LABORATÓRIO DE INSTRUMENTAÇÃO E FÍSICA EXPERIMENTAL DE PARTÍCULAS

Transcription

1 LABORATÓRIO DE INSTRUMENTAÇÃO E FÍSICA EXPERIMENTAL DE PARTÍCULAS Preprin LIP/ July 2002 Perspecives for Posiron Emission Tomography wih RPCs A. Blanco 1,2, V. Chepel 1,3, R. Ferreira-Marques 1,3, P. Fone 1,4,*, M.I. Lopes 1,3, V. Peskov 5, A. Policarpo 1,3 1 LIP Laboraório de Insrumenação e Física Experimenal de Parículas, Porugal 2 GENP, Dep. Fisica de Pariculas, Univ. Saniago de Composela, Spain. 3 Deparameno de Física da Universidade de Coimbra, Coimbra, Porugal. 4 Insiuo Superior de Engenharia de Coimbra, Coimbra, Porugal. 5 Royal Insiue of Technology, Sockholm, Sweden. Absrac In his sudy we address he feasibiliy and main properies of a posiron emission omograph (PET) based on RPCs. The concep, making use of he converer-plae principle, akes advanage of he inrinsic layered srucure of RPCs and is simple and economic consrucion. The exremely good ime and posiion resoluions of RPCs also allow he TOF-PET imaging echnique o be considered. Mone-Carlo simulaions, suppored by experimenal daa, are presened and he main advanages and drawbacks for applicaions of poenial ineres are discussed. Presened a RPC2001-VI Workshop on Resisive Plae Chambers and Relaed Deecors, Coimbra, Porugal, November 2001 * Corresponding auhor: Paulo Fone, LIP-Coimbra, Deparameno de Física da Universidade de Coimbra, Coimbra, PORTUGAL. el: (+351) , fax: (+351) , fone@lipc.fis.uc.p

2 1 Inroducion Posiron Emission Tomography (PET) is a radioracer imaging echnique in which racer compounds labelled wih posiron emiing radionuclides are injeced ino he objec of sudy. These racer compounds can hen be used o rack biomedical and physiological processes, wih applicaions ranging from he early deecion of cancer o neurophysiology sudies. Afer a shor pah lengh he posiron annihilaes wih an elecron of he medium emiing simulaneously wo almos ani-parallel 511 kev phoons. The coinciden deecion of boh phoons idenifies he occurrence of an annihilaion along he chord joining he deecion poins and he accumulaion of such daa allows he reconsrucion of he aciviy disribuion in he issues. If, addiionally, he coincidence ime difference (difference in fligh ime beween he pair of phoons) is measured (TOF-PET) wih a FWHM accuracy he posiion of he annihilaion along he chord may be idenified wih a FWHM accuracy L given by L= c 2 = c( σ )2 (1) where σ is he rms ime accuracy per phoon. The numerical facors combine o yield Lmm [ ] 0.15 ps [ ] σ [ ps]2. (2) In any foreseeable sysem he localizaion accuracy given by eq.(2) is much coarser han he desired image granulariy (few mm) and herefore he image reconsrucion procedure canno be avoided. However, by including he TOF informaion, along wih oher advanages [1], one gains a sensiiviy improvemen of he order of L L, where L is he ypical objec lengh, hus reducing he number of evens needed for he image reconsrucion [2]. For his reason he invesigaion of TOF-PET sysems is being acively pursued (e.g. [3], [4]). In order o compare a TOF-PET sysem wih a convenional (non-tof) one and accoun for he above-menioned increase in sensiiviy when using TOF informaion, one may inroduce an inrinsic sensiiviy meri facor defined as 2 εtof Lmm [ ] f ε 0.15 ps [ ]. (3) where ε TOF and ε are he quanum efficiencies for a single gamma phoon in TOF and non-tof sysems, respecively. However, in a pracical PET omograph many addiional facors deermine he overall sysem sensiiviy in erms of dose given o he paien for an exam requiring a cerain image qualiy (see for insance [5], [6] and references herein). Examples of such facors would include: 1) Sysem price, which largely deermines he affordable field of view (FOV: axial deecor lengh); 2) Localisaion accuracy of he phoon ineracion poins, which includes he deph-of-ineracion (DOI) informaion, he posiion resoluion along he axial and ransaxial direcions of he omograph and he errors due o Compon scaering in he deecor; 3) The coun rae capabiliy, including deecor occupancy and daa acquisiion (DAQ) hroughpu; 4) The ime resoluion of he couner and of he DAQ, which influences he number of random coincidences; 5) Couner geomery, in paricular he use (2D-PET) or no (3D-PET) of collimaing sepa dividing he FOV in nearly independen rings. For mos of hese performance facors here will be considerable differences beween he RPC TOF-PET and he crysal PET approaches. However a full discussion of such aspecs and of heir inerplay canno be done in he presen repor, being only oulined he mos prominen differences o be expeced beween hese approaches based on simulaions and on experimenal resuls. 2 The RPC TOF-PET concep The RPC TOF-PET concep is based on he converer plae principle [7] and akes advanage of he naurally layered srucure of RPCs, of is simple and economic consrucion, excellen ime resoluion (60 ps σ even for single gaps equipped wih posiion-sensiive readou [8], [9]) and very good inrinsic posiion accuracy (50 µm online in digial readou mode [10]). A possible RPC srucure for TOF-PET applicaions is shown in Figure 1. The incoming gamma phoons will inerac wih he elecrode maerials hrough he phooelecric and Compon processes. The resuling elecrons srongly scaer in he maerial and evenually emerge ino he gas gap, ionising he gas and iniiaing avalanches along heir pah. The exponenial dependence of he avalanche final charge wih he posiion of he iniial charges assures ha only hose avalanches iniiaed

3 close o he cahode will be deeced. (In Figure 1 he lead foil was chosen as cahode.) d... Lead-acrilic (0.6 mm) e - e - Lead-acrilic γ Spacers defining he gas gap (d=0.3mm) Lead foil (30 µm) Time and posiion signal pickup elecrodes on a hin PCB Figure 1 Couner consrucion assumed in he GEANT4 simulaions presened below wih a schemaic indicaion of he mos relevan physical processes. Each layer has a oal hickness of 1 mm resuling in a 10 cm hick 100-layer couner. Oher approaches based on he converer plae principle, alhough offering a much poorer ime resoluion, may be found in [11], [12]. 3 Experimenal and Mone-Carlo resuls 3.1. Quanum Efficiency and Energy Sensiiviy Clearly he weaker poin of he converer plae approach o medium-energy phoon deecion is he low quanum efficiency (probabiliy of deecion per inciden phoon) relaive o inorganic scinillaors. Calculaions show ha a sack of very hin lead foils obains he bes resuls. However, for keeping he mechanical accuracy of he couner, he foils mus be mouned on a rigid resisive elecrode. A possible maerial may be leadloaded acrilic (LA - used for ransparen X-ray proecions), which offers is bes performance a a convenien hickness of 0.5 o 1 mm. We measured a quanum efficiency of for single lead foils in good agreemen wih ([7], [12]), while he corresponding GEANT4 simulaion yielded a value of , validaing GEANT4 for his applicaion. Furher simulaions sugges ha a 140-layer couner wih he srucure shown in Figure 1 should reach an efficiency per phoon ε TOF =22% (Figure 2), yielding a relaive meri facor eq.(3) for a 24 cm diameer objec f = 0.33, (4) where a coincidence ime resoluion of 300 ps FWHM was used (see secion 3.3). Quanum Efficiency Nr. of gaps All ineracions Secondary ineracions Energy (KeV) Figure 2 Simulaed (GEANT4) quanum efficiency as a funcion of he inciden phoon energy for a sack of couners (see Figure 1). For a large number of plaes he values scale almos linearly wih he energy above 100 kev. Only a small fracion of he secondary ineracions (from couner-scaered phoons) is visible. The proposed sysem would show a sensiiviy handicap by a facor of hree when compared wih a sandard crysal PET, all oher variables being he same. However for some special applicaions (see secion 3.5) here are advanages in he presen proposal ha may easily ouweigh his handicap. Using phoon energy disribuion daa generaed by he public domain SIMSET simulaion package [13] i was deermined ha he energy sensiiviy feaure of he presen approach is equivalen, in erms of rejecion of coincidences involving a scaered phoon, o an energy discriminaion hreshold of 300 kev Posiion resoluion A srong poin of he converer plae approach is is posiion resoluion capabiliy (e.g. [14]). In Figure 3 we show he GEANT4-simulaed image of a disc phanom in waer imaged by a 16-layer couner wih a srucure similar o ha shown in Figure 1. The inner radius was 4 cm and he sack was 16 mm hick. All high-energy processes were simulaed including he posiron pah range (using he β energy disribuion from 18 F) and phoon non-colineariy. The FWHM RPC posiion resoluion was assumed o be wice he widh of he gas gap (2 0.3=0.6 mm) for elecrons ejeced from he anode and 0.2 mm for elecrons ejeced from he cahode. The image was reconsruced by he sandard algorihm of filered backprojecion. The conras of he resuling image was enhanced by subracion of he low frequency background.

4 Figure 3 Mone-Carlo simulaion (GEANT4) of 40 µm diameer discs spaced by 1 mm. The discs are clearly resolved, in agreemen wih he corresponding 0.38 mm FWHM Poin Spread Funcion. Since he deph of ineracion of he phoons is milimerically deermined by he deph of he firing gap no parallax effec was observed in he simulaions. The 40 µm diameer discs spaced by 1 mm are clearly resolved. The corresponding Poin Spread Funcion was deermined by reconsrucing poin sources and shows a widh of 0.38mm FWHM Time resoluion Measuremens aken wih 0.3 mm single-gap RPCs made wih elecrodes of glass and aluminium yielded a FWHM coincidence ime resoluion = 300 ps, equivalen o σ = 90 ps, for 511 kev phoon pairs. Since such couners are known o deliver a resoluion σ 60 ps [8] when irradiaed wih minimum ionising paricles (MIPs) and no opimisaion of he mechanical and operaional parameers for his applicaion was done ye, furher improvemens are reasonably o be expeced. Anyhow, hese resuls already compare favourably wih he ypical TOF-PET coincidence resoluion of 500 ps FWHM achievable wih BaF 2 or LSO segmened crysals [3] and even wih large BaF 2 crysals yielding 340 ps FWHM [15]. 3.4.Rae capabiliy Since glass (or even bakelie) RPCs are amongs he gas couners wih lower rae capabiliy, i should be asked wheher hese couners migh wihsand he very large couning raes desirable for PET. The couning rae capabiliy of iming glass RPCs under coninuous irradiaion is around 200 Hz/cm 2 [16], corresponding o 20 khz/cm 2 for a 100-layer couner. For he large area applicaion considered in he nex secion, he corresponding maximum singles coun rae would be 1.2 GHz, which is likely unreachable due o oher reasons such as random coincidences and DAQ hroughpu limiaions. Therefore i seems ha for large area couners he maximum sysem couning rae will no be limied by he RPCs rae capabiliy. 3.5 Field of view and overall sysem sensiiviy The presen approach may allow he realizaion of affordable full-body FOV scanners. Simulaions by oher auhors [17] sugges ha increasing he axial FOV of a sandard crysal PET from 20 o 60 cm would improve he noise-equivalen coun rae (NEC 1 [18]) by a facor of five o eigh. Such resuls are shown in Figure 4 along wih our own simulaions, based on he SIMSET package [13]. Our curves correspond o a 80 lier waer cylinder wih a diameer of 24 cm and 1.8 m lengh. The simulaion is mean o deermine only geomerical effecs and doesn' include rae-dependen effecs like sysem dead ime or random coincidences, while he resuls of ref. [17] where obained wih he Zubal phanom, using larger couning raes and modelling rae-dependen effecs. To allow a comparison beween boh resuls he curve from [17] was normalized o a NEC sensiiviy of 10 khz/(kbq/ml) for a 20 cm FOV (a common value for 3D-PET: e.g.[19]) and our curve o 3.3 khz/(kbq/ml) following eq.(4). Two accepance angles 2 were considered kbq/ml120 NEC (khz) a 1 Accepance=±45º Accepance=±15º Field of View (cm) Figure 4 Simulaed noise-equivalen coun raes as a funcion of he axial field of view. The circles correspond o sandard 3D PET [17] and he hick lines o RPC TOF-PET. A sensiiviy increase by a facor 10 over sandard 3D PET (20 cm FOV) may be possible for whole-body examinaions. The resuls sugges ha a 2.5 meer FOV RPC TOF-PET would show a NEC sensiiviy improvemen by a facor 10 over he sandard PET sensiiviy, vasly ouweighing he inrinsic sensiiviy handicap for whole-body examinaions. 1 The number of rue coincidences ha would creae an image of similar qualiy in he absence of noise (scaered and random coincidences). 2 The maximum angle of he incoming phoons relaive o he radial direcion.

5 4. Conclusions We assessed several key feaures of an economic and posiionally accurae RPC-based approach o he TOF-PET medical imaging echnique. Efficiency up o 22% for 511 kev phoons may be achievable for a 140-layer couner along wih an energy sensiiviy equivalen in erms of scaer rejecion o an energy cuoff a 300 kev. A coincidence ime resoluion of 300 ps FWHM was measured for 511 kev phoons, comparing favourably wih he values achievable wih segmened fas crysals. The excellen (380 µm FWHM) parallax-free posiion resoluion expeced for he reconsruced images (from small diameer couners) may have direc ineres for he high-resoluion imaging of small animals required by pharmaceuical research. The simple srucure of RPC-ype deecors and he consequen possibiliy of economic consrucion in large areas opens he way o consider also whole-body field-of-view TOF- PET couners. Simulaions sugges ha such sysems may reach an overall sensiiviy one order of magniude larger han presen day crysal-based PET scanners. Such improvemens may open new medical applicaions by lowering considerably he radiaion hazard involved in whole-body PET examinaions and simulaneously increase he paien hroughpu. 5. Acknowledgemens This sudy was suppored by he FCT projec CERN/P/FIS/40111/2000. The auhors graefully acknowledge he Cenro de Supercompuación de Galicia (CESGA) for providing very subsanial compuaional resources and he very special collaboraion of Dr. Carmen C. Bueno. 6. References [1] S. E. Derenzo e al., "Criical insrumenaion issues for <2 mm resoluion, high sensiiviy brain PET," in Quanificaion of Brain Funcion: Tracer Kineics and Image Analysis in Brain PET, K. Uemura, N. A. Lassen, T. Jones, and I. Kanno, Eds. Amserdam: Elsevier Science Publishers, 1993, pp (hp://cfi.lbl.gov/insrumenaion/pubs/quanbrainfunc.pdf). [2] T. F. Budinger, J. Nucl. Med., vol. 24, pp , [3] W. W. Moses and S. E. Derenzo, IEEE Transacions on Nuclear Science NS-46, pp (1999). [4] T. Yamaya e al., Phys. Med. Biol. 45 (2000) [5] W. W. Moses, Nucl. Insrum. and Meh. A471 (2001) [6] D. Croseo, "A modular VME or IBM PC based daa acquisiion sysem for muli-modaliy PET/CT scanners of differen sizes and deecor ypes", presened a he IEEE Nuclear Science Symposium and Medical Imaging Conference, Lyon, France, 2000, published in he conference record. hp:// [7] J.E.Baeman, Nucl. Insr. and Meh 221 (1981) 131. [8] A. Blanco e al., Nucl. Insr. and Meh. A478 (2002) 170. [9] A. Blanco e al., "Single-gap iming RPCs wih bidimensional posiion sensiive readou for very accurae TOF sysems", hese proceedings. [10] V. Peskov and P. Fone, Gain, Rae and Posiion Resoluion Limis of Micropaern Gaseous Deecors, presened a he PSD99-5h Inernaional Conference on Posiion-Sensiive Deecors, London, England, Also preprin LIP/01-06 (hp://xxx. lanl. gov/abs/physics/ ). [11] A Jeavons e al., IEEE Trans. Nucl. Sci. NS-30, pp , [12] J.Lacy e al., Nucl. Insr. Meh. A 471 (2001)88. [13] T.K Lewellen e al., The SimSET program, in Mone Carlo calculaions in nuclear medicine: applicaions in diagnosic imaging, M. Ljungberg; S-E. Srand, and M. A. King, Eds. Philadelphia: Insiue of Physics, 1998, pp [14] A. Jeavons e al., IEEE Trans. Nucl. Sci. Vol 46, No 3, pp , June [15] J.Cederk. e al., Nucl. Insr. and Meh. A471 (2001) [16] A.Akindinov e al., Nucl. Insr. Meh. A456 (2000) 16. [17] R.D. Badawi e al., IEEE Trans. Nucl. Sci. NS-47, pp , [18] S.C.Sroher e al., IEEE Trans. Nucl. Sci. Vol. 37, pp , [19] Lars-Eric Adam e al, Journal of Nuclear Medicine Vol. 42 No. 12 (2001)