Monte Carlo Analyses of X-ray Absorption, Noise and Detetive Quantum Effiieny Considering Therapeuti X-ray Spetrum in Portal Imaging Detetor G. Cho', H.K. Kim', Y.H. Chung', D.K. Kim', H.K. Lee', T.S. Suh' and K.S. Joo4 'Dept. of Nulear Eng. and 'Material Eng., Korea Advaned Institute of Siene and Tehnology, Taejon 305-701 KOREA 'Dept. of Biomedial Eng., Catholi niversity Medial College, Seoul 137-040 KOREA 'Dept. of Physis, Myongji niversity, Yongin 449-728 KOREA Abstrat The bremsstrahlung spetrum from 6 MV line'ar aelerator (LINAC) as obtained and used as an input x-ray soure in the simulation to estimate several important physial quantities of the detetor in therapeuti x-ray portal imaging system, suh as quantum and energy absorption effiienies, Sank fator and the detetive quantum effiieny (DQE) et. In addition, e have obtained a spatial distribution of energy deposit ithin the detetor, resulting in a spatial-frequeny dependent DQE. From the simulation results, it is found that the use of metal plate largely enhanes the energy absorption, leading to the large output signal. Hoever, onsidering the noise properties, the presene of metal plate degrades the DQE at non-zero spatial frequenies beause the detetor absorption noise is dominated by the quantum absorption. For the verifiation of our simulation results, e have ompared ith experimental measurements and results of other literatures. I. INTRODCTION As alternatives to the onventional x-ray film used for the verifiation of radiation field plaement in the external radiation therapy, many different types of eletroni portal imaging devies (EPIDs) have been developed suh as a matrix ion hamber, a video-based EPID and the most reently developed flat-panel type devies (amorphous silion- or amorphous selenium-based pixel detetors) in the last fe deades [I]. Among them, the video-based EPID is the most idely used no due to the simple and inexpensive fabriation ith use of harge-oupled devies (CCDs) or omplementary metal-oxide-semiondutor (CMOS) image sensor of hih groing tehnology ill give higher resolution and sensitivity at lo ost. In a video-based EPID, a phosphor sreen pasted to a metal plate, as an x-ray onverter or so-alled a portal detetor, is the first stage in the asaded imaging hains of transferring anatomial information to the final user. Therefore it is largely responsible for the total performane of the system. Sevral theoretial and experimental studies for the optimal design of the metal plate/phosphor sreen ombination have been performed [2-71. Hoever, those studies ere based on simple approximations, for example, the mono-energeti x-ray beam as the input. Also some papers onsidered the detetive quantum effiieny (DQE) only at zero frequeny. Based on Monte Carlo tehnique, e have investigated the overall performanes of the portal detetor in terms of x-ray absorption and assoiated noise haratristis onsidering the breinsstrahlung spetrum from 6 MV linear aelerator (LINAC). In addition, e simulated the spatial distribution of the energy deposit ithin the detetor, resulting in the frequeny-dependent DQE, or DQE(n. The simulation results ill be disussed in omparison ith the experimental measurements and other literatures in detail. 11. MATERIAL AND METHODS The phosphor film is the first signal transfer stage in the x- ray imaging system and its performane is haraterized by many terms, suh as the energy absorption effiieny, the detetive quantum effiieny, the noise poer spetrum, the modulation transfer funtion et. Sank [8] and Jaffray et al. [4] have desribed these properties using the energy-moments of the absorbed energy distribution (AED) in the ase of the mono-energeti x-ray beam. The AED is the distribution of deposited energy from eah x-ray quantum in the phosphor film through the interation mehanisms suh as photo-eletri absorption, Compton sattering and the pair prodution. The AED is hardly measured diretly so it is normally simulated by Monte Carlo tehnique. The deposited or absorbed energy annot exeed the energy of the inident x-ray. One e kno the AED, then e an alulate its energy moments of various degrees by the folloing equation. M,, = jes(e)e"de (1) here E is the amount of the x-ray energy absorbed by the film, S(E) is the AED, and M,, is the n-th moments. Based on this expression, e an desribe several physial quantities important in estimating the performane of the phosphor film as an x-ray detetor as follos: AQ =MO, (2) A, =M,lMo, (3) A, = M; IMoM2, (4) DQE = M ; i ~,, (5) here, A,, AE, and As represent the quantum absorption, energy absorption, and Sank or statistial fator, respetively, and the DQE stands for the detetive quantum effiieny. We took this approah, but extended to the poly-energeti x-ray beam. The AED of a phosphor film from x-rays having the energy spetrum of @(E') is simply expressed by the folloing integral [9] S(E) = joeo@(e')k(e, E')fE' (6) O-78O3-6jO3-8/0 I /$ 10.00 200 1 I EF3: 19-64
~ 0.3 0.4.- v) + = 0.2 V - N m E 0.1 z 0.0 0 1 2 3 4 5 6 Photon energy (MeV) Figure 1 : Normalized bremsstrahlung spetrum from the gold target of 6 MV LINAC ith the inident eletron kineti energy of 5.58 MeV. Confirming of Monte Carlo simulation, the intensity spetrum alulated by Shiff formula is also inluded. here Eo is the maximum energy in the spetrum and the kernel R(E,E ) is the response funtion of the phosphor, hih is the probability that a single inident x-ray quantum ith energy E ill deposit the amount of energy E ithin the detetor. As the first stage of analyses, a bremsstrahlung spetrum from the gold target of a 6 MV LINAC (Siemens Digital Mevatron KD, Serial # M2498, the inident eletron kineti energy of 5.58 MeV), @(E ), as simulated by MCNP4B ode [IO] based on the material and geometri data [ll], as an inident x-ray soure in the simulation. Then, in order to alulate AED of the phosphor, e onsidered a penil beam inident perpendiularly onto the portal detetor ith the bremsstrahlung spetrum. The portal detetor is a ombination of a metal plate and a phosphor film. The funtion of metal plate is to stop the energeti therapeuti x-rays and to produe seondary eletrons, of hih esaped ones ill deposit energy in the phosphor film. We used a opper plate of various thiknesses (0-50 mm) and a phosphor film of Gd202S (0.1-5 mm) ith redued density of 3.67 g/m3 [2,4], aounting for other ompositions like polyurethane polymer-binder and air pokets. We used MCNP4B ode again to simulate the interations of x-rays and seondary eletrons in the metal plate and phosphor film. The obtained AED of the phosphor film for various thiknesses of the metal plate and the phosphor film as used to estimate the energy moments and performane quantities disussed in above by simple numerial integration and division. Aounting for the spatial spread of sattered x-rays and seondary eletrons in the phosphor film, leading to the DQE(f), the loal distribution of energy absorption as estimated in pre-defined voxels of the phosphor layer in the simulation [7]. Then, the enter point of eah ell as assumed to be the representative position of the isotropi optial photon radiator of hih intensity is the proportional to the absorbed energy of the hole ell [7]. Finally the DQEO as alulated using the frequeny-dependent quantum aounting theory [ 12,131. In order to onfirm our simulation results, e onduted several experimental measurements suh as the relative sensitivity and spatial resolution of the hand-made detetor ith the x-ray beam from 6 MV LINAC. The detailed experimental proedures an be found in referene [14]. In addition, some of data from Dr. Munro s alulation and measurements [4,5,13] ere used for the omparison ith our results. 111. RESLTS AND DISCSSION A. Bremsstrahlung X-rays Simulated the intensity spetrum of bremsstrahlung x-ray beam produed from 6 MV LINAC is presented in Fig. 1. All omponents of LINAC-head are onsidered in this simulation, suh as the gold target inluding the ooling system, the primary ollimator, a flattening filter and a dose hamber. In order to onfirm the simulation reliability, Shiff intensity spetrum [15], hih is idely used in the analysis of experimental results in high-energy or aelerator physis, as also alulated and inluded in the same figure. The mean energy of x-rays is 1.57 MeV. In the first energy bin or belo 558 kev, the simulation underestimates. This disrepany an be understood as the effet of the eletron energy-utoff at 500 kev used in the simulation. Hoever, sine the bremsstrahlung radiation yield at the energy belo 500 kev is less than 5.3% in gold, the appliation of this energy-utoff is tolerable. 15,....,....,....,....,.... i Phosphor thikness 10 20 30 40 50 Figure 2: Quantum absorption effiienies of the portal detetors. The use of metal plate degrades the quantum absorption beause the seondary quanta emitted from the metal plate gradually derease. 19-65
40 50 F W._ - m 0.6-0.4 6 MV LINAC (Siemens) LanexTM sreen (1 34 mg/m2) Figure 3: Energy absorption effiienies of the portal detetor. The energy absorption is diretly related to the detetor signal output. The introdution of metal plate largely enhanes the energy absorption effiieny, hih is strongly related to the seondary quanta fluene and their ranges ithin the detetor. B. X-ray Absorption For the inident bremsstrahlung x-rays Fig. 2 and 3 sho the quantum and energy absorption effiieny, respetively, as a funtion of the metal thikness for three ases of phosphor thiknesses. These ere alulated using the energy spetrum of bremsstrahlung x-ray rather than the intensity spetrum shon in Fig. 1. These x-ray absorption effiienies are very sensitive to the phosphor thikness as expeted. As shon in figures, the energy absorption, resulting in the final signal output of phosphor, is largely enhaned by the introdution of metal plate even for the thikness of a fe millimeters, hile the quantum absorption is degraded. In addition, as the phosphor is thinner, the absorption effiienies beome independent of the metal plate thikness beause the relatively long eletron range in phosphor. Hoever the eletron energy spetrum does not vary ith the metal thikness beause of the relatively short eletron range in meta. Sine the light signal from the portal detetor is diretly proportional to the total absorbed energy ithin the detetor, it an be estimated by multiplying the quantum and energy absorption. Therefore, e an easily expet that there exists the maximum value of the metal thikness from Fig. 2 and 3. We irradiated the LanexT fast hak sreen (Eastman Kodak, GdlO,S:Tb, 134 mgkm ) oupled to the opper plate ith various thiknesses for the LINAC (Siemens) operated at 6 MV [ 141, and then ompared ith our simulation results. As shon in Fig 4, the simulation result is in a good agreement ith the measured data. C. Sank Fator Figure 4: Relative detetor sensitivity as a funtion of the metal thikness. It is assumed that the signal output from the detetor is diretly proportional to the total energy deposited ithin the detetor. Y v) 0.6,....,....,.,..,,.., I,,,,, Phosphor thikness 0.0 Figure 5: Sank fators of the portal detetor. Sine the Sank fator is proportional to AE, the overall feature is similar to the results ofae. Hoever, the dependeny of the phosphor thikness is reversed beause the variane of S(E) beomes large as the phosphor thikness inreases. Generally Sank fator is understood as the normalized DQE over the quantum effiieny. The alulated Sank fator, as presented in Fig. 5, has a saturation feature as a funtion of metal thikness hih is similar to the energy absorption. But its magnitude dereases ith the phosphor thikness, hih is opposite to the energy absorption. These harateristis an be understood easily if e rearrange (4) as follos using the definition of A, and AE ; A, = AQ(A,)l /MI. (7) 19-66
0 v ' n 0.03 t --- Phosphor thikness -1 mm Figure 6: Detetive quantum effiienies of the portal detetor at zero spatial frequeny. DQE is resulted from the ompetition beteen A, and As. Above 5 mm-thik metal plate, the quantum absorption dominates the DQE. I L3 lo-' 4 Detetor ombinations metal thikness of about 5 mm beause the Sank fator is almost onstant at that region. In interest of metal thikness (0-5 mm), the DQE is almost onstant beause the rapidly hanging of the quantum absorption and the Sank fator. Thus, referring the results of energy absorption for maximized signal output, the metal thikness of 5 mm is onluded as an upper limit. It is important to note that maximizing the DQE at zero spatial frequeny is not suffiient in portal imaging. In order to be able to detet anatomial strutures in the portal images reliably, it as reported that the DQEV) should be maximized up to spatial frequenies of 0.4 mm-' [16]. Considering the spatially absorbed energy distribution ithin the detetor, the DQEV), as shon in Fig. 7, as alulated for several portal detetors. The spatial distribution of the energy absorption in portal detetor as onfirmed by the measurement of the linespread funtion ith 6 MV x-ray beam 1141. From the simulation results, to important onlusions have been extrated: (i) the spatial frequeny-"independent" DQE values for thinner phosphor layer, and (ii) the presene of metal plated degrades the DQE for non-zero spatial frequenies. This onlusion is different from other literatures based on mono-energeti x-ray approximations [4,6]. In order to onfirm our Monte Carlo alulation of DQE(f), e ompared ith theoretial and experimental data from the literatures 15,131. For the omparison, e alulated DQEO of their detetor [4-5,131 for therapeuti x-ray beams from another type of 6 MV LINAC (Varian) [17]. Figure 8 shos our alulation result hih gives exellent agreement to the referene data. Munro's group [4] alulated DQE(0) by using EGS4 ode, and resulted in 0.017 hih is very lose to our value, 0.018 of DQECf) at zero frequeny. -0-1 mm-metal/l mm-phosphor 0 3 mm-phosphor only 0.5 1.o 1.5 2.0 Gd202S sreen (358 mg/m2) Spatiat freqeny Figure 7: DQEV) for several portal detetors. It IS noted that the presene of metal plate degrades the DQECf). From the above expression, the Sank fator is a stronger funtion of the energy absorption than the quantum absorption so its shape as a funtion of metal thikness follos the energy absorption. Hoever, the variane of AED, M2, beomes large hen the phosphor thikness inreases, so its magnitude has the reversed tendeny as a funtion of phosphor thikness in omparison ith the energy absorption. D. Detetive Quantum Effiieny, DQE As mentioned before, the DQE is determined in terms of the quantum absorption and the Sank fator. As shon in Fig. 6, the DQE follos the quantum absorption above the a 0.5 1.o 15 2.0 Spatiat frequeny (mm") Figure 8: DQE(0 for 1 mm Gd202S ith 1 mm opper for 6 MV x- ray beam from Varian LINAC. The simulated result is exellent agreement ith the experiment [5,13]. The value, 0.018 at zero frequeny is omparable to 0.01 7 by ECS4 ode [4]. 19-67
IV. CONCLSION sing the Monte Carlo tehnique, e have studied' the x- ray absorption and its assoiated noise properties for the typial portal detetor onsidering the realisti x-ray sptrum from a linial LINAC. From the simulation rsults, it as found that: (a) There is an optimum thikness of metal plate to maximize the detetor sensitivity. For example, 1.5 mm-thik opper plate for GdlOzS sreen ith 134 mg/m' (or thik) gives the highest detetion sensitivity for x-ray beam from 6 MV LINAC. I (b) The DQEO beomes onstant as the thikness of phosphor layer dereases, hih is due to the uniform energy deposition of seondary quanta by relatively long range, resulting in minimized absorption noise. () The use of metal plate degrades the DQEO beause the detetor absorption noise is dominated by the quantum absorption. The suggested simulation ill be useful for the optimal design of typial portal detetors for different energy input, and also an be extended to any sintillating phosphor detetor for other x-ray imaging systems suh as flat-panel type devies. V. REFERENCES [ 11 P. Munro, "Portal imaging tehnology: Past, present, and future," Sem in Radiat Onol 5, 1 15-133 (1 995). [2] T. Radliffe, G. Barne B. Wok, R. Rajapakshe and S. Shalev. "Monte Carlo optimization of metal/phosphor sreens at megavoltage energies," Med Phys 20, 1161-1169 (1993). [3] B. Wok, T. Radliffe, K.W. Leszzynski, S. Shalev and R. Rajapakshe, "Optimization of metal/phosphor sreens for on-line portal imaging," Med Phys 21, 227-235 (1994). [4] D.A. Jaffray, J.J. Battist A. Fenster and P. Munro, "Monte Carlo studies of x-ray energy absorption and quantum noise in megavoltage transmission radiography," Med Phys 22, 1077-1088 (1995). [5] J. Bissonnette, LA. Cunningham and P. Munro, "Optimal phosphor thikness for portal imaging," Med Phys 24, 803-814 (1997). 161 C. Kaush, B. Shreiber, F. Kreuder, R. Shmidt and 0. Dossel, "Monte Carlo simulations of the imaging performane of metal plate/phosphor sreens used in radiotherapy," Med Phys 26, 2 1 13-2 124 (1999). [71 H.K. Kim, G. Cho, Y.H. Chung, H.K. Lee and S.C. Yoon, "Monte Carlo studies of metal/phosphor sreen in therapeuti x-ray imaging," Nul Instr Meth A422, 713-717 (1999). [S] R.K. Sank, "Absorption and noise in x-ray phosphors," J Appl Phys 44,4199-4203 ( I 973). [91 G. Cho, H.K. Kim, H. Woo, G. Oh and D.K. H "Eletroni dose onversion tehnique using a NaI(TI) detetor for assessment of exposure dose rate from environmental radiation," IEEE Trans Nul Si 45, 981-984 (1998). [IO] J.F. Briesmeister, MCNP-A General Monte Carlo N- Partile Transport Code, Version 4B, LANL Report, LA- 12625-M (1997). 1 I] B.L. Helmer, private ommirniation, Siemens Medial Systems, In. 121 LA. Cunningham, M.S. Westmore and A. Fenster, "A spatial-frequeny dependent quantum aounting diagram and detetive quantum effiieny model of signal and noise propagation in asaded imaging systems," Med Phys 2 I, 4 17-427 ( 1994). [ 131 J. Bissonnette, I.A. Cunningham, D.A. Jaffray, A. Fenster and P. Munro, "A quantum aounting and detetive quantum effiieny analysis for video-based portal imaging," Med Phys 24, 815-826 (1997). 141 Y.H. Chung, H.K. Kim, G. Cho, S.K. Ahn, H.K. Lee and S.C. Yoon, "Charaterization of x-ray detetor for CCDbased eletroni portal imaging devie," J Biomed Eng Res 21, 119-127 (2000). 151 G.E. Desobry and A.L. Boyer, "Bremsstrahlung revie: An analysis of the Shiff spetrum," Med Phys 18, 497-505 (1991). [16]F. Kreuder, B. Shreiber, C. Kaush and 0. Dossel, "A struture based method for on-line mathing of portal images for an optimal patient set-up in radiotheraphy," Philips J Res 51, 317-337 (1998). [17]S.S. Kubsad, R. Makie, M.A. Gehring, D.J. Misiso, B.R. Palial, M.P. Mehta and T.J. Kinsell "Monte Carlo and onvolution dosimetry for stereotati radiosurgery," Int J Radiation Onology Biol Phys 19, 1027-1035 (1990). 19-68