The effect of copper conversion plates on low-Z target image qualityDavid Parsonsa)

Department of Physics and Atmospheric Science, Dalhousie University, 5820 University Avenue, Halifax,Nova Scotia B3H 1V7, Canada

James L. RobarDepartment of Radiation Oncology, Dalhousie University, 5820 University Avenue, Halifax,Nova Scotia B3H 1V7, Canada

(Received 8 May 2012; revised 11 July 2012; accepted for publication 12 July 2012; published 13August 2012)

Purpose: Common electronic portal imaging devices (EPIDs) contain a 1.0 mm copper conversionplate to increase detection efficiency of a therapeutic megavoltage spectrum. When used in imagingwith a photon beam generated with a low atomic number (Z) target, the conversion plate attenuatesa substantial proportion of photons in the diagnostic range, thereby reducing the achievable imagequality. In this work, we measure directly dependence on low-Z target image quality as a function ofcopper plate thickness, for both planar imaging and cone beam computed tomography (CBCT).Methods: Monte Carlo modeling was used to quantify changes to the diagnostic spectrum and de-tector response for low-Z target beams generated with either 2.35 or 7.00 MeV electrons incident ona carbon target. Planar contrast-to-noise ratio (CNR) and spatial resolution measurements were madeas a function of copper thickness. CNR measurements were made for CBCT imaging as a function ofdose both with and without the copper plate present in the EPID.Results: The presence of copper in the EPID decreased the diagnostic photon population by up to20% and suppressed the peak detector response (dose deposited in the scintillator) at 60 keV bya factor of 6.4. Planar CNR was increased by a factor ranging from 1.4 to 4.0, depending on thematerial imaged, with no copper present compared to a standard 1.0 mm thickness. Planar spatialresolution was only slightly degraded with increasing copper thickness. Increases in CBCT imageCNR ranged from a factor of 1.3–2.1 with the copper plate removed.Conclusions: It is possible to increase the proportion of photons in the diagnostic energy range(25 keV–150 keV) reaching the phosphor screen by as much as 20% when removing the cop-per conversion plate. This results in significant increases of planar and CBCT image CNR. Con-sequently, we suggest that the copper conversion plate be removed from the EPID when used forlow-Z target planar or CBCT imaging. © 2012 American Association of Physicists in Medicine.[http://dx.doi.org/10.1118/1.4742052]

Key words: low-Z target, EPID, planar imaging, CBCT

I. INTRODUCTION

Megavoltage planar or cone beam computed tomography(MV CBCT) with a therapeutic 6 MV radiation beam hasbeen used to provide information for image guided radiationtherapy.1 However, MV imaging suffers from poor contrast-to-noise (CNR) characteristics and involves a relatively highimaging dose. While the introduction of a kilo-voltagex-ray tube and detector orthogonal to the treatment beamline2 has greatly improved available image quality, it has alsoincreased the associated quality assurance of matching imag-ing and treatment isocenters, added additional equipment, in-creased maintenance costs, and does not allow beam’s-eye-view imaging. The replacement of the high atomic number (Z)target with of a low-Z target3–13 has been shown to increasethe low energy component of the beam, thereby improvingthe available image quality in MV-CBCT.7, 10, 14–16 While themajority of the research regarding low-Z targets has focusedon target composition, incident electron energy and their re-sulting effects on image quality for either planar or CBCTimaging, relatively little has been reported with regard to op-

timizing the detector for a low-Z target energy spectrum. Atypical electronic portal imaging device (EPID) is composedof a metal conversion plate (typically copper or steel), a de-tection receptor (typically a scintillating phosphor), and pho-todiode array. In MV imaging, photons interact with the metalconversion plate to produce Compton recoil electrons, whichinteract with the scintillating phosphor to produce visible lightthat can be detected by the photodiodes.17 While a typical flat-tened 6 MV beam contains less than 0.5% of photons in thediagnostic energy range (e.g., 25–150 keV), this fraction in-creases to as much as 50% for a low-Z target beam.11 Whilethis component of the spectrum is the most useful from thepoint of view of image formation, they also have a high prob-ability of absorption in the copper conversion plate.

Several groups have investigated the effect of detectorcomposition on image quality in combination with low-Z tar-gets. Orton and Robar11 modeled 4 MeV and 6 MeV elec-trons incident on aluminum and beryllium low-Z targets in ahigh-energy linac, with an aS500 detector (Varian Medical,Inc.), demonstrating that the removal of the copper conver-sion plate can result in a contrast enhancement factor of 1.2, at

5362 Med. Phys. 39 (9), September 2012 © 2012 Am. Assoc. Phys. Med. 53620094-2405/2012/39(9)/5362/10/$30.00

5363 D. Parsons and J. L. Robar: Effect copper conversion plates low-Z target image quality 5363

both 4 MeV and 6 MeV. Faddegon et al.7 qualitatively showedthat MV-CBCT image quality could be increased with the re-moval of the copper conversion plate when using a carbon tar-get with 4.2 MeV incident electrons. Roberts et al.9 demon-strated that altering the detector receptor from Gb2S2O:Tb toCsI:Tl crystal can significantly increase the detector responsefrom 4 MeV electrons incident on nickel and carbon target.Compared to a 6 MV therapeutic beam, they showed a con-trast enhancement factor ranging from 1.3 for a thick waterphantom (25.8 cm) to 4.62 for a thin water phantom (5.8 cm).More recently, Roberts et al.15 showed a twofold increase incontrast by reducing the incident electron energy to 1.4 MeVin combination with a CsI:Tl crystalline scintillator detectorcompared to a 6 MV therapeutic beam and a standard mega-voltage EPID in the thick water phantom. However, with theincrease in image quality the associated cost of a crystallinescintillator is greatly increased, as is the complexity of man-ufacturing a crystal without defects. In lieu of a crystallinescintillator, Breitbach et al.18 showed that the copper conver-sion plate and Gb2S2O:Tb could be replaced with a segmentedscintillator composed of a Gb2S2O ceramic. Using the sameimaging beam line as Faddegon et al.,7 they showed that itis possible to increase the CNR in CBCT images by a factorof 1.64 compared to the unaltered EPID with no statisticallysignificant change in spatial resolution.

It is clear that by altering the detector composition, it ispossible to significantly affect image contrast and CNR. How-ever, to date no one has quantitatively measured the effectthe copper conversion plate has on image CNR in standardMV EPID detectors. Through measurement we investigate(i) the effect that the thickness of the copper conversion platehas on a 2.35 and 7.00 MV/carbon spectra and the associ-ated detector response, and (ii) the resultant changes in imagequality, both qualitatively and quantitatively. This paper in-cludes the following: In Sec. II, we present the methodologyon beam generation, the imaging system, Monte Carlo mod-els, and imaging protocols. In Sec. III, we present results anddiscussion, which include Monte Carlo simulations, planar,and CBCT image quality analysis.

II. MATERIALS AND METHODS

II.A. Low-Z target beam production

The low-Z target beams were generated using a Varian2100EX linear accelerator (Varian Medical, Inc., PaloAlto, CA). Two separate carbon targets were investigated(i) 7.6 mm thick (2.35 MV/carbon) and (ii) 22.8 mm thick(7.00 MV/carbon). The thicknesses of the carbon targets werecalculated to coincide with the continuous slowing downapproximation range for 2.50 MeV and 7.00 MeV electrons.While it has been shown that an increase in low-Z targetthickness slightly degrades resolution,12 a “full-thickness”target alleviates the need for a polystyrene filter to absorbtransmitted electrons and increases the production of lowerenergy photons.3 As described previously,10–12 when usedexperimentally, the carousel is operated in manual modeand the appropriate target is rotated into the beam line via

a rotary switch. For the 2.35 MV/carbon beam, the linac isoperated in electron mode and equipped with a customized4 MeV program board (used for research purposes only). Forthe 7.00 MV/carbon beam, a standard 6 MeV program boardis employed. To generate the 2.35 MeV beam, the bendingmagnet was adjusted to tune the beam energy down from thenominal operating point of 4.5 down to 2.35 MeV. To accountfor the diminished beam current observed with the reducedmean electron energy, the electron gun current, grid voltage,solenoid, and buncher cavity steering set points were adjustedto maximize electron current. The design and installation ofsimilar targets similar to those used in this study have beenpublished recently11, 12 and will not be explained in full here.Briefly, targets were mounted to the beam side of the carouseland vertically offset 0.9 cm from the beryllium exit windowof the primary collimation vacuum. By keeping the target asclose as possible to the beryllium exit window, the amountof electron scattering in air is minimized.11 It is possible todecrease mean electron energy below 2.35 MeV, however,this significantly increases acquisition times, degrades spatialresolution, and has no noticeable advantage with respect toimage CNR.

II.B. Imaging system

An aS1000/IAS3 imaging system (Varian Medical Sys-tems, Inc.) was used for all imaging. The system was setup asa stand-alone configuration separate from the clinical imag-ing system on the treatment unit. The aS1000 panel (IDU20)has an active area of 30 × 40 cm2 and includes a 1.0 mm Cubuildup plate, a Gd2O2S:Tb scintillating phosphor layer and a1024 × 768 array of photodiodes switched by thin-film tran-sistors deposited on a glass substrate. The number of monitorunits (MU) per exposure was varied to control the dose to thephantoms, where the IAS3 allows centi-MU control of thisparameter. Prior to imaging, flood field, and dark field cali-brations were performed where the former has the effect ofcorrecting for the forward peaked nature of the low-Z target,flattening filter free beam.

II.C. Monte Carlo simulations

In order to study the effects of the copper conversion plateon the 2.35 and 7.00 MV spectra, Monte Carlo simulationswere run using BEAMnrc.19 The linac and IDU20 weremodeled in accordance to the physical setup using validatedmodels provided by Orton and Robar.11 It should be notedthat these models were originally of an aS500, howeverbetween the two detectors, the physical composition isrelatively consistent with the main difference being thenumber of pixels within the active area of the detector. 2.0× 107 incident electron histories were run for both 2.35 and7.00 MeV beams. Linac collimation was set to match theactive area of the detector at a source to detector distance(SDD) of 140 cm. Selective bremsstrahlung splitting wasused, and as recommended by Rogers et al.,20 an effectivesplitting field size of 50 × 50 cm2 was used for source to sur-face distance (SSD) of 140 cm with minimum and maximumbremsstrahlung splitting values of 200 and 1000, respectively.

Medical Physics, Vol. 39, No. 9, September 2012

5364 D. Parsons and J. L. Robar: Effect copper conversion plates low-Z target image quality 5364

FIG. 1. Diagram of the full imaging system. (a) A schematic drawing on the linac and detector. (b) A photo of the detector with the cover off with copper platesand Cerrobend weights.

Global electron (ECUT) and photon (PCUT) cut-off energiesof 0.700 MeV and of 0.010 MeV, respectively, were used.The resulting phase space was recorded at the exit side of thecopper conversion plate, above the phosphor. To study theeffect of the copper conversion plate thickness on the energyspectra, the thickness was varied in 0.2 mm increments from0 mm to 1.6 mm, corresponding with the physical setup ofexperiments, as described below. To determine the spectraldistributions, BEAMdp (Ref. 21) was used to analyze thephase space data. The resulting data were analyzed andplotted using MATLAB (Mathworks, Natick, MA).

The IDU20 was also modeled in DOSXYZnrc (Ref. 22)to determine the effect the thickness of the copper conver-sion plate has on detector response. Similar to the methodsused by Roberts et al.,9, 15 mono-energetic pencil beams ofvarying energy from 0.010 MeV to 7.00 MeV were inci-dent on the detector, and the detector response was quanti-fied by scoring dose deposited in the phosphor screen perincident particle. 5.0 × 107 incident photon histories wereused for all simulations, with PCUT and ECUT set to 0.010MeV and 0.521 MeV, respectively. Detector response sim-ulations were repeated for 0 mm, 1.0 mm, and 1.6 mmcopper thicknesses. The dose distributions were analyzedwith MATLAB and plotted with energy fluence for 2.35 and7.00 MV/carbon beams.

II.D. Planar imaging

To observe changes in image quality with varying thick-nesses of copper, the gantry was rotated to 0◦ (in IEC coordi-nates) and the detector was oriented perpendicular to the cen-tral axis at a SDD of 140 cm, as shown in Fig. 1. To decreasethe amount of backscatter radiation, Styrofoam blocks sup-ported the detector from the floor. The detector cover, foam

spacing materials, and copper conversion plate were removed.To vary the thickness of copper, 14 × 14 × 0.02 cm3 coppersheets were stacked on the paper sheet covering the scintilla-tor. This allowed for the thickness of copper to range from 0mm to 1.6 mm in 0.2 mm increments. To compress the cop-per sheets, a customized Cerrobend frame was manufacturedto apply pressure around the periphery of the stacked cop-per sheets, and the weight of the frame was increased until itwas observed that all layers of copper were compressed firmlyagainst the panel. To quantify changes in image quality, a pla-nar CNR and QC3 spatial resolution phantom23 were alter-nately positioned on the central axis. The CNR phantom con-sisted of various tissue substitutes surrounded by polystyrene(electron density relative to water, ρe

w = 1.08). The tissuesubstitutes were 2.5 cm in diameter and 3.0 cm thick and sim-ulated cortical bone 60% (ρe

w = 1.69), cortical bone 30%(ρe

w = 1.28), inner bone (ρew = 1.09), brain (ρe

w = 1.05),and lung (ρe

w = 0.44). A relatively thin phantom was used asit allows sensitivity in measuring changes of CNR with cop-per thickness. The beam was collimated to 10 × 10 cm2 ata SSD of 140 cm. An imaging dose of 0.02 cGy was deliv-ered to the center of the CNR phantom to be representative ofthe approximate dose per projection given in CBCT. A higherdose of 0.19 cGy was delivered to center of the QC3 phan-tom for spatial resolution measurements. A higher dose wasused as a consequence of high noise at 0.02 cGy resulting inspurious results. CNR was calculated as

CNR =∣∣Pmaterial − Ppolystyrene

∣∣√

σ 2material + σ 2

polystyrene

, (1)

where Pmaterial is the average pixel value within the tis-sue substitute, Ppolystyrene is the average pixel value in the

Medical Physics, Vol. 39, No. 9, September 2012

5365 D. Parsons and J. L. Robar: Effect copper conversion plates low-Z target image quality 5365

surrounding polystyrene, σ material is the average noise withinthe tissue substitute, and σ polystryene is the average noise withinthe surrounding polystyrene. Error bars were found by calcu-lating the mean and standard deviation of CNR measured inmultiple images. Spatial resolution was analyzed according tothe method outlined by Rajapakshe et al.,23 where the relativemodulation transfer function (RMTF) is defined as

RMTF = M(f )

M(f1), (2)

where M(f) is the output modulation of the line pair in a re-gion of interest and M(f1) is the output modulation for thelowest frequency line pair region. As suggested by Droegeand Morin,24 the output modulation is obtained by using therelationship between the signal amplitude and its variance de-fined as

M2 (f ) = σ 2m (f ) − σ 2(f ), (3)

where σ m2(f) is the measured total variance within a region of

interest, and σ 2(f) is the variance due to random noise withinthe region of interest. The variance due to random noise ascalculated by Rajapakshe et al.23 is

σ 2 (f ) = σ 2sub

2, (4)

where σ 2sub(f) represents the variance within the region of in-

terest of two subtracted images. Uncertainty in these measure-ments was estimated by calculating the mean and standarddeviation of RMTF measured in multiple image sets.

Planar imaging of a sheep head was done to qualitativelyobserve the effect the copper conversion plate has on planarimage quality at both low and high energies. The setup forimaging of the head was as follows: The gantry and couchwere rotated to 90◦. The imaging panel was placed in a standlocated on the couch with fine adjustments screws and lev-eled to ensure orthogonality of the panel and the beam axis.The head was centered on isocenter and a SDD of 130 cm wasused. All imaging involved an approximate dose of 0.26 cGydelivered to the center of the head. Images were first takenwith the copper conversion plate within the detector. The cop-per plate was then removed and replaced with a 2.1 mm sheetof foam (to ensure adequate compression of the panel assem-bly) and the detector was reassembled. Images were then ac-quired without the copper conversion plate present within thedetector. Images were analyzed in MATLAB and were com-pared with identical gray level window settings.

II.E. CBCT imaging

The CBCT images were acquired using the method de-scribed previously by Robar et al.10 Briefly, this involvedarranging the beam and detector as described above for planarimaging of the sheep head and rotating the object on a rotationstage consisting of a microstepping motor (Intelligent Mo-tion Systems, Inc., Marlborough, CT). The rotation stage wasaligned such that the axis was at the isocenter. Image acquisi-tion control was accomplished using MATLAB with the DigitalI/O toolbox. All projection images were saved in DICOM for-

mat and subsequently read by the reconstruction software. Im-age reconstruction was done using the Feldkamp Davis Kress(FDK) algorithm25 implemented in MATLAB. The phantomused for CBCT was cylindrical, 12.1 cm in diameter by 6.1 cmin length, composed of polystyrene and contained the sametissue substitutes as described above for planar imaging. Tis-sue substitute materials were 2.5 cm in diameter and 7 cmin length. CBCT images were acquired both with and with-out the copper conversion plate present in the detector, withcentral doses to the phantom ranging from 1.1 to 11.8 cGy.Imaging was repeated for both the 2.35 MV/carbon and7.00 MV/carbon beams. CNR was measured as described inEq. (1), with the exception that error bars were calculated bymeasuring the mean and standard deviation of CNR in threeslices above and below the analyzed slice.

To observe qualitative changes, CBCT imaging of thesheep head was done as well, for both beam energies, withand without the copper plate. An approximate imaging doseof 4.1 cGy was delivered to the center of the head, locatedon isocenter. The resultant images were analyzed in MATLAB

and were compared with identical gray level window settings.

III. RESULTS AND DISCUSSION

III.A. Monte Carlo characterization

Figure 2(a) shows modeled fractional fluences at the exitside of the copper conversion plate, illustrating the effectof various copper thicknesses on the energy spectrum, overthe range from 25 keV to 150 keV, for both 2.35 and7.00 MV/carbon beams. When increasing copper thicknessgreater than 0 mm, the fluence below 100 keV is significantlyattenuated [Fig. 2(a)], and accordingly the percentage of di-agnostic photons is reduced. It should be noted that for thecopper-free cases, the increase in fluence below 38 keV for the0 mm copper spectra was due to the presence of barium(k-edge at 37.45 keV) within the detector cover. For the2.35 MV/carbon beam, the diagnostic photon population(25 keV–150 keV) was decreased by 20% [Fig. 2(b)] withthe standard 1.0 mm copper conversion plate in place, com-pared to 0 mm of copper. A corresponding 15% decreasewas observed for the 7.00 MV/carbon beam. Figure 3 illus-trates the response of the IDU20 with various thicknesses ofcopper as well as the photon spectra produced with the 2.35and 7.00 MV/carbon beams. With no copper present in theIDU20, detector response peaks at approximately 60 keV anddecreases rapidly with increasing photon energy. When cop-per was placed within the IDU20, the response at 60 keV wassuppressed by factors of 6.4 and 8.5 for 1.0 mm and 1.6 mmthicknesses, respectively. At 0.8 MeV, the response was ap-proximately the same regardless of copper thickness, abovewhich the responses diverge with the copper responses in-creasing with increasing energy. Figure 3 demonstrates thatthe 2.35 MV/carbon beam contains a substantial fluence ofphotons within the peak detector response region (approxi-mately 30–120 keV). For the 7.00 MV/carbon beam peak flu-ence occurs at approximately 400 keV, at which point the re-sponse is locally minimized for the copper-less detector. In

Medical Physics, Vol. 39, No. 9, September 2012

5366 D. Parsons and J. L. Robar: Effect copper conversion plates low-Z target image quality 5366

FIG. 2. (a) The diagnostic spectral distribution with 0 mm, 0.4 mm, 1.0 mm, and 1.6 mm of copper present in the detector for 2.35 MV/carbon and7.00 MV/carbon beams. (b) The percentage of photons within the diagnostic energy domain (25–150 kV) at the phosphor surface as a function of copperthickness.

summary, the results of Figs. 2 and 3 indicate that copperremoval is advantageous for either low-Z target beam withregard to maximizing the diagnostic population reaching thephosphor, and particularly for the 2.35 MV/carbon beam, al-

lows the energy spectrum to be well matched to the detectorresponse. The detector response curves shown here agree withsimilar results given by Roberts et al.9 and Faddegon et al.7

for an Elekta iViewGT (Elekta AB, Stockholm, Sweden) and

FIG. 3. Response curves for 0 mm, 1.0 mm, and 1.6 mm thick copper layers within the IDU20 (left axis) and photon energy fluence produced with the2.35 MV/carbon and 7.00 MV/carbon beams (right axis).

Medical Physics, Vol. 39, No. 9, September 2012

5367 D. Parsons and J. L. Robar: Effect copper conversion plates low-Z target image quality 5367

FIG. 4. Planar image contrast-to-noise ratio as a function of copper thickness for both 2.35 MV/carbon and 7.00 MV/carbon beams, for cortical bone 60%(top-left), cortical bone 30% (top-right), inner bone (middle-left), brain (middle-right), and lung (bottom-left).

a Perkin Elmer XRD 1640 AG9 panel (Perkin Elmer, Inc.,Waltham, MA), respectively. Both of these detectors includea 1.0 mm copper plate, a gadolinium oxysulphate scintillat-ing layer, and an array of photodiodes.16, 26 Both papers haveshown that a high-Z scintillating material strongly interactswith the diagnostic portion of the energy spectrum. Roberts

et al.9, 15 also show the detector response for an Eleckta XVIpanel (Elekta AB, Stockholm, Sweden) which differs substan-tially in design, with the omission of a copper conversionplate and replacement of the gadolinium oxysulphate scin-tillating layer by columnar, thallium doped, cesium iodide(CsI:Tl) crystal. The responses of this detector and the IDU20

FIG. 5. Relative modulation transfer function (RMTF) for 0.0 mm, 1.0 mm, and 1.6 mm thick copper layers in the IDU20 detector for 2.35 MV/carbon and7.00 MV/carbon beams.

Medical Physics, Vol. 39, No. 9, September 2012

5368 D. Parsons and J. L. Robar: Effect copper conversion plates low-Z target image quality 5368

FIG. 6. Planar imaging of a sheep head at 0.26 cGy (at isocenter) for both0.0 mm (right) and 1.0 mm (left) of copper within the detector at both 2.35(bottom) and 7.00 MV/carbon (top). For comparison, the images were nor-malized to the pixel value at a point within all the images and with identicalgray level window settings.

without copper are similar in that they share a peak responseat approximately 50–70 keV and a local minimum betweenapproximately 500–600 keV, with the main difference in re-sponses being in energies less than 40 keV.

III.B. Planar image quality

Figure 4 shows image CNR as a function of copper thick-ness for various tissue equivalent materials. For most materi-als, with the exception of lung, image CNR decreases quicklywith the addition of the first few 0.2 mm thick layers of cop-per and approaches a relatively constant CNR value beyond0.8 mm of copper. The rapid degradation in CNR with thefirst few layers corresponds to the decrease in the diagnosticfluence with copper shown in Fig. 2; both the fluence in thediagnostic range and CNR show the largest decreases withthe addition of the first 0.4 mm of copper. With no copperpresent, image CNR was increased by factors of 1.6, 2.3, 3.2,1.4, and 2.3 for cortical bone 60%, cortical bone 30%, innerbone, brain, and lung, respectively, compared to the standard1.0 mm of copper. Similar gains in CNR were observed forthe 7.00 MV/carbon beam, with factor increases ranging from1.6 to 4.0; however, the absolute values of CNR were typicallyless than half of those observed at 2.35 MV. The results pre-sented here confirm the modeled results of Orton and Robar,11

in that a gain in image contrast was observed with the removalof the copper layer.

Figure 5 shows RMTF at various thicknesses of copper forthe 2.35 and 7.00 MV/carbon beams. At most frequencies,there appears to be little variation with changing copper thick-ness. At 0.25 lp/mm, there is a detectable but small changein RMTF, with relative improvement of RMTF by 18% and15% for the 2.35 and 7.00 MV/carbon beams, with removal ofthe 1.0 mm copper layer. The frequency where RMTF equals0.5, or f50, did not vary significantly with copper removal, foreither low-Z target beam. With the 1.0 mm copper layer, here

FIG. 7. Profiles of Fig. 6 with and without the copper conversion plate within the detector for 2.35 MV/carbon and 7.00 MV/carbon beams.

Medical Physics, Vol. 39, No. 9, September 2012

5369 D. Parsons and J. L. Robar: Effect copper conversion plates low-Z target image quality 5369

FIG. 8. Cone beam computed tomography images at approximately 4.1 cGywith copper plate (top) and without copper (bottom) in the detector, with the7.00 MV/carbon and 2.35 MV/carbon beams. The tissue equivalent materialsimaged are (clockwise starting from the darkest) lung, cortical bone 60%,cortical bone 30%, inner bone, and brain.

we see decrease of f50 0.03 lp/mm when reducing electronenergy from 7.00 to 2.35 MeV. In comparison, Connell andRobar12 reported a change in f50 of approximately 0.02 and

0.03 lp/mm for beryllium and aluminum target beams, whenreducing the incident electron energy from 7.0 to 4.5 MeV.

Figure 6 shows planar sagittal sheep images both withand without the copper conversion plate with the 2.35 and7.00 MV/carbon beams. The images show increased imagecontrast with the removal of the copper conversion plate, com-pared to images with the copper conversion plate in place.This is most noticeable within the squamous, nasal, andmandible regions. These differences are highlighted in theprofiles shown in Fig. 7, in which the contrast is notably in-creased with the removal of the copper conversion plate forthe 2.35 MV/carbon beam. Figure 7 also shows slight lossof fine detail when using the 2.35 MV/carbon beam com-pared to 7.00 MV/carbon. As noted by Connell and Robar,12

this is largely a result of electron beam broadening when de-creasing incident electron energy. However, the changes arerelatively minor between images with and without the cop-per conversion plate, which agrees with the results shown inFig. 6.

III.C. CBCT image quality

Figure 8 shows CBCT images of tissue equivalent materi-als in polystyrene both with and without copper in the IDU20for the 2.35 MV/carbon and 7.00 MV/carbon beams. TheCBCT images show greater contrast with the copper conver-sion plate removed both at 7.00 and 2.35 MV/carbon, withthe greatest improvement seen for 2.35 MV/carbon. This is

FIG. 9. Contrast-to-noise ratios of tissue substitutes in CBCT images as a function of dose, comparing the 2.35 MV/carbon and 7.00 MV/carbon beams bothwith and without copper present in the detector.

Medical Physics, Vol. 39, No. 9, September 2012

5370 D. Parsons and J. L. Robar: Effect copper conversion plates low-Z target image quality 5370

FIG. 10. CBCT images of a sheep head at approximately 4.1 cGy (at isocen-ter) with copper plate (top) and without copper (bottom) in the detector, withthe 7.00 MV/carbon and 2.35 MV/carbon beams.

evident also in the corresponding CNR curves as a functionof dose, shown in Fig. 9. At approximately 4.1 cGy and forthe 2.35 MV/carbon beam, for example, copper removal pro-duced improvement of CNR by factors of 1.5, 1.9, 2.1, 1.4,and 1.3 for cortical bone 60%, cortical bone 30%, inner bone,brain, and lung, respectively. Similar factor increases are seenfor the 7.00 MV/carbon beam, but as for the planar imag-ing results, CNR values are typically half of that observedat 2.35 MV/carbon. The CNR improvement with copper re-moval seen here is comparable in magnitude to the factor1.64 gain in CNR observed by Breitbach et al.18 when using asegmented ceramic Gb2S2O scintillator, compared to a stan-dard 1.0 mm copper conversion plate with Gb2S2O scintillat-ing phosphor. However, this group used a different beam-line7

within a Siemens linac at 4.2 MV.Figure 10 shows CBCT images of the sheep head with and

without the copper conversion plate with the 2.35 MV/carbonand 7.00 MV/carbon beams. These images highlight the effectthe presence of copper in the IDU20 has on image contrast-to-noise characteristics. This can be seen in the bony structuresof the skull, and in the nasal cavity, where at 2.35 MV/carbonare clearly visualized without copper, compared to the imageacquired with the copper layer. These trends are also presentin the 7.00 MV/carbon images; however, the CNR is clearlydegraded. These images taken here agree with the qualitativefindings reported by Faddegon et al.,7 in which there is anoverall improvement in bone and soft tissue visibility withcopper removal; however, in this work we show that the re-duction of energy to 2.35 MeV offers an overall improve-ment in image quality. While we believe that with a moreefficient detector such as a crystalline scintillator would bepreferable,9, 15 these results demonstrate that substantial im-provements in image quality may be realized with a standardMV EPID, following removal of the copper conversion plate.

IV. CONCLUSIONS

In this work, we have shown that the removal of the copperconversion plate within an IDU20 EPID can significantly im-prove image contrast and CNR when used in conjunction withlow-Z target imaging beams. The presence of the copper con-version plate caused a decrease in the diagnostic populationof 20% and 15% for 2.35 MV/carbon and 7.00 MV/carbonbeams, respectively. The peak response within the detectorwithout copper was reduced by a factor of 6.4 when 1.0 mmof copper was present in the detector. Planar images acquiredwith a copper-less detector showed a factor gain in CNR rang-ing from 1.4 to 3.2 and 1.6 to 4.0, for 2.35 MV/carbon and7.00 MV/carbon beams, respectively, compared to the stan-dard 1.0 mm of copper. Similar gains in CNR were observedin CBCT images, where CNR was improved by a factor rang-ing from 1.3 to 2.1. Spatial resolution was not affected signif-icantly by copper removal, for either beam. Qualitative imag-ing of a sheep head showed that there was an increase in pla-nar and CBCT image contrast with a copper-less detector.

ACKNOWLEDGMENTS

The authors are grateful for support provided by VarianMedical Systems for the funding of this project and the Nat-ural Sciences and Engineering Research Council of Canadafor their additional financial support. We would like to thankRobert Moran for his vital electronic and technical support,Ian Porter for fabrication of apparatus, Dr. Robin Kelly andDr. Mammo Yewondwossen for their knowledge and teach-ing of techniques utilized in this project.

a)Author to whom correspondence should be addressed. Electronic mail:david.parsons@dal.ca

1O. Morin, A. Gillis, J. Chen, M. Aubin, M. K. Bucci, M. Roach, andJ. Pouliot, “Megavoltage cone-beam CT: System description and clinicalapplications,” Med. Dosim. 31, 51–61 (2006).

2D. A. Jaffray and J. H. Siewerdsen, “Cone-beam computed tomographywith a flat-panel imager: Initial performance characterization,” Med. Phys.27, 1311–1323 (2000).

3D. M. Galbraith, “Low-energy imaging with high-energy bremsstrahlungbeams,” Med. Phys. 16, 734–746 (1989).

4D. W. Mah, D. M. Galbraith, and J. A. Rawlinson, “Low-energy imagingwith high-energy bremsstrahlung beams: Analysis and scatter reduction,”Med. Phys. 16, 653–665 (1993).

5O. Z. Ostapiak, P. F. O’Brien, and B. A. Faddegon, “Megavoltage imagingwith low Z targets: Implementation and characterization of an investiga-tional system,” Med. Phys. 25, 1910–1918 (1998).

6A. Tsechanski, A. F. Bielajew, S. Faermann, and Y. Krutman, “A thin targetapproach for portal imaging in medical accelerators,” Phys. Med. Biol. 43,2221–2236 (1998).

7B. A. Faddegon, V. Wu, J. Pouliot, B. Gangadharan, and A. Bani-Hashemi,“Low dose megavoltage cone beam computed tomography with an unflat-tened 4 MV beam from a carbon target,” Med. Phys. 35, 5777–5786 (2008).

8S. Flampouri, P. M. Evans, F. Verhaegen, A. E. Nahum, E. Spezi, andM. Partridge, “Optimization of accelerator target and detector for portalimaging using Monte Carlo simulation and experiment,” Phys. Med. Biol.47, 3331–3349 (2002).

9D. A. Roberts, V. N. Hansen, A. C. Niven, M. G. Thompson, J. Seco, andP. M. Evans, “A low Z linac and flat panel imager: Comparison with theconventional imaging approach,” Phys. Med. Biol. 53, 6305–6319 (2008).

10J. L. Robar, T. Connell, W. Huang, and R. G. Kelly, “Megavoltage planarand cone-beam imaging with low-Z targets: Dependence of image qual-ity improvement on beam energy and patient separation,” Med. Phys. 36,3955–3963 (2009).

Medical Physics, Vol. 39, No. 9, September 2012

5371 D. Parsons and J. L. Robar: Effect copper conversion plates low-Z target image quality 5371

11E. J. Orton and J. L. Robar, “Megavoltage image contrast with low-atomicnumber target materials and amorphous silicon electronic portal imagers,”Phys. Med. Biol. 54, 1275–1289 (2009).

12T. Connell and J. L. Robar, “Low-Z target optimization for spatial res-olution improvement in megavoltage imaging,” Med. Phys. 37, 124–131(2010).

13D. Sawkey, M. Lu, O. Morin, M. Aubin, S. S. Yom, A. R. Gottschalk,A. Bani-Hashemi, and B. A. Faddegon, “A diamond target for megavoltagecone-beam CT,” Med. Phys. 37, 1246–1253 (2010).

14D. A. Roberts, V. N. Hansen, M. G. Thompson, G. Poludniowski, A. Niven,J. Seco, and P. M. Evans, “Comparative study of a low-Z cone-beam com-puted tomography system,” Phys. Med. Biol. 56, 4453–4464 (2011).

15D. A. Roberts, V. N. Hansen, M. G. Thompson, G. Poludniowski, A. Niven,J. Seco, and P. M. Evans, “Kilovoltage energy imaging with a radiotherapylinac with a continuously variable energy range,” Med. Phys. 39, 1218–26(2012).

16B. A. Faddegon, M. Aubin, A. Bani-Hashemi, B. Gangadharan,A. R. Gottschalk, O. Morin, D. Sawkey, V. Wu, and S. S. Yom, “Com-parison of patient megavoltage cone beam CT images acquired with an un-flattened beam from a carbon target and a flattened treatment beam,” Med.Phys. 37, 1737–1741 (2010).

17D. A. Jaffray, J. J. Battista, A. Fenster, and P. Munro, “Monte Carlo studiesof x-ray energy absorption and quantum noise in megavoltage transmissionradiography,” Med. Phys. 22, 1077–88 (1995).

18E. K. Breitbach, J. S. Maltz, B. Gangadharan, A. Bani-Hashemi, C. M.Anderson, S. K. Bhatia, J. Stiles, D. S. Edwards, and R. T. Flynn, “Imagequality improvement in megavoltage cone beam CT using an imaging beamline and a sintered pixelated array system,” Med. Phys. 38, 5969–5979(2011).

19D. W. Rogers, B. A. Faddegon, G. X. Ding, C. M. Ma, J. We, andT. R. Mackie, “BEAM: A Monte Carlo code to simulate radiotherapy treat-ment units,” Med. Phys. 22, 503–524 (1995).

20D. W. O. Rogers, B. Walters, and I. Kawrakow, “BEAMnrc users manual,”NRCC Report No. PIRS-0509(A)revL (NRCC, Ottawa, 2011).

21C. M. Ma and D. W. O. Rogers, “BEAMdp users manual,” NRCC ReportNo. PIRS-0509(C)revA (NRCC, Ottawa, 2009).

22I. Kawrakow, “Accurate condensed history Monte Carlo simulation of elec-tron transport. I. EGSnrc, the new EGS4 version,” Med. Phys. 27, 485–498(2000).

23R. Rajapakshe, K. Luchka, and S. Shalev, “A quality control test for elec-tronic portal imaging devices,” Med. Phys. 23, 1237–1244 (1996).

24R. T. Droege and R. L. Morin, “A practical method to measure the MTF ofCT scanners,” Med. Phys. 9, 758–760 (1982).

25L. Feldkamp, L. Davis, and J. Kress, “Practical cone-beam algorithm,”J. Opt. Soc. Am. A 1, 612–619 (1984).

26L. Parent, J. Seco, P. M. Evans, A. Fielding, and D. R. Dance, “MonteCarlo modelling of a-Si EPID response: The effect of spectral variationswith field size and position,” Med. Phys. 33, 4527 (2006).

Medical Physics, Vol. 39, No. 9, September 2012