Kanai3, Masao Imaizumi2,Eku Shimosegawa2 and Jun Hatazawa2,3

1 Kobe City College of Technology, Kobe, Japan2 Department of Nuclear Medicine and Tracer Kinetics, Osaka University, Graduate School ofMedicine, Osaka, Japan3 Department of Molecular Imaging in Medicine, Osaka University Graduate School ofMedicine, Osaka, Japan

E-mail: [email protected]

Received 29 December 2010, in final form 3 March 2011Published 8 April 2011Online at stacks.iop.org/PMB/56/2873

AbstractThe silicon-photomultiplier (Si-PM) is a promising photodetector for thedevelopment of new PET systems due to its small size, high gain and relativelylow sensitivity to the static magnetic field. One drawback of the Si-PM is that ithas significant temperature-dependent gain that poses a problem for the stabilityof the Si-PM-based PET system. To reduce this problem, we developed andtested a temperature-dependent gain control system for the Si-PM-based PETsystem. The system consists of a thermometer, analog-to-digital converter,personal computer, digital-to-analog converter and variable gain amplifiers inthe weight summing board of the PET system. Temperature characteristics ofthe Si-PM array are measured and the calculated correction factor is sent tothe variable gain amplifier. Without this correction, the temperature-dependentpeak channel shifts of the block detector were −55% from 20 ◦C to 35 ◦C.With the correction, the peak channel variations were corrected within ±8%.The coincidence count rate of the Si-PM-based PET system was measuredusing a Na-22 point source while monitoring the room temperature. Withoutthe correction, the count rate inversely changed with the room temperatureby 10% for 1.5◦ C temperature changes. With the correction, the count ratevariation was reduced to within 3.7%. These results indicate that the developedtemperature-dependent gain control system can contribute to improving thestability of Si-PM-based PET systems.

(Some figures in this article are in colour only in the electronic version)

4 Author to whom any correspondence should be addressed.

0031-9155/11/092873+10$33.00 © 2011 Institute of Physics and Engineering in Medicine Printed in the UK 2873

2874 S Yamamoto et al

1. Introduction

The silicon-photomultiplier (Si-PM) is a recently developed photodetector that is also calleda Geiger-mode avalanche photodiode (G-APD), a multi-pixel photon counter (MPPC), ora solid-state photon counter (SSPC) (Ottea et al 2005, Finocchiaro et al 2009, Hong et al2008). Si-PM is a promising photodetector for PET, especially for its possible use in magneticresonance imaging (MRI) systems because it is less sensitive to static magnetic fields. Sincethe gain of the Si-PM is much higher than an avalanche photodiode (APD), it is easy to use. Inaddition, its timing property combined with a fast scintillator is excellent (Schaart et al 2010),and some block detector designs have already been reported (Schaart et al 2009, Kolb et al2010, Llosaa et al 2009a, 2009b, Henseler et al 2009).

We have recently developed a Si-PM-based depth-of-interaction (DOI) PET system forsmall animals (Yamamoto et al 2010). It employs 16 detector blocks consisting of 4 × 4Si-PM arrays and 2 types of LGSO scintillators. The performance of the Si-PM-based PEThas a spatial resolution of 1.6 mm FWHM and sensitivity of 0.6%. High-resolution mouseand rat images were successfully obtained using the PET system.

A drawback of the Si-PM is that it has significant temperature-dependent gain whichposes a problem for the stability of the Si-PM-based PET system. The room temperaturechanges produce an energy photo-peak shift, thus changing the sensitivity.

One solution for minimizing this problem is to keep the temperature of the detector blocksof the Si-PM-based PET system at the same level by such means as active circulation of theair. However, this type of system is bulky, noisy and expensive. If we develop a temperature-dependent gain control system for a Si-PM, these drawbacks will be reduced and it willcontribute to improving the stability of Si-PM-based PET systems. For Si-PM-based radiationdetectors, temperature-dependent gain control was reported for a Cherenkov light detectorby controlling the Si-PM bias voltage (Kim et al 2009, Marrocchesi et al 2009, Anderhubet al 2011). However, no temperature-dependent gain compensation system for PET has beenreported. We developed a temperature-dependent gain control system for the Si-PM arrayblock detectors using another approach, variable gain amplifiers to compensate for the changein Si-PM gain because the approach is easier for us to achieve. We implemented the gaincontrol system in our Si-PM PET and evaluated the stability improvement of the PET system.

2. Materials and methods

2.1. Measurements of temperature dependence of the Si-PM array

2.1.1. Si-PM array used for the experiments. The Si-PM array used for the experimentswas made by Hamamatsu Photonics, type S11064-025P (figure 1). This Si-PM array consistsof 4 × 4, 3 mm square channels. Each channel of the Si-PM contains 25 μm x 25 μm and14 400 pixels with an aperture ratio of 30.8%. The supply voltage is 60–80 V, and the outsidediameter of the array is 16 mm x 18 mm (Hamamatsu Photonics data sheet). The 16 outputsof the Si-PM array were fed to a 16-channel amplifier board by 16 small coaxial cables.

2.1.2. Block detector used for the measurement of the temperature characteristics of the Si-PMarray. To measure the temperature characteristics of the Si-PM array, a Si-PM array-basedblock detector was made. Gadolinium oxyorthosilicate (GSO) was selected for the scintillatorsdue to its small temperature characteristics (Utsu and Akiyama 1991). GSOs (0.5 mol% Ce:1.9 mm × 1.9 mm × 9 mm) were arranged in a 4 × 4 matrix and optically coupled one toone to the center of each pixel of the Si-PM array. All surfaces of the GSO scintillators were

A temperature-dependent gain control system for Si-PM PET 2875

Figure 1. Hamamatsu Photonics Si-PM array (S11064-025P) used for experiments set on a printedboard with output coaxial cables.

chemically etched. These GSO scintillators were wrapped in Teflon tape ∼10 times (reflectorthickness: ∼1 mm) and combined to form a GSO block. The difference in the light outputamong the GSO scintillators was less than ±3%.

The GSO block was optically coupled to the Si-PM array by a silicon compound (Sin-Etsu Silicon, KE 420, Tokyo, Japan). Because the center-to-center spacing between theGSO scintillators was almost the same as that of the channels of the Si-PM array, each GSOscintillator could be optically coupled to the center of each channel of the Si-PM.

2.1.3. Measurement of the temperature characteristics of the Si-PM array. The GSO blockdetector was encased in a black box and the temperature was increased from 20 ◦C to 40 ◦C.Following the natural decrease of the temperature in the black box, position histograms of theGSO block detector were acquired at several temperatures. A Cs-137 point source was usedfor the gamma source. Photo-peak channels for all 16 GSOs were calculated as a functionof temperature. After averaging the curve, a correction factor of the linear function as roomtemperature was calculated for the correction factor of the variable gain amplifiers.

2.2. Temperature-dependent gain control system of the Si-PM-based PET

2.2.1. Temperature-dependent gain control system. Figure 2 shows a schematic drawing ofthe temperature-dependent gain control system of the Si-PM-based PET system. The signalsfrom the Si-PM array are fed to the weight-summing amplifiers (W Sum) and to the variablegain amplifiers (Variable Gain AMP), where the correction factor for compensating the gainvariations of the Si-PM is set.

The thermometer installed outside the weight-summing board measures the temperaturearound the Si-PM array. The analog signal form the thermometer is fed to the analog-to-digitalconverter (ADC) and then to the personal computer (PC). The PC calculates the correctionfactor for the variable gain amplifier to compensate the temperature-dependent gain of theSi-PM array, and the correction factor is converted to an analog signal by the digital-to-analogconverter (DAC), set to the variable gain amplifiers. The compensated analog pulses are thenfed to the data acquisition system of the PET system (Yamamoto et al 2007).

2876 S Yamamoto et al

Figure 2. Schematic drawing of the temperature-dependent gain control system.

For the thermometer, a thermocoupler (TZ-1CA-K22, Asahi Keiki Co., Tokyo, Japan)was selected. For the A-D and D-A converters, a USB-based interface containing both A-D and D-A converters (CONTEC. AIO-16082AY-USB, Osaka, Japan) was selected. Thecorrection factor was transferred to the variable gain amplifier in the weight-summing boardevery 1 s.

2.2.2. Temperature-dependent gain control system for the Si-PM PET system. Figure 3shows a photograph of the temperature-dependent gain compensation system for the Si-PM-based PET system. The sensing point of the thermocoupler is positioned at the side of thedetector ring to measure the nearest temperature of the Si-PM arrays used for the PET system.The tip of the thermo-coupler was positioned inside the detector ring holder to minimize thedirect air blowing from such sources as air conditioners.

2.3. Performance measurements of the temperature-dependent gain control system

2.3.1. Performance measurements of the temperature-dependent gain control system for theGSO block detector. To evaluate the performance of the temperature-dependent gain controlsystem for the block detector, the corrected temperature characteristics were measured usingthe same GSO block detector used for the temperature characteristic. The GSO block detectorwas encased in a black box, and the temperature was increased from 20 ◦C to 40 ◦C Afterfollowing the temperature changes in the black box, position histograms were acquired andphoto-peak channels were calculated as a function of temperature in the black box.

2.3.2. Performance measurements of the temperature-dependent gain control system for theSi-PM-based PET system. To evaluate the performance of the temperature-dependent gaincontrol system for the PET system, the count rates for PET were monitored for 8 h with andwithout the temperature-dependent gain control system using a Na-22 point source positionedat the center of the axial field of view of the PET system. Prompt, delayed and prompt-minus-delayed count rates were monitored every 1 s. Furthermore, temperature during theacquisition was also monitored every 1 s.

A temperature-dependent gain control system for Si-PM PET 2877

Figure 3. Temperature-dependent gain control system combined with the Si-PM PET system.

3. Results

3.1. Temperature dependence of the Si-PM array

3.1.1. Temperature characteristic of the Si-PM array. The temperature characteristics of theSi-PM array investigated for all channels measured using the GSO block detector are shownin figure 4. Although there is some gain variation among the channels of the Si-PM array, thetemperature characteristics of all channels were similar. The temperature characteristics ofSi-PM ranged from −3.4 to 3.8% /◦C.

The normalized temperature characteristics were then averaged for all channels relativeto the gain at 20 ◦C (figure 5). The averaged temperature characteristic was −3.6% / ◦C. Thistemperature characteristic is similar to that measured by Knob et al (2010). The average gainreduction in these temperature ranges (from 20 ◦C to 35 ◦C) was −55%.

3.1.2. Correction factor for the temperature-dependent gain control system. The correctionfactor derived from the normalized temperature-dependent gain is shown in figure 6. Fromthe graph, the following equation is derived by linear approximation:

y = 0.08x−0.63, (1)

where x is the temperature (◦C) measured by the thermometer and y is the correction factorfor the variable gain amplifiers.

3.2. Performance measurements of the temperature-dependent gain control system

3.2.1. Performance measurements of the temperature-dependent gain control system forthe GSO block detector. The temperature-dependent gain variation of all channels andthe normalized temperature-dependent average gain of the Si-PM array-based GSO blockdetector after correction are shown in figure 7(A) and (B), respectively. The gain variation

2878 S Yamamoto et al

Figure 4. Temperature-dependent gain variations of all channels of the Si-PM array-based blockdetector.

Figure 5. Normalized temperature characteristics averaged for all channels of the Si-PM array.

after correction was effective for all channels and the averaged variation was ±8% from20 ◦C to 35 ◦C.

3.2.2. Performance measurements of the temperature-dependent gain control system for theSi-PM-based PET system. Coincidence count rates (prompt-minus-delayed) of the Si-PMPET system and temperature changes for 8 h without and with the temperature-dependent gaincompensation system are shown in figure 8(A) and (B), respectively. Without the correction,

A temperature-dependent gain control system for Si-PM PET 2879

Figure 6. Averaged correction factor for the Si-PM array.

(A) (B)

Figure 7. Temperature-dependent gain variation of 16 channels of the Si-PM block detector (A)and normalized temperature characteristics averaged for all channels of the Si-PM array (B) aftercorrection.

the count rate inversely changed with room temperature by 10% for 1.5 ◦C temperaturechanges. With the correction, the count rate variation was reduced to be within 3.7%.

4. Discussion

We successfully developed a temperature-dependent gain control system for a Si-PM-basedPET system. Without the correction, the coincidence count rate (prompt-minus-delay) changed∼10% for 1.5 ◦C temperature changes. For quantitative measurement in PET studies, 10%sensitivity change is not acceptable, and 1.5 ◦C is not an extreme temperature change.

2880 S Yamamoto et al

(A) (B)

Figure 8. Count rates and temperature changes of the Si-PM PET system for 8 h without (A) andwith the temperature-dependent gain compensation system (B).

Consequently, quantitative measurement of PET will be difficult without a temperature-dependent gain compensation system.

The photo-peak shifts due to the temperature change also contribute to the sensitivityvariation of the detector blocks. If the temperature during normalization measurement andanimal studies is different, the normalization data will not accurately correct the detector’ssensitivity, resulting in the increased noise in the reconstructed images. Consequently, thedeveloped temperature-dependent gain compensation system may help to keep the signal-to-noise ratio (S/N) in the reconstructed images.

In the system, a single correction factor was applied for all channels of the Si-PM array.If we apply the correction factors to each channel of the Si-PM array independently, thecorrection accuracy may be improved. However, the temperature characteristics of the Si-PMarray are similar among all channels as shown in 4. Therefore, the single correction factor isadequate for the Si-PM array.

The results in this paper showed the precision of the correction over 20 ◦C However, thedeveloped temperature-dependent gain compensation system can be used lower than 20 ◦Cbecause the correction factor is calculated by linear approximation. Such cooled conditionwill provide an advantage in higher S/N ratio for Si-PM-based detectors.

The system also used a single thermometer for monitoring the temperature. If weuse multiple thermometers, the correction accuracy will be improved by compensating theposition-dependent temperature differences. However, the cost of the system would be higherand the system would be more complex with so many temperature sensors.

The temperature-dependent gain control system was effective for relatively slowtemperature change as shown in figure 8. One difficulty for this temperature compensationsystem is the temporal differences of the temperatures between the Si-PM-based blockdetectors and temperature sensor. Since the Si-PM arrays are coupled to LGSO blocks,the temperature of Si-PM arrays changes relatively slow while the temperature change of thethermometer is usually quick when it is positioned in the air without any surrounding material.Consequently, the position and surrounding material of the thermometer is important and sometemporal smoothing of the temperature data may improve the precision of the correction inrelatively short temperature changes.

A temperature-dependent gain control system for Si-PM PET 2881

Another possible method for the gain control system is changing the bias voltage of theSi-PM-based block detectors. Compared to controlling the bias voltage approach with ourvariable gain amplifier approach, the former system needs a precise computer-controllablevoltage supply which is more expensive. Also controlling the bias voltage has a risk ofaccidentally supplying higher voltage than absolute maximum rating of the sensors or otherelectronics parts. In our approach, we do not need to consider these risks. The disadvantageof the variable gain amplifier-based approach is that high speed variable gain amplifiers areneeded in the electronics circuit.

Although the peak channels can be compensated with the developed system, it is notpossible to correct the performance degradation due to the increased noise of the Si-PMarray-based block detector at high temperature. As the temperature increases, the noise of theSi-PM array also increases and thus may degrade the energy, spatial and timing resolutionsof the block detectors, because the breakdown voltage of the Si-PM has linear relationwith temperature (Hamamatsu Photonics data sheet) and the noises also increase with thetemperature. Consequently, it is better to use the Si-PM-based PET system in a temperature-controlled room even when the temperature-dependent gain compensation system is adapted.However, even in a temperature-controlled condition, the developed temperature-dependentgain compensation system helps to improve the system stability and it would be especiallyvaluable during accidental events like air conditioner problems in a room.

5. Conclusion

Our developed temperature-dependent gain control system successfully compensated thephoto-peak shifts of a Si-PM array-based block detector and changes in the coincidence countrate of the Si-PM-based PET system. We conclude that the developed temperature-dependentgain control system will contribute to the stability of Si-PM-based PET systems.

Acknowledgments

This work was partly supported by the Ministry of Education, Science, Sports and Culture,Japan and the National Institute of Biomedical Innovation, Osaka, Japan.

References

Anderhub H et al 2011 A G-APD based camera for imaging atmospheric Cherenkov telescopes Nucl. Instrum.Methods A 628 107–10

Finocchiaro P, Pappalardo A, Cosentino L, Belluso M, Billotta S, Bonanno G and Mauro S D 2009 Features of siliconphoto-multipliers: precision measurements of noise, cross-talk, after pulsing, detection efficiency IEEE TransNucl. Sci. 56 1033–41

Hamamtsu Photonics, MPPC S11064 data sheetHenseler D, Grazioso R, Nan Z and Schmand M 2009 SiPM performance in PET applications: an experimental and

theoretical analysis IEEE Nuclear Science Symp. Conf. Record pp 1941–8Hong S J, Song I C, Ito M, Kwon S I, Lee G S, Sim K-S, Park K S, Rhee J T and Lee J S 2008 An investigation into

the use of Geiger-mode solid-state photomultipliers for simultaneous PET and MRI application IEEE Trans.Nucl. Sci. 55 882–88

Kim M Y et al 2009 Scintillation and Cherenkov light detection with a 3 mm × 3 mm silicon photomultiplier Nucl.Phys. B 197 325–30

Kolb A, Lorenz E, Judenhofer M S, Renker D, Lankes K and Pichler B J 2010 Evaluation of Geiger-mode APDs forPET block detector designs Phys. Med. Biol. 55 1815–32

Llosaa G et al 2009a Energy, timing and position resolution studies with 16-pixel silicon photomultiplier matrices forsmall animal PET IEEE Trans. Nucl. Sci. 56 2586–93

2882 S Yamamoto et al

Llosaa G et al 2009b First results in the application of silicon photomultiplier matrices to small animal PET Nucl.Instrum. Methods A 610 196–9

Marrocchesi P S et al 2009 Active control of the gain of a silicon photomultiplier Nucl. Instrum. Methods A 602 391–95Ottea A N et al 2005 A test of silicon photomultipliers as readout for PET Nucl. Instrum. Methods A 545 705–15Schaart D R, Seifert S, Vinke R, van Dam H T, Dendooven P, Lohner H and Beekman F J 2010 LaBr(3):Ce and

SiPMs for time-of-flight PET: achieving 100 ps coincidence resolving time Phys. Med. Biol. 55 N179–89Schaart D R, van Dam H T, Seifert S, Vinke R, Dendooven P, Lohner H and Beekman F J 2009 A novel, SiPM-array-

based, monolithic scintillator detector for PET Phys. Med. Biol. 54 3501–12Utsu T and Akiyama S 1991 Growth and applications of Gd2SiO5: Ce scintillators J. Cryst. Growth 109 376–84Yamamoto S, Imaizumi M, Watabe T, Watabe H, Kanai Y, Shimosegawa E and Hatazawa J 2010 Development of a

Si-PM-based high-resolution PET system for small animals Phys. Med. Biol. 55 5817–31Yamamoto S, Mashino H, Kudo H, Matsumoto K and Senda M 2007 A dual layer GSO PET system for small animals

IFMBE Proc. 14 1712–15