Radiation Measurements

Transcription

1 Radiation Measurements 85 (2016) 111e115 Contents lists available at ScienceDirect Radiation Measurements journal homepage: Immersion cooling of silicon photomultipliers (SiPM) for nuclear medicine imaging applications R.R. Raylman *, A.V. Stolin Center for Advanced Imaging, Department of Radiology, One Medical Center Dr., Box 9236, West Virginia University, Morgantown, WV 26506, United States highlights Immersion cooling is new, simple and inexpensive method to cool solid state based nuclear medicine scanner. Method successfully tested on a scaled version of an SiPM-based PET detector module. Can be scaled up to cool a complete PET scanner. article info abstract Article history: Received 6 November 2015 Received in revised form 17 December 2015 Accepted 29 December 2015 Available online 1 January 2016 Keywords: Immersion cooling PET Silicon photomultipliers Nuclear medicine instrumentation Silicon photomultipliers (SiPM) are compact, high amplification light detection devices that have recently been incorporated into magnetic field-compatible positron emission tomography (PET) scanners. To take full advantage of these devices, it is preferable to cool them below room temperature. Most current methods are limited to the cooling of individual detector modules, increasing complexity and cost of scanners made-up of a large number of modules. In this work we investigated a new method of cooling, immersion of the detector modules in non-electrically conductive, cooled liquid. A small-scale prototype system was constructed to cool a relatively large area SiPM-based, scintillator detector module by immersing it in a circulating bath of mineral oil. Testing demonstrated that the system rapidly decreased and stabilized the temperature of the device. Operation of the detector illustrated the expected benefits of cooling, with no apparent degradation of performance attributable to immersion in fluid Elsevier Ltd. All rights reserved. 1. Introduction The combination of MRI and PET systems has received substantial interest over the last two decades. Although this combination was first proposed in the mid-1990s (Christensen et al., 1995; Raylman et al., 1996), it wasn't until the introduction of arrays of avalanche photodiodes (APD) that practical scanners were constructed (Pichler et al., 2006; Grazioso et al., 2006). The development of MR-suitable PET scanners has been advanced by the introduction of arrays of silicon photomultipliers (SiPM), which possess signal gains comparable to photomultiplier tubes (on the order of ), and have the same insensitivity to magnetic fields as APDs (Roncali and Cherry, 2011). A number of investigators have created MR-compatible PET detector modules from which MRI/PET scanners can be constructed (Chagani et al., 2009, Schaart et al., * Corresponding author. Department of Radiology One Medical Center Dr., Box 9236, HSC West Virginia University, Morgantown, WV , United States. address: rraylman@wvu.edu (R.R. Raylman). 2009, Yamamoto et al., 2010, Llosa et al., 2011, Schulz et al., 2011, Zorzi et al., 2011; Wang et al., 2012, Yoon et al., 2012). While each of these efforts produced good PET detector modules, they did not take full advantage of the potential performance of SiPMs by cooling them. As with most solid-state devices, the performance of SiPMs is affected by temperature. Lowering their temperature reduces thermal (dark current) noise and increases gain. The gain increase is due to the reduction of the breakdown voltage and thus overvoltage at lower temperatures (assuming a constant bias voltage). A number of cooling methods are suitable for application with SiPMs. Perhaps the most common method is use of Peltier coolers (Sokolov et al., 2002), which transfer heat from one of the device's surfaces to the opposite surface. In applications to MR-compatible PET detectors, the presence of these devices can interfere with MR imaging. The strong magnetic fields present in an MR scanner also may affect operation of the Peltier devices. Furthermore, a secondary cooling system is required to remove heat transferred from the cooling surface, so additional hardware is required. Finally, / 2016 Elsevier Ltd. All rights reserved.

2 112 R.R. Raylman, A.V. Stolin / Radiation Measurements 85 (2016) 111e115 these devices are usually mounted on the backside of the SiPM arrays, thus the cooling of the individual SiPMs is indirect, relying on heat transfer from the SiPM, through the PCB mounting board, to the cooling surface of the Peltier cooler. Other cooling methods utilize the circulation of refrigerated liquid. For example, Raylman et al. (Raylman et al., 2014) created an MR-compatible PET detector by encasing the SiPMs and readout electronics in a copper jacket cooled by circulating fluid. Cooling jackets contact the SiPM arrays at their edges, so they are cooled indirectly. The jackets draw heat through the periphery of the devices, eventually cooling the entire array. A more direct method for reducing the temperature of SiPM arrays is to utilize a cooled light guide (Stolin et al.,2013). In this design, the solid acrylic light guide used to spread scintillation light among numerous SiPMs to enhance event positioning, is replaced by a light guide consisting of a thin film of optically transparent, cooled liquid. Therefore, heat is drawn directly from the SiPMs. Due to the complexity of maintaining a uniform circulation of fluid within the light guide, this method is limited to use with relatively small, planar detectors. In this work, we investigated a new method for reducing the temperature of SiPMs, immersion cooling. As the name indicates, this method entails submersion of the detector (scintillator, SiPM array and readout electronics) in a cooled, electrically non-conductive liquid. The potential advantages of immersion cooling compared to other techniques include the ability to cool any size or shape detector, its simplicity and relatively low cost. Delrin plastic was placed around the scintillator array (see Fig. 1). To make the unit light tight, it was wrapped in thin vinyl tape. A temperature sensor was mounted on the scintillator array to monitor its temperature. Fig. 2 shows the complete detector module. In order to reduce the number of output analog channels, multiplexing electronics were used. These devices consisted of an Anger-type readout, which reduceed the sixty-four output channels from the SB-8 to four channels (two X- and two Y-directional channels) (AiT Instruments, Newport News, VA). Diode coupling of SiPM signals was implemented to minimize the effects of capacitance. The readout electronics were connected to a control module (AiT Model SiPM-IM16). This unit houseed an adjustable bias voltage supply and analog output amplifiers. The module also produced a sum of the amplitudes of all of the analog signals, which is a measure of the total amount of energy deposited in the scintillator. The summed output was discriminated with a Philips Scientific 6915 constant-fraction discriminator. The logical pulse produced by the discriminator initiates digitization of the analog signals from the readout electronics by an FPGA-based analog-to- 2. Material and methods The goal of this project was to demonstrate the ability of immersion cooling to effectively reduce and stabilize the temperature of scintillation detectors, enhancing their performance. Specifically, a prototype cooling system was constructed and used in conjunction with an SiPM-based PET detector module PET detector module The PET detector module consisted of a array of 1 mm 1 mm 10 mm LYSO detector elements (pitch ¼ 1.07 mm) coupled to an array of SiPMs (SB-8, SensL Technologies, LTD, Cork, Ireland) through a 2.7 mm-thick piece of ultraviolet transparent acrylic. This light guide facilitated spread of the scintillation light from each detector element to a number of SiPMs, enabling accurate determination of event position in the scintillator array. The SensL SB-8 consists of an 8 by 8 matrix of 6 mm 6 mm individual SiPMs. Each SiPM contains 18, mm cells resulting in a packing fraction of 64% and a total active area of 52 mm 52 mm. For all measurements, the bias voltage was 28.8 V. Since the scintillator is much smaller than the SiPM array, a spacer constructed from Fig. 2. Picture showing the complete PET module (detector was wrapped in vinyl tape to make the unit light tight). The ribbon cable carries the analog signals from the readout electronics to the control module. The other cable in the figure connects the temperature sensor to its control unit. Fig. 1. PET detector module construction. (a) Left-to-right: SensL SB-8 SiPM array and light guide, Delrin scintillator spacer and LYSO scintillator array. (b) Top view of the assembled detector.

3 R.R. Raylman, A.V. Stolin / Radiation Measurements 85 (2016) 111e digital conversion (ADC) module developed by AiT Instruments. The digitized data were transmitted to the control computer via a USB-2 connection to the acquisition control computer. Data acquisition and processing were controlled by custom Java-based software, with an interface created with the Kmax Scientific software package (Sparrow Corp, Port Orange, FL). Fig. 3 shows a schematic of the data acquisition electronics Detector cooling system The closed-loop, prototype, cooling system consisted of a fluid reservoir, pulsating pump (Pulsatron Series E, Pulsafeeder, Punta Gorda, FL), cooler and immersion tank. The reservoir had a horizontal copper plate with two other copper plates attached vertically to cool the liquid. The bottom copper plate was in thermal contact with the cold side of a Peltier device (60 W, Vktech, Inc., Korea, TEC ). The cooler incorporated two 60 W Peltier devices coupled to a two-piece aluminum block. A snake-like pattern was machined inside the aluminum block through which fluid transited and was cooled prior to exiting the unit. Barbed fittings were attached to entry and exit holes in the block to permit attachment of tubing (see Fig. 4). Finally, the fluid circulated through an insulated immersion tank that contained the detector module (see Fig. 5). Fig. 4. Picture showing the apparatus used assess the effectiveness of immersioncooled radiation detectors Temperature measurements To assess the efficacy of immersive cooling of a SiPM-based PET detector, the temperature of the scintillator was measured as a function of time. Specifically, the system shown in Fig. 4 was operated as the temperature was monitored over a period of approximately 4 h Detector module testing The effect of cooling on the SiPM array was demonstrated by acquiring maps of scintillation light as a function of position at room temperature and at the lowest attainable temperature (12.6 C). Six 22 Na sources (3.5 mci each) were positioned above the detector and data acquired using the system shown in Fig. 3. Focal areas of signal in these maps, which indicate positions of scintillator elements, were first automatically identified and then manually corrected. A set of Voronoi diagrams was constructed using detector element positions as vertices. Boundaries of the diagrams were defined as the boundaries of individual crystals. The ability to identify individual detector elements was assessed by calculating the resolvability index (RI) (Stolin et al., 2014): RI ¼ GM FWHM : (1) D GM-FWHM is the geometric mean of the full width at half maximum of two-dimensional Gaussian function fits of individual element event distributions averaged over all of the elements and D is the average distance between adjacent detector elements in the scintillator array. Fig. 5. Picture of the detector module in the insulated immersion tank. The top of the readout electronics board, in addition to connections to the temperature sensor and readout control unit, are shown in the picture. Energy spectra were obtained from the data acquired to map the scintillation light distribution described above. Prior to processing, the spectra were corrected for output non-linearity produced by Fig. 3. Schematic of the electronics used to acquire data.

4 114 R.R. Raylman, A.V. Stolin / Radiation Measurements 85 (2016) 111e115 signal clipping in the readout electronics caused by use of diodes in the circuitry. These correction factors were calculated by applying a Gaussian function fit to the photo-peaks in the energy spectra ( 22 Na emits 511 kev and 1274 kev photons). Linear correction factors were calculated to align the known photo-peak energies with the measured energies. A range of ±20% of a photo-peak position was chosen as the fitting interval. The ratio of FWHM of the fit curves to photo-peak energy is reported as energy resolution. 3. Results Fig. 6 shows the results of the cooling-timing test. It took approximately 2 h for the temperature of the detector to be reduced from room temperature (22 C) to the coldest temperature achievable with the current version of the system (12.6 C). This temperature remained constant (±0.1 C) for the rest of the experiment (an additional 2 h). Fig. 7 shows event distribution maps acquired with and without cooling. The RI for the un-cooled detector is ± (it was possible to fit 494 of 576 (86%) peaks present in the crystal map to a 2D-Gaussian function). For the cooled detector, the RI is ± (549 out 576 (95%) could be fit to a 2D-Gaussian function). Fit failures were not included in the calculation of RI or its standard deviation. Note that a smaller RI indicates more precise localization of detector elements compared to maps with larger RIs. Fit failures are mostly caused by the inability to separate signals from adjacent detector elements, which occurred mainly at the edges of the scintillator array where the shape of the light pulses are non-gausssian. Non-uniform light distribution is improved somewhat by increased signal amplitudes and reduced noise achieved by cooling the SiPMs, thus increasing the number of peaks successfully fit to a 2D-Gaussian function. Improved separation of the light distribution achieved by cooling the SiPMs is illustrated by the intensity profiles shown in Fig. 8(a). Fig. 8(b) shows plots of the first three elements of the profiles shown in Fig. 8(a). The energy resolution for the un-cooled detector is 14.2 ± 0.4% versus 13.8 ± 0.3% for the cooled detector. 4. Discussion Cooling is an effective method for enhancing and stabilizing the Fig. 6. Plot of detector temperature as a function of cooling time. The curve was fit with an exponential function to guide the eye. performance of solid-state light detection devices, such as SiPMs, often achieved by utilizing Peltier coolers. In this investigation we studied another method, immersion cooling. Immersion cooling is known mostly in the high performance-computing field (Bergles and Bar-Cohen, 1994). To assess the potential application of immersion cooling to scintillator-based, radiation detectors, a smallscale prototype was constructed and tested. Fig. 6 shows a plot of temperature of the scintillator as a function of time from initiation of cooling. It required approximately 2 h to reduce and stabilize the temperature of the detector from room temperature to the minimum temperature of 12.6 C, which was maintained for the remaining 2 h of the experiment. The temperature limit is determined by the power of the Peltier coolers, effectiveness of the thermal insulation and the power dissipated by the detector module (~1.2 W). A lower temperature could be achieved by increasing the cooling power of the system by replacing the improvised system used in this investigation with a desktop circulation cooler, which can cool the fluid over a wide range of temperatures. Thus, it would be possible to optimize temperature based on the characteristics of the type of SiPM and application without any additional cost (other than the expense of the cooler). The shape of this curve is dependent upon a number of factors, including the cooling power of the system, insulation of the immersion tank, pumping rate, flow rate of fluid through the tank (5 ml/min) and heat conductivity of the fluid (0.162 W m 1 K 1 for mineral oil). The event distribution maps shown in Fig. 7 demonstrate, first, that it is indeed possible to operate an SiPM-based, scintillation detector immersed in mineral. While this finding is not unexpected, it is nonetheless important to confirm this ability. Furthermore, these maps illustrate the effect of cooling on the ability to identify the individual elements in the scintillator array. The focal areas in the maps appear more localized, with less background noise, in the cooled versus room temperature measurements. This observation is supported by the reduction of the resolvability index (RI) and increase in the number of Gaussianshaped (non-distorted) peaks for the cooled detector compared to the un-cooled detector. The profiles in Fig. 8(a) illustrate the ways in which the intensity distribution maps (Fig. 7) are changed by cooling. Specifically, the peak-to-valley distances are increased and the positions of the peaks are shifted when the SiPMs are cooled. Shifting of peak positions is due to increase in SiPM array gains, increasing the amplitude of their outputs and consequently altering results of the calculations used to determine the locations of events in the scintillator array. Note, that a shift in the position of the peaks is not important as long as a new pixel mapping procedure is performed prior to use. The plots in Fig. 8(b) demonstrate the improvement of separation of the light signals originating from elements at the edges of the scintillator array when the device is cooled. This phenomenon is the reason for the higher number of successful 2D-Gaussian fits and lower error in RIs reported for the cooled detector. In summary, this investigation demonstrated the successful application of immersion cooling with scintillation detectors. This method has some potential advantages compared other current techniques. For example, immersion cooling is relatively simple to apply to any scanner size and geometry. Current cooling systems are designed to cool individual detector modules within a scanner. Since there are often hundreds of detectors in a scanner, this approach can rapidly become cumbersome and expensive. By replacing discrete coolers (Peltier devices, for example) with a single system by immersing all of the scanner's detectors in cooling fluid, complexity and cost can be reduced. Introduction of fluid into a scanner gantry would require the addition of leak tight enclosures, inclusions of penetrations to permit an in- and out-flow of

5 R.R. Raylman, A.V. Stolin / Radiation Measurements 85 (2016) 111e Fig. 7. (a) Event distribution map acquired from the detector at room temperature (22 C). (b) Event distribution map acquired from the cooled detector (12.6 C). Fig. 8. (a) Representative profiles from the maps shown in Fig. 7 of the cooled and un-cooled detector. (b) Expanded part of the plot shown in (a) illustrating the changes in light distribution maps at the edges of the scintillator array (highlighted by the arrow) produced by cooling. cooling liquid and a chiller/pump. It is important to note that while mineral oil was used in this investigation, it is possible to use virtually any non-electrically conductive fluid for immersive cooling of radiation detectors. The next step in development of the immersion cooling system is to scale it for use with a small, preclinical PET scanner. References Bergles, A.E., Bar-Cohen, A., Immersion cooling of digital computers. Cool. Electron. Sys. Proceed. NATO Adv. Study Inst. 539e621. Chagani, H., Dolenec, R., Korpar, S., Krizan, P., Pestotnik, R., Stanovnik, A., Verheyden, R., Tests of silicon photomultiplier PET modules. IEEE Nucl. Sci. Symp. Med. Imag. Conf. Rec. 1518e1520. Christensen, N.L., Hammer, B.E., Heil, B.G., Fetterly, K., Positron emission tomography within a magnetic field using photomultiplier tubes and lightguides. Phys. Med. Biol. 40, 691e697. Grazioso, R., Hang, N., Corbeil, J., Schmand, M., Ladebeck, R., Vester, M., Schnur, G., Renz, Fischer, H., APD-based PET detector for simultaneous PET/MR imaging. Nucl. Instrum. Meth. A 569, 301e305. Llosa, G., Barrio, J., Cabello, J., Lacasta, C., Oliver, J.F., Rafecas, M., Solaz, C., Barrillon, P., de La Taille, C., Bisogni, M.G., Del Guerra, A., Piemonte, C., Development of a PET prototype with continuous LYSO crystals and monolithic SiPM matrices. IEEE Nucl. Sci. Symp. Med. Imag. Conf. Rec. 3631e3634. Pichler, B.J., Judendofer, M.S., Catana, C., Walton, J.H., Kneiling, M., Nutt, R.E., Siegel, S.M., Claussen, C.D., Cherry, S.R., Performance test of an LSO-APD detector in a 7-T scanner for simultaneous PET/MRI. J. Nucl. Med. 47, 639e647. Raylman, R.R., Hammer, B.E., Christensen, N.L., Combined MRI-PET scanner: a Monte Carlo evaluation of the improvements in PET resolution due to the effects of a static homogeneous magnetic field. IEEE Trans. Nucl. Sci. 43, 2406e2412. Raylman, R.R., Stolin, A.V., Majewski, S., Proffitt, J., A large area, silicon photomultiplier-based PET detector module. Nucl. Instrum. Meth. A 735, 602e609. Roncali, E., Cherry, S.R., Application of silicon photomultipliers to positron emission tomography. Ann. Biomed. Engin. 39, 1358e1377. Schaart, D.R., van Dam, H.T., Seifert, S., Vinke, R., Dendooven, P., Lohner, H., Beekman, F.J., A novel, SiPM-array-based, monolithic scintillator detector for PET. Phys. Med. Biol. 54, 3501e3512. Schulz, V., Weissler, B., Gebhardt, P., Solf, T., Lerche, C.W., Fischer, P., Ritzert, M., Mlotok, V., Piemonte, C., Golschmidt, B., Vandenberghe, S., Salomon, A., Schaeffer, P.K., SiPM based preclinical PET/MR insert of a human 3T MR: first imaging experiments. IEEE Nucl. Sci. Symp. Med. Imag. Conf. Rec. 4467e4469. Sokolov, A., Gostilo, V., Loupilov, A., Zalinkevich, V., Performance improvement of Si(Li) Peltier cooled detectors. IEEE Trans. Nucl. Sci. 49, 2427e2430. Stolin, A.V., Majewski, S., Raylman, R.R., Novel method of temperature stabilization for SiPM-based detectors. IEEE Trans. Nucl. Sci. 60, 3181e3187. Stolin, A.V., Majewski, S., Jaliparthi, G., Raylman, R.R., Proffitt, J., Evaluation of imaging modules based on SensL array SB-8 for nuclear medicine applications. IEEE Trans. Nucl. Sci. 61, 2433e2438. Wang, Y., Zhang, Z., Li, Wang, B., Shuai, L., Feng, B., Chai, P., Liu, S., Tang, H., Li, T., Liao, Y., Huang, X., Chen, Y., Liu, Y., Zhang, Y., Long design and performance evaluation of a compact, large-area PET detector module based on silicon photomultipliers. Nucl. Instrum. Meth. A 670, 49e54. Yamamoto, S., Imaizumi, M., Watabe, T., Watabe, H., Kanai, Y., Shimosegawa, E., Hatazawa, J., Development of a Si-PM-based high-resolution PET system for small animals. Phys. Med. Biol. 55, 5817e5831. Yoon, H.S., Ko, G.B., Kwon, S.I., Lee, C.M., Ito, M., Chan Song, I., Lee, D.S., Hong, S.J., Lee, J.S., Initial results of simultaneous PET/MRI experiments with an MRIcompatible silicon photomultiplier PET scanner. J. Nucl. Med. 53, 608e614. Zorzi, N., Melchiorri, M., Piazza, A., Piemonte, C., Tarolli, A., Development of large-area silicon photomultiplier detectors for PET applications at FBK. Nucl. Instrum. Meth. A 636, S208eS213.