A dedicated high resolution PET imager for plant sciences
Qiang Wang, Aswin J. Mathews, Ke Li, Jie Wen, Sergey Komarov, Joseph A. O'Sullivan, Yuan-Chuan Tai
aa r X i v : . [ phy s i c s . i n s - d e t ] J a n A dedicated high resolution PET imager for plantsciences
Qiang Wang , Aswin J. Mathews , Ke Li , Jie Wen , SergeyKomarov , Joseph A. O’Sullivan and Yuan-Chuan Tai Department of Radiology, Washington University in St Louis, MO 63110, USA Department of Electrical and Systems Engineering, Washington University in StLouis, MO 63130, USAE-mail: [email protected],[email protected]
Abstract.
PET provides in vivo molecular and functional imaging capability that iscrucial to studying the interaction of plant with changing environments at the whole-plant level. We have developed a dedicated plant PET imager that features highspatial resolution, housed in a fully controlled environment provided by a plant growthchamber (PGC).The system currently contains two types of detector modules: 84 microPET R (cid:13) R4 blockdetectors with 2.2 mm crystals to provide a large detecting area; and 32 Inveon
T M blockdetectors with 1.5 mm crystals to provide higher spatial resolution. Outputs of thefour microPET R (cid:13) block detectors in a modular housing are concatenated by a customprinted circuit board to match the output characteristics of an Inveon T M detector.All the detectors are read out by QuickSilver
T M electronics. The detector modulesare configured to full rings with a 15 cm diameter trans-axial field of view (FOV) fordynamic tomographic imaging of small plants. Potentially, the Inveon
T M detectorscan be reconfigured to quarter-rings to get a 25 cm FOV using step-and-shoot motion.The imager contains 2 linear stages to position detectors at different heights for multi-bed scanning, and 2 rotation stages to collect coincidence events from all angles. Thedetector modules and mechanical components of the imager are housed inside a PGCthat regulates the environmental parameters.The PET system has been built and integrated into the PGC. The system has atypical energy resolution of 15% for Inveon
T M blocks and 24% for R4 blocks; timingresolution of 1.8 ns; and sensitivity of 1.3%, 1.4%, 3.0% measured at the center ofFOV, 5 cm off to R4 half-ring and 5 cm off to Inveon half-ring, respectively(with a350-650 KeV energy window and 3.1 ns timing window). System spatial resolution issimilar to that of commercial microPET R (cid:13) sytems, with 1.25 mm rod sources in themicro-Derenzo phantom resolved using ML-EM reconstruction algorithm. Preliminaryimaging experiments using soybean and wild type and mutant maize labeled with C-CO produced high-quality dynamic PET images that reveal the translocation anddistribution patterns of photoassimilates. Wang et al
1. Introduction
Rapid growing population brings in a fast increasing requirements on food, energy andother natural resources. Advances in molecular biology technique have made geneticmodified corps that are more resistant to biotic(like insects, virus, microorganisms)and abiotic stress (such as drought, temperature extremes, nutrient limitation and soon) widely available to improve food and energy production. Practical corps yieldand biomass growth is a complex phenotypic trait determined by the interactions of agenotype with the growth environment. Photonic-based techniques are widely appliedin plant phenomics[1], X-ray computed tomography(CT) and MRI scanners are usedto non-destructively image the inner structures of plant or its roots under soil[2][3].Besides the structural imaging tools, new physiological imaging methods is in greatdemands to rich the tool sets for future plant phenomics[4][5]. Short-lived radioisotopetechnique can provide data that are crucial for developing more precise models thatquantitatively link the underlying biochemical reactions to physiological response andbetter predictive models can be built to better understand the plant development andgrowth at the whole-body scope. This new method can also greatly save time and moneyfor delivering new products to the farm compared with conventional field trails.PET is a noninvasive functional and molecular imaging technique that can providequantitative information of dynamic radio-tracer distribution with spatial resolution inthe order of millimeter. It is already commonly used to diagnosis human disease[6] andhigh spatial resolution scanners with small crystal size are also widely applied for smallanimal study[7]. Using PET detector modules to collect coincidence event counts ofdifferent part of plant is a conventional method at early time[8]. Then large planardetector modules with larger detection area are used to acquire projection images ofinteresting plant sections[9]. Commercial PET scanner has been used by some researchgroups for plant imaging[10]. Most of currently built PET scanners dedicated for plantimaging are based on these small crystal size PET detector modules which are usuallyadopted from small animal PET research projects. The Japanese group built a radio-tracer imaging system based on two large area planar detector head(120.8 x 186.8 mm )to acquire 2D dynamic projection PET images[11]. The detector head is composed of 4 x6 detector modules each composed of 10 x 10 array of 2 x 2 x 20 mm BGO scintillators.The German research group utilize 8 small animal PET detector modules developed bythe ClearPET
T M project[12] to form two partial ring detector sets and achieve a 10.1cm FOV. Completed data sets for 3D tomographic image reconstruction are acquired byrotational motion of the detector sets mounted on a rotation table[13]. The plant PETscanner under development at Brookhaven National Lab is based on the RatCAP PETproject[14] with larger scanner bore size(100 mm in diameter and 18 mm in height) withmore detector modules[15]. The proposed Jefferson National Lab’s PhytoPET scannerfeatures modular design concept to achieve re-configurable system geometry. The basicbuilding module is a 48 x 48 array of 1.0 x 1.0 x 10 mm LYSO crystals coupled to the5 cm x 5 cm Hamamatsu H8500 position sensitive photomultiplier(PS-PMT)[16].
Wang et al C, N, O) to many days using long half-life isotopes( Cu, Na), as a result, scanner shouldprovide high sensitivity and low noise level; most positrons will escape from the thin leafwhich needs special consideration[17][18][19]; As plant are very sensitive to environmentchanges, controlled environment is crucial even at labeling and imaging time. Andplants grow vertically, the plant PET bore axis must be vertical which is different fromthe human and small animal scanners. To address issues specific to functional plantimaging, a dedicated plant PET imager has been built with the aim of providing aregional resource for plant sciences.
2. Material and methods
The proposed functional plant PET system is designed with two major features in mind:(1) high spacial resolution and sensitivity, potentially configurable system geometry toaccommodate plants of different sizes and shapes and (2) the ability to control theenvironment in which the plants will be studied. To achieve these goals, we designed aPET system that is composed of high performance modular detectors for small animalPET scanners. Detector modules can be reconfigured to various geometry that aresuitable for acquiring projection or tomographic images for different parts of a plant.These detectors are mounted to translation and rotation stages that are controlled by acomputer remotely. The above components will be installed in a plant growth chamberwith full environmental control. Radio-tracer delivery system has been developed toenable real-time radio-labeling and imaging capability for plant imaging studies.
We use two types of detector modules (shown in figure 2a,b) in the imager with differentgeometry to accommodate different imaging needs. The first group consist of 8 SiemensInveon
T M detector modules, each containing 4 PS-PMTs to readout 4 LSO crystalarrays. Each LSO array contains 20 x 20 crystals each measuring 1.51 x 1.51 x 10 mm in 1.59 mm pitches. The second group consist of microPET R (cid:13) R4 detector modules,each containing 4 PS-PMTs to readout 4 LSO arrays. Each LSO array contains 8x8crystals each measuring 2.2 x 2.2 x 10 mm in 2.4 mm pitch. 21 R4 modules are used tobuild a detector panel with large solid angle coverage for imaging large plants. SiemensQuickSilver T M [20] electronics is used for detector readout and coincidence processing.figure 3 illustrates the data flow of the QuickSilver
T M readout electronics system. The16 output signals from 4 PS-PMT in each R4 module are multiplexed using a customPCB to mimic a single PS-PMT with 4 position-encoded signals. This allow us to use
Wang et al Figure 1.
Shows the plant PET system that is composed of (a) two sets of detectormodules; (b) readout electronics system; (c) positioning stages; (d) mechanical supportcomponents; (e) plant growth chamber; (f) radio-tracer delivery port, and (g) motionmonitoring and position indicators. Components a, c and d are inside the PGC thatcontrols the imaging environment.
Figure 2. (a) R4 detector module, (b) Inveon
T M detector module, (c) The floodhistogram of the whole R4 detector block. The outputs of four R4 block detectors aremultiplexed by a custom PCB to re-map four 8x8 crystal arrays to form a 16 x 16array to reduce the number of readout channels, (d) Inveon flood flood histogram.
Wang et al Figure 3.
Block diagrams of system readout electronics the QucikSilver
T M electronics to readout 32 Inveon
T M detectors and 84 R4 detectors,and still have additional 11 readout channels for future silicon photomultiplier (SiPM)based sub-millimeter detectors that are under developed[21]. Two segments of flatflexible cables(total length of 3.5 m) are used to connect the detector output to theQuickSilver
T M electronics through a custom designed junction board mounted on thePGC. figure 2c,d shows the flood histogram of a typical Inveon
T M detector and amultiplexed R4 module read out by system electronics. No observable signal degradationis observed with these long cables.
Eight Inveon
T M detector modules (32 blocks) and 21 microPET R (cid:13) R4 detector modules(84 blocks) can be arranged to either form a half-ring (of different radii) to provide high-resolution dynamic imaging capability with an imaging FOV of 15 cm diameter by 10 cmtall(figure 4.a). Alternatively, the Inveon
T M modules can be arranged as a quarter-ringfigure 4.b. With a step-and-shot motion, this configuration provides tomographic imagesof objects up to 25 cm in diameter. With Configuration in figure 4.c, the Inveon
T M modules are arranged in a plane to get projection images of even larger objects.The motion of the detectors is controlled by the use of 2 linear stages and 2 rotationstages, mounted on a optical table (60 cm x 90 cm). As shown in figure 4, the 2 verticalstages control the height of two sets of detector in order to track radiotracer throughoutan entire plant. The radius of the R4 modules based half-ring is fixed while the distancebetween Inveon
T M detector set to the center of the 2 concentric rotation stages can beadjusted by fixing the detector holding panel to different sets of holes drilled on the
Wang et al Figure 4.
Three different configurations of system geometry
Table 1.
Geometry parameters for different configurations
Config Actual FOV Imaging type Geometry/Radius (mm)uration Axial Transverse Inveon microPET R41 10 cm * 15 cm 4D half-ring/86.1 half-ring/140.72 8 cm * 25 cm 3D or 4D ** quarter-ring/166.63 12 cm * (up to) 40 cm 2D projection planar/variable
Notes: * larger axial FOV (up to 60 cm) can be achieved with multi-bed scan** depends on plant’s traslocation speed aluminum arm to accommodate plants of different size. A plant is typically centered atthe top of ration stage. The quarter-ring detector group and the plant can be rotatedindependently to form lines of response(LOR) from all angles for tomographic imaging.Two rotation controllers are connected in a daisy-chain via a RS232 serial port to thehost compute. The detailed geometry parameters of the three different configurationsare shown in Table 1.
The plant PET system have two different types of detector modules and it need somemotion control function to image different size of plants with different detector geometry.A custom designed imaging console software (figure 5) designed with MFC applicationframework provides the following main functions: (1) Detector modules and systemsetup; (2) Scanning angle and duration calculation based on system geometry and radio-nuclide half-life; and (3)Automated motion control an data acquisition.Detector modules setup includes ASIC working parameters setup, crystal lookuptable(CLU), energy lookup table(ELU), time alignment lookup table(TLU) generation.The supply voltages for Inveon and R4 modules are set to 700V and 800V individually.
Wang et al Figure 5.
Graphic User Interface (GUI) for detector and system setup. The screensnapshot shows an interactive crystal look-up table generating process
Tube gain difference of the four R4 detector blocks housed in the same module are mainlycompensated by manually adjusting the high voltage divider resistors on the multiplexerboards. For Inveon
T M modules, CLU, ELU and TLU are generated automatically. Setupof microPET R (cid:13) R4 modules requires minor manual effort to ensure correct identificationof corner crystals.Automated motion control part calculates the needed motion steps and sends commandsto the stepping motor controllers. The control codes also read back the positioninformation from the decoders mounted on the stepping motors’ axis to check ifrequested motions are completed.
The list-mode data are collected with QuickSilver
T M electronics and sorted by customsorting codes to sinogram data set. As multiplier boards are used for R4 detectormodules, remapping codes are needed to convert the index ordered with flood histogrampeaks to pre-defined reconstruction geometry coordinators.Reconstruction of PET images is based on the Maximum Likelihood estimation ofactivity concentration through the Expectation Maximization (ML-EM) algorithm. Thesystem matrix is factored into a normalization component, attenuation componentand geometric component. The geometric component of the matrix is computed bysubdividing the detector crystals and forming sub-LORs joining the sub-crystals. UsingSiddon’s algorithm, the average intersection in each voxel and divided by the squareof the length of the LOR to obtain the emission system matrix weights. Currently,as we do not have a method to estimate the attenuation of the subject under study,the attenuation component is ignored. For plants that have narrow stems and thinleaves, we suspect that the attenuation component is minor. The goal of normalizationis to estimate the not modeled parameters in the system matrix. For this, we scan a
Wang et al
T M ,Inveon
T M -Inveon
T M data andalso, efficiency of individual crystals. Randoms and scatter estimates are added intothe forward model. Random events rate is estimated through a delayed windowapproach. Scatter is estimated using a Single Scatter Simulation, whereby an imageis reconstructed first, down-sampled, and scatter component estimated under a singlescattering approximation. The tails of the scatter estimates are scaled and fitted todata.
As plant is very sensitive to environment changes, plant growth environment needs to becontrolled before and during imaging experiments. The plant PET imager is designedto be integrated in a plant growth chamber. The plant growth chamber (made byConviron) has an exterior dimension of 79.5 ′′ x 33.25 ′′ x 79 ′′ (WxDxH) and a interior 10ft growth area. The growth environment can be controlled with a temperature rangesfrom 4 ◦ Celsius to 45 ◦ Celsius, light intensity up to 1000 umoles/m /s, humidity levelfrom 40% to 90%, and CO level from the ambient level to above. The entire systemis located in a plant-imaging lab ( 24 m ) above our cyclotron facility for easy accessof a wide range of radionuclides and tracers. Gas isotopes are delivered via dedicatedgas turnings directly from the cyclotron facility. The redundant radioactive gas orthose flushing out from the labeling chamber can be recollected and delivered back tothe cyclotron for further administration. Custom labeling chambers of different size orshape are made of conventional polyvinyl chloride(PCV) or transparent acrylic tubes.The radioactive gas are delivered either directly into the labeling chamber inside thePGC or to the radio-labeling system resided in a fume hood beside the PGC.
3. Results
The detector block energy resolution was measured using a Ge-68 line source, and wasfound to be 15.1+/-1.4% and 23.8+/-6.2% FWHM at 511 keV for Inveon
T M and R4detector, respectively. These results agree with thoese published in literatures[22][23].Coincidence timing resolution between an Inveon
T M detector and a R4 detector wasfound to be 1.8 ns FWHM. The coincidence timing window of the system was set to3.1 ns, accordingly. For this particular detector geometry, system sensitivity was muchdifferent from conventional full ring systems. The sensitivity of the system shown infigure 6 was roughly measured with a 70 uCi Ge-68 point source at 3 different trans-axialposition (center, 5 cm off center to R4 half-ring, 5 cm off center to Inveon
T M half-ring)crossing the whole axial FOV with a 1.6 mm step size. The system sensitivity at theaxial center of those 3 selected positions is 1.3%, 1.4% and 3.0% for energy window
Wang et al Figure 6.
The system sensitivity measure with different energy windows at 3 differentpositions crossing the axial FOV
Figure 7.
The flood histograms of the two type of detector modules acquired at at30 ◦ Celsius and the photon peaks and crystal lookup tables are generated based thedata acquired at 20 ◦ Celsius (left:MicroPET R4 module, right:Inveon module). of 350-650 KeV, and 2.0%, 2.0%, 4.3% for 250-750 KeV respectively with 3.1 ns timewindow. The sensitivity along the center axis within +/-25 cm offset maintains a peakvalue which may be useful for small plants study. The system sensitivity is limitedby the solid angle coverage of the half-ring R4 modules. For the following imagingexperiments, data was acquired using an energy window of 350 KeV to 650 KeV and atiming window of 3.1 ns.The plant PET scanner is located inside the PGC where the temperature may variesto mimic the change caused by the alternation of day and night. The LSO light outputand the PMT’s gain and quantum efficiency of the photo-cathode are affected by the
Wang et al Figure 8.
Reconstructed uniform cylindrical phantom images and their profiles intangential, radial and axial direction. environmental temperature[24],[25]. This variation may result in a shift of the photonpeaks in the detector block’s flood histogram or crystal’s energy peaks. 25 ◦ Celsius and30 ◦ Celsius are the common selected temperatures for day and night time respectively.A 70 uCi Ge-68 point source was used to acquire single events with 20 ◦ Celsius and 30 ◦ Celsius respectively. The detector modules was kept in the two temperature conditionsfor at lease 2 hours before taking data. The single events are sorted with our consoleprogram. figure 7 shows the crystals photon peaks found based on the data set acquiredat 20 ◦ Celsius match very well with the same module’s flood histogram acquired at 30 ◦ Celsius. The energy peaks of individual crystals from the two selected detector modulesare compared and there a slightly decrease of energy peak value at 30 ◦ Celsius in theenergy spectrum, but the energy resolution keeps the same. The evaluation shows thatthe detector module works stably with 10 ◦ Celsius of temperature variation which ismeets the requirements from practical applications.
A 6-cm diameter Ge-68 cylindrical phantom with a uniformactivity concentration of 616 nCi/cc was used to normalize the system. A separate scanof the same phantom (with offset) was reconstructed with normalization and calculatedattenuation correction. Images in figure 8 show good uniformity in the whole FOV.
A home-made phantom with Derenzo-like hot rodpattern was scanned to evaluate the spatial resolution of the plant PET system. Theinner core of the phantom has a diameter of 32 mm and contains fillable hot rods ofdifferent size (0.80, 1.00, 1.25, 1.50, 2.00 and 2.50 mm) arranged into 6 segments. The
Wang et al Figure 9.
Central slice of reconstructed micro-Derenzo phantom image with an innercore diameter of 3.2 cm and hot rod diameters of 0.80, 1.00, 1.25, 1.50, 2.00, 2.5 mm distance between adjacent rods in each segment is twice the rod diameter. The phantomwas filled with 0.50 mCi(18.4MBq) of F-18 solution and scanned for 20 minutes. List-mode data was sorted into custom defined 3-dimensional sinograms and reconstructedwith ML-EM algorithm. The 3-dimensional image size is 200 x 200 x 320 pixels with a0.4 x 0.4 x 0.4 mm voxel size. For this phantom study, a conventional energy windowof 350 to 650 KeV and time window of 3.1 ns were applied. Phantom attenuation andscatter were not correction in reconstruction.The reconstructed transverse slice of the phantom is show in figure 9. The rodswith 2.5, 2.0, 1.5 and 1.25 mm diameter are clearly separated and the 1.0 mm diametergroup are moderately resolved. To evaluate the performance of this dedicated PET scanner for real plant imagingapplications, three pilot experiments were conducted, which also provided the dataof different plants’ C-CO absorption and translocation pattern. All of the followingimages are reconstructed with ML-EM algorithm and the image size is 400 x 400 x 160pixels with a 0.8 x 0.8 x 0.8 mm voxel size. A young cucumber plantwas labeled with 10 mCi C-CO in a cylindrical chamber(shown in figure 10.a). Thetotal uptake is 0.3 mCi at the end of the 15-minute labeling. The plant was imagedfor 10 minutes. Different parts of the cucumber plant are clearly delineated in thereconstructed image as shown in figure 10.c. The flowers appear to be the sinks of thephotosynthates. The top 3 leaves of a dwarf soybean plant werelabeled with 12 mCi C-CO using a homemade rectangular labeling chamber(figure 11)for 13 minute. The total uptake was estimated to be 6 mCi after decay correction back Wang et al Figure 10.
The first PET imaging experiment with the system: (a) a cucumber plantbeing labeled with C-CO , (b) the plant is being imaged, (c) maximum intensityprojection of the 3D PET images of the cucumber plant clearly shows uptake in leaf,petiole, flowers and stem. Figure 11. spot labeling of top 3 leaves of a dwarf soybean plant to the beginning of the labeling time. The plant was imaged 1 hour later (because itwas too hot), beginning at time points 0, 60, 85 and 140 minutes, respectively.Five volumes of interest(VOIs) were selected at(1) the junction of leaves and stem,(2) stem, (3) junction of stem and a soybean pod, (4) edge of the pod, and (5) a beaninside the pod. Each VOI is 2.4 x 2.4 x 2.4 mm (3 x 3 x 3 pixels in image). The meanvalue of the 27 voxels is plotted over time and shown as time-activity-curve of the VOIin figure 12. Most photosythates were translocated to the seed at the late frames. As shown in figure 13 and figure 14, a youngmutant and wild type maize (8 days after sowing) were labeled with about 10mCi(370MBq) of C-CO inside a custom made labeling chamber for about 10 minutes.The chamber was light up as soon as the C-CO gas was injected by a high luminosity Wang et al Figure 12. reconstructed dynamic images at different time points (imaging startedat 76 min post-injection), and time-activity-curve of selected VOIs
Figure 13. shows the imaging protocol
Wang et al Figure 14.
Different translocation pattern shown with wild type and mutant maizedynamic PET image in 1 hour, upper: mutant, lower: wild type
LED light source mounted on the chamber’s top cap. After the radio-labeling, theactivity was flushed out and the plant was moved into the plant PET imager for scanningfor about 2 hours. Raw data was binned into time frames with 5 minutes duration. Noattenuation and scatter correction was applied in the following image reconstruction.Dynamic images with 5-minute frames revel the different translocation and distributionpatterns of photoassimilates in the 2 types of maize plants. The PET images at latertime point (after 60 minutes) clearly show the root structures in regular soil. Small hotspots appeared at those small root ends, which possibly relates to a physiological truththat seeding root ends need more energy for new roots growth.
4. Discussion
This dedicated plant PET system features arbitrary geometry which practically reducethe total cost by reusing the detector modules from the old microPET R (cid:13) scanners. Onthe other hand, it provide more flexibility of detector modules can be used and detectorgeometry that would better fit our imaging objects, which is even more importantfor plant imaging. As mentioned above, the plant to be studied is of different sizeand shapes, a fixed detector geometry can not fulfill these requirements. Even themicroPET R (cid:13) scanner that is dedicated for small animal study also has two different boresize to better target different imaging objects. Our study already shows that detectorwith different crystal size can work well in one readout system, and we have already Wang et al N, C have very short half-life.The measured sensitivity at the center of FOV is almost the same when changing thesetting from 4 Inveon
T M detector half rings to 2 detector half rings. A third half-ringdetector set can easily be added to the system to acquire coincidence events with theInveon
T M detector set and improve system sensitivity. The third detector set can bemounted apart from the existed R4 detectors, as a result, we can simultaneously acquirecoincidence events from two separated FOV with adjustable distance. This will be veryuseful for tall plant study, as the most important two parts of a plant are the top whereflowers and young leaves are growing out and the bottom part with roots in soil.
The experience from our preliminary plant imaging experiments suggests that biggertrans-axial FOV is very useful. As show in figure 15, the maize roots growth very fastand the roots can go through the 10 inch long tube with in less than two weeks. Withthe multi-bed scan mode, we can increase the axial FOV, but some fast photoassimilatetranslocation information will be lost. As the activity is highly accumulated in the smallvolume of root structure, good image can be reconstructed with around one million ofcoincidence events which can be acquired within one minute. We can still achieve these5-minute frames dynamic images using step-and-shot motion in trans-axial directionwith 1.25 minute of duration per step. The corresponding trans-axial FOV will close to16 inch FOV which is four fold of current static system.
When talking about system throughput, it is not comparable with the conventionaloptical imaging system already applied to plant phenotyping studies. In fact, one ortwo minutes of scan can acquire enough coincidence events for reconstructing a good 3DPET image, which is comparable with the CT scanner. Depends on the study purpose,some studies need to continues monitor several hours long physiological process whilesome studies just need to acquire the PET image at a dedicated time point. One possiblehigh-throughput multi-modality imaging system is to combine the plant PET scannerthat work at the snapshot mode with a compact X-ray camera system to image plantswith both structure and functional information.For plant study, the environmental changes during the labeling and imaging stagesneed to be controlled to avoid the corresponding effects on plant growth. A completeautomated plant labeling system that on one hand will improve system’s throughputand on the other hand will mostly get rid of the fluctuation of environmental conditions.The plants to be studied are of different size, as a result, it is not easy to build a
Wang et al Figure 15.
Multi-bed scan of a 13-day wild type maize grown in a 10 inch long PVCpot and the segmented PET images of the whole plant, left: the plant and the fourbed positions marked with different colored rectangles, right: the PET images of thefour different bed positions, the corresponding acquired time points are 4 min, 17 min,37 min and 64 min as refer to the scan beginning time common labeling system. Some custom made labeling chambers are being or to be builtto accommodate more plants in our system. The automatic labeling system also workstogether with the automatic radio-tracer delivery system so that we can precisely controlthe radio-labeling process which is very important for a series of repetitive experiments.
5. Conclusion
We have developed a dedicated high-resolution plant PET scanner based the detectormodules from small-animal PET systems. The scanner features re-configurable systemgeometry and full control of plant growth environment. The system is composed oftwo different detector modules with different crystal sizes which shows a reasonablesensitivity of 1.3% at center of FOV, 18% system energy resolution, and 1.8 ns time
Wang et al R (cid:13) systems. Phantom studies and preliminary plant imaging experiments show thathigh quality 3D tomographic and dynamic PET images can be acquired with the fullring configuration. These initial plant imaging studies also clearly demonstrated thefunctional imaging capability of the plant PET system. Additional studies using N-13,C-11 and other radionuclides are being conduct by collaborating with regional plantscientists. This dedicated plant PET scan possesses a open system geometry, so newhigh resolution detector modules and optimized geometry configuration can be appliedto the system to improve the system performance in terms of spatial resolution, FOVand sensitivity.
6. Acknowledgments
We would like to than Dr. Stefan Siegel and Dr. Dongming Hu of Siemens MolecularImaging, Inc. for technical support in using Siemens QuickSilver
T M electronics, LeeSobotka, Carmen Dence at Washington University in St. Louis for helpful discussion,and Kinda Abdin, Tom Voller, Lori Strong, Laforest Richard, Patrick Zerkel, BillMargenau, Greg Gaehle for providing access to radioactive sources. This researchwas support by U.S. National Science Foundation, grant DBI-1040498. Imagingreconstruction was performed using the facilities of Washington University Centerfor High Performance Computing, which were partially supported by grant NCRR1S10RR022984-01A1.
References [1] Wanneng Yang, Lingfeng Duan, Guoxing Chen, Lizhong Xiong, and Qian Liu. Plant phenomicsand high-throughput phenotyping: accelerating rice functional genomics using multidisciplinarytechnologies.
Current opinion in plant biology , 16(2):180–7, May 2013.[2] S. J. Mooney, T. P. Pridmore, J. Helliwell, and M. J. Bennett. Developing X-ray ComputedTomography to non-invasively image 3-D root systems architecture in soil.
Plant and Soil ,352(1-2):1–22, November 2011.[3] Ljudmilla Borisjuk, Hardy Rolletschek, and Thomas Neuberger. Surveying the plant’s world bymagnetic resonance imaging.
The Plant journal : for cell and molecular biology , 70(1):129–46,April 2012.[4] Fabio Fiorani, Uwe Rascher, Siegfried Jahnke, and Ulrich Schurr. Imaging plants dynamics inheterogenic environments.
Current opinion in biotechnology , 23(2):227–35, April 2012.[5] Stijn Dhondt, Nathalie Wuyts, and Dirk Inz´e. Cell to whole-plant phenotyping: the best is yet tocome.
Trends in plant science , 18(8):428–39, August 2013.[6] M E Phelps. Positron emission tomography provides molecular imaging of biological processes.
Proceedings of the National Academy of Sciences of the United States of America , 97(16):9226–33, August 2000.[7] S R Cherry and S S Gambhir. Use of positron emission tomography in animal research.
ILARjournal / National Research Council, Institute of Laboratory Animal Resources , 42(3):219–32,January 2001.[8] Peter E. H. Minchin and Michael R. Thorpe. Using the short-lived isotope 11C in mechanisticstudies of photosynthate transport.
Functional Plant Biology , 30(8):831, 2003.
Wang et al [9] Matthew R Kiser, Chantal D Reid, Alexander S Crowell, Richard P Phillips, and Calvin R Howell.Exploring the transport of plant metabolites using positron emitting radiotracers. HFSP journal ,2(4):189–204, August 2008.[10] Amin Garbout, Lars J. Munkholm, Søren B. Hansen, Bjørn M. Petersen, Ole L. Munk, andRadoslaw Pajor. The use of PET/CT scanning technique for 3D visualization and quantificationof real-time soil/plant interactions.
Plant and Soil , 352(1-2):113–127, September 2011.[11] H Uchida, T Okamoto, T Ohmura, K Shimizu, N Satoh, T Koike, and T Yamashita. A compactplanar positron imaging system.
Nuclear Instruments and Methods in Physics Research SectionA: Accelerators, Spectrometers, Detectors and Associated Equipment , 516(2-3):564–574, January2004.[12] K. Ziemons, R. Barbier, G. Brandenburg, P. Bruyndonckx, Y. Choi, D. Christ, N. Costes,Y. Declais, O. Devroede, C. Dujardin, a. Fedorovd, U. Heinrichs, M. Korjik, M. Krieguer,C. Kuntner, G. Largeron, C. Lartizien, H. Larue, P. Lecoq, S. Leonard, J. Marteau, Ch.Morel, J.B. Mosset, Ch. Parl, Ch. Pedrini, a.G. Petrosyan, U. Pietrzyk, M. Rey, S. Saladino,D. Sappey-Marinier, L. Simon, M. Streun, S. Tavernier, and J.M. Vieira. The ClearPETproject: development of a 2nd generation high-performance small animal PET scanner.
NuclearInstruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectorsand Associated Equipment , 537(1-2):307–311, January 2005.[13] S Beer, M Streun, T Hombach, J Buehler, S Jahnke, M Khodaverdi, H Larue, S Minwuyelet, C Parl,G Roeb, U Schurr, and K Ziemons. Design and initial performance of PlanTIS: a high-resolutionpositron emission tomograph for plants.
Physics in medicine and biology , 55(3):635–46, February2010.[14] C Woody, a Kriplani, P OConnor, J.-F Pratte, V Radeka, S Rescia, D Schlyer, S Shokouhi, S Stoll,P Vaska, a Villaneuva, N Volkow, and B Yu. RatCAP: a small, head-mounted PET tomographfor imaging the brain of an awake RAT.
Nuclear Instruments and Methods in Physics ResearchSection A: Accelerators, Spectrometers, Detectors and Associated Equipment , 527(1-2):166–170,July 2004.[15] M. Budassi and S. Stoll. First results from the BNL plant imaging system. In
Nuclear ScienceSymposium and Medical Imaging Conference (NSS/MIC) , pages 3530–3532. Ieee, October 2012.[16] A.G. Weisenberger, B. Kross, S.J. Lee, J. McKisson, J.E. McKisson, W. Xi, C. Zorn, C.R. Howell,A.S. Crowell, C.D. Reid, and M. Smith. Nuclear physics detector technology applied to plantbiology research.
Nuclear Instruments and Methods in Physics Research Section A: Accelerators,Spectrometers, Detectors and Associated Equipment , 718:157–159, August 2013.[17] Marine Soret, Stephen L Bacharach, and Ir`ene Buvat. Partial-volume effect in PET tumor imaging.
Journal of nuclear medicine : official publication, Society of Nuclear Medicine , 48(6):932–45,June 2007.[18] David L Alexoff, Stephen L Dewey, Paul Vaska, Srilalan Krishnamoorthy, Richard Ferrieri, MichaelSchueller, David J Schlyer, and Joanna S Fowler. PET imaging of thin objects: measuring theeffects of positron range and partial-volume averaging in the leaf of Nicotiana tabacum.
Nuclearmedicine and biology , 38(2):191–200, February 2011.[19] Heyu Wu and Yuan-Chuan Tai. A novel phoswich imaging detector for simultaneous beta andcoincidence-gamma imaging of plant leaves.
Physics in medicine and biology , 56(17):5583–98,September 2011.[20] DF Newport and SB Siegel. QuickSilver: a flexible, extensible, and high-speed architecture formulti-modality imaging.
Nuclear Science Symposium Conference Record, 2006. IEEE , pages2333–2334, 2006.[21] Tae Yong Song, Heyu Wu, Sergey Komarov, Stefan B Siegel, and Yuan-Chuan Tai. A sub-millimeter resolution PET detector module using a multi-pixel photon counter array.
Physicsin medicine and biology , 55(9):2573–87, May 2010.[22] Y C Tai, A Chatziioannou, S Siegel, J Young, D Newport, R N Goble, R E Nutt, and S R Cherry.Performance evaluation of the microPET P4: a PET system dedicated to animal imaging.
Wang et al Physics in Medicine and Biology , 46(7):1845–1862, July 2001.[23] Qinan Bao, Danny Newport, Mu Chen, David B Stout, and Arion F Chatziioannou. Performanceevaluation of the inveon dedicated PET preclinical tomograph based on the NEMA NU-4standards.
Journal of nuclear medicine : official publication, Society of Nuclear Medicine ,50(3):401–8, March 2009.[24] M. Moszyski, a. Nassalski, a. Syntfeld-Kauch, T. Szcz´sniak, W. Czarnacki, D. Wolski, G. Pausch,and J. Stein. Temperature dependences of LaBr3(Ce), LaCl3(Ce) and NaI(Tl) scintillators.
Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers,Detectors and Associated Equipment , 568(2):739–751, December 2006.[25] S. Weber, D. Christ, and M. Kurzeja. Comparison of LuYAP, LSO, and BGO as scintillatorsfor high resolution PET detectors.