A.V.

Klyachko
Nuclear et al. / Nuclear
Instruments Instruments
and Methods andinMethods
Physicsin Physics
ResearchResearch
A 628 A 628 (2011)434–439
(2011) 434–439 434

Contents lists available atScienceDirect

Nuclear Instruments and Methods in

Physics Research A
journal homepage:www.elsevier.com/locate/nima

Dose imaging detectors for radiotherapy based on gas electron multipliers$

A.V. Klyachko a,, D.L. Friesel b, C. Kline b, J. Liechty b, D.F. Nichiporov a, K.A. Solberg a
a
Indiana University Cyclotron Operations, Bloomington, Indiana, USA b
PartTec, Ltd, Bloomington, Indiana, USA

Available online 8 July 2010 New techniques in charged particle therapy and widespread use of modern dynamic beam delivery systems demand new beam
monitoring devices as well as accurate two-dimensional dosimetry systems to verify the delivered dose distribution. We are
Keywords:
GEM detector
developing dose imaging detectors based on gas electron multipliers (GEM) with the goal of improving dose measurement
Hadron therapy linearity, position and timing resolution, and to ultimately allow pre-treatment verification of dose distributions and dose
Dose imaging delivery monitoring employing scanning beam technology. A prototype 10 10 cm 2 double-GEM detector has been tested in the
205 MeV proton beam using electronic and optical readout modes. Preliminary results with electronic cross-strip readout
demonstrate fast response and single-pixel (4 mm) position resolution. In optical readout mode, the line spread function of the
detector was found to have s¼ 0:7 mm. In both readout modes, the detector response was linear up to dose rates of 50 Gy/min,
with adequate representation of the Bragg peak in depth-dose profile measurements.
& 2010 Elsevier B.V. All rights reserved.

articleinfo abstract

1. Introduction Existing detectors and measurement techniques used in the clinical

practice of proton therapy are not well suited for dynamic dose
New methods of beam delivery in charged particle therapy such as distribution monitoring. Solutions exist, but as one-of-a-kind,
beam scanning and energy stacking are becoming increasingly popular as experimental systems [1–3]. The difficulties with monitoring and dose
more new clinical proton facilities include Intensity Modulated Proton verification in IMPT could be resolved by implementation of gaseous
Therapy (IMPT) in their specifications. Promising better coverage of amplification devices, such as the Gas Electron Multiplier (GEM) [4].
clinical targets with dose fields, the IMPT technology implies dynamic GEMs offer fast performance, robustness and flexibility in the detector
variations in beam’s lateral position, intensity and energy during the beam design, allowing for both electronic and optical readout schemes.
delivery. A typical radiotherapy treatment delivers about 2 Gy of absorbed Recently, the results of characterization of scintillating GEM detectors,
dose into a 1 l target volume and, implemented with an IMPT technique, developed for pre-treatment dose distribution verification in particle
will require 10,000 individual beams delivered in about 100 s, i.e. 100 Hz therapy, has been reported for proton [5], aparticle [6] and carbon [7]
spot delivery frequency or, with re-scanning, even 300 Hz. To control the beams.
dose in each spot to better than 5% on a daily basis, one requires a readout We are developing prototype GEM-based detectors for
and feedback system with a response time of 0:1ms. Dose distributions twodimensional dose imaging in radiotherapy with the goal of
created in IMPT are characterized by high lateral and depth dose gradients. improving dose measurement linearity, position and timing resolution,
Verification of such fields therefore requires accurate measurements of and to ultimately allow pre-treatment verification of dose distributions
lateral and depth dose profiles as well as absorbed dose. Main and dose delivery monitoring employing scanning beam technology. In
requirements for a beam monitor are: response time 0:1ms, position the present work, we report on the first results obtained in detector
resolution of 1–2mm, and dose measurement linearity 1%. For a dose prototype tests using a 200 MeV proton beam.
verification detector, sub-millimeter spatial resolution and tissue-
equivalence of the dose response are desirable.

$
Funded by: NIH.
Corresponding author.
E-mail address: [email protected] (A.V. Klyachko).

0168-9002/$ - see front matter & 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.nima.2010.07.019
2. Materials and methods narrow strips were connected together to form 4 mm wide anodes
providing a 4 4 mm2 readout pixel size. A 12 13 array of strips was
2.1. Detector setup, electronic readout read out using gated integrator cards, developed at the Indiana
University Cyclotron Facility (IUCF), and a VME-based data
The detector, schematically shown in Fig. 1, consists of a double- acquisition system. During the beam measurements, the GEM 1 and
GEM amplification structure with a copper-clad Kapton cathode and a GEM2 were operated at 350 and 340 V, respectively, with the drift,
transfer and induction fields set, respectively, at 1.5, 1.7 and 1.7

Fig. 1. Schematic diagram of the detector setup in electronic readout mode.

crossed-strip collector electrode, mounted in an air-tight aluminum kV/cm, each stage being powered from an individual power supply.
housing with thin 87 81 mm2 stainless steel windows, continuously
flushed with an Ar/CO2 (70/30%) gas mixture. The 100 100 mm2 GEM
foils produced by Tech-Etch [8] have a triangular hole pattern with 2.2. Detector setup, optical readout
equidistant holes at a pitch of 140 mm and hole diameters in the metal
layers of 70mm. The GEMs were mounted on 2.4 mm thick Rexolite The same detector housing, cathode and double-GEM amplification
frames. The multilayer two-dimensional strip anode with 340 mm structure were used in the optical readout mode, as shown in Fig. 2.
The stainless steel windows were replaced by aluminized Mylar
wide Y-strips (lower layer) and 80mm wide X-strips (upper layer),
entrance and transparent Mylar exit windows. The strip anode was
both at 400 mm pitch, is similar in design to the COMPASS [9] removed and the bottom copper layer of GEM 2 served as a charge
readout electrode. It was also produced by Tech-Etch. Groups of 10 collector electrode, its signal was read out by a IUCF-designed
 A.V. Klyachko et al. / Nuclear Instruments and Methods in Physics Research A 628 (2011) 434–439 436

recycling integrator card. The detector volume was continuously 2.3. Experimental setup
flushed with an Ar/CF4 (95/5%) gas mixture. Both GEMs were
operated at 300 V, with the same drift and transfer fields as described in
Section 2.1. The light produced by the electron avalanches was
detected by an SBIG ST-6 astronomical camera with thermoelectric
cooling from the Santa Barbara Instrument Group [10] and a Tamron
CCTV CS zoom lens. The camera was positioned away from the beam
and was shielded with lead and concrete blocks to reduce radiation
damage to the CCD sensor. The light path was enclosed in light-tight
shielding made from a black aluminum foil. The Texas Instruments
TC-241 CCD sensor, with a quantum efficiency of 62% at 650 nm, is
well matched to the emission spectrum of the Ar/CF 4 gas mixture [11].
The sensor has 375 241 pixels with dimensions of 23 27 mm2, which
translates to 0.36 0.42 mm 2 at the GEM2 location. During the
measurements, the camera was cooled to 30 1C. The images were read
out by the camera’s native SBIG software and analyzed using SBIG
[10] and ImageJ [12] software packages.
 The detector was tested in the IUCF Proton Dosimetry Test Facility
with a 205MeV proton beam. The experimental setup is shown in Fig. 3.
The Ionization Chamber Beam Monitor (ICBM) was used to measure
beam intensity and beam profile, and to control the beam delivery,
terminating the beam upon the delivery of a pre-set dose. The dose rate
was varied by changing the beam current in the cyclotron. A 2.4mm thick

436 A.V. Klyachko et al. / Nuclear Instruments and Methods in Physics Research A 628 (2011) 434–439

Fig. 2. Schematic diagram of the detector setup in optical readout mode.

 A.V. Klyachko et al. / Nuclear Instruments and Methods in Physics Research A 628 (2011) 434–439 438

Fig. 3. Schematics of the experimental setup. 1—ICBM; 2—scatter foil; 3—collimator; 4—acrylic phantom; 5—GEM detector; 6—Markus chamber; 7—therapy dosimeter/
copper scatter foil spreaded the beam to about 6cm diameter FWHM at the determine the light yield by integrating over an area of interest identical
detector location, also degrading the beam energy to 198MeV. For for all measurements.
depthdose measurements (see Section 3), the thickness of an acrylic picoampermeter; 8—VME DAQ; 9—mirror; 10—CCD camera.
phantom placed in front of the detector was varied by adding/ removing
acrylic sheets, with 1–10mm steps. The dose linearity tests were carried
out without any phantom material. Brass collimators were used to shape
the beam impinging on the detector. A commercial Markus ionization
chamber with NISTtraceable calibration in absorbed dose to water (PTW,
Model TN23343) with a 5.3mm diameter active area, positioned behind
the detector, was used to estimate the dose rate.

2.4. Backgrounds and data analysis

In the electronic readout mode, the main sources of background are the
noise pickup in the detector and in the cables. The background was
measured before each beam measurement and subtracted from the data. In
the optical readout mode, the main background sources are: (a) camera
offset; (b) dark current (noise); (c) ambient light. To take into account the
backgrounds (a–c), an ambient light image was taken without beam and
with the same exposure time as in beam measurements, and then
subtracted from the images taken with the beam. Corrections for additional
possible backgrounds (such as interactions of scattered beam and
secondary particles with the camera’s sensor, and scintillations in the
detector gas and exit window caused by beam particles) were estimated at
less than 1% total [5] and therefore were not applied. The images have
been processed offline to correct for background and to remove extremely Fig. 4. Dose-rate response of the detector in electronic readout mode. The lines are
linear fits to the data below 40 Gy/min.
hot and cold isolated pixels, using the routines provided with the SBIG
camera. The images then have been analyzed with ImageJ software to
 Fig. 5. Depth-dose response of the detector in electronic readout mode. Insert:
expanded view of the Bragg peak.

3. Results and discussion Bragg peaks from detector and Markus chamber. In future, we intend to
increase the readout channel count and to use a wide aperture parallel
In the electronic readout mode, to estimate the position resolution plate ionization chamber instead of the Markus chamber for
of the detector, we performed series of measurements with different comparative measurements.
diameter collimators. The results obtained in the proton beam and with In the optical readout mode, collimators of different shapes were used
radioactive sources suggest that the spatial resolution is close to the to estimate the spatial resolution of the detector. An image of a 1.4 20
single-pixel size, i.e. 4 mm. The dose-rate response has been measured mm2 collimator obtained with 5 nA beam and 3 s exposure is shown in Fig.
with a 20 mm diameter collimator. The summed response of three X- 6 (insert) together with the light
and three Y-strips corresponding to the central part of the beam is Fig. 6. The 1.4 20 mm2 collimator image (insert) and the light intensity profile along the
one-pixel wide horizontal line. The data are fitted with a superposition of 3 Gaussians.
shown in Fig. 4.
To evaluate the energy dependence of the detector’s output, we
measured the depth-dose response of the detector, with the beam
current of about 6 nA corresponding to dose rate of 12Gy=min without
any material in the phantom, and up to 40Gy=min at the Bragg peak.
The depth-dose curves (summed response of six X- and six Y-strips)
are shown in Fig. 5. The curves are normalized to 1 at zero phantom
thickness. The acrylic thickness is converted to water-equivalent depth.
Note that the detector and the Markus chamber have sensitive areas of
different size and both detect only a small (and different) portion of all
particles coming out of the phantom, so one should not expect identical
438 A.V. Klyachko et al. / Nuclear Instruments and Methods in Physics Research A 628 (2011) 434–439
 A.V. Klyachko et al. / Nuclear Instruments and Methods in Physics Research A 628 (2011) 434–439 440

Fig. 8. Dose rate response of the detector in optical readout mode. The line is a linear fit to
the data below 55 Gy/min.

Fig. 7. Top—an image of a 20 mm diameter collimator. Bottom—an image profile along

the one-pixel wide horizontal line (shown in the upper part).
Fig. 9. Bragg curves measured in optical readout mode. Compared are the light signal, the
charge signal from the last GEM foil, and the Markus chamber signal.
Insert—expanded view of the Bragg peak.

intensity profile along a single-pixel line. The main peak has been fitted were taken with 3 s exposures, shorter than the duration of the beam
with a Gaussian curve with s¼ 0:7mm. An image of a 20 mm diameter delivery. The depth-dose measurements in the optical readout mode were
collimator is shown in Fig. 7 as well as a light intensity profile along a carried out with a 50 mm diameter collimator and a beam current of 5 nA.
one-pixel wide line (insert). We attribute the 3% variations of intensity on Depth-dose curves, shown in Fig. 9, have been normalized to 1 at zero
the flattop to GEM gain nonuniformities caused by hole size variations. depth. At the peak, the charge signal exceeds the light signal by about
The detector dose-rate response in optical readout mode, measured with 3.5%. One of the reasons for this might be the difference of the integration
the same 20 mm diameter collimator, is shown in Fig. 8. Beam images areas for charge and light signals.
4. Conclusions

We have developed a prototype detector system for twodimensional

dose imaging in hadron therapy based on a doubleGEM amplification
structure, using either electronic or optical readout. In both modes, the
detectors exhibit a linear dose rate response up to about 50 Gy/min and
reproduce the Bragg peak in depth-dose measurements. In electronic
readout mode, a position resolution of 4 mm (single-pixel) was
observed. We expect that, using a multi-pad readout electrode with
smaller pitch, the position resolution of the detector can be
significantly improved, at the cost of a significant increase in the
number of readout channels. In the optical readout mode, the line
spread function of the detector was found to have s¼ 0:7 mm. The
position resolution in this mode also can be improved by using a higher
pixel count CCD camera.
The GEM-based detectors are promising candidates for the creation
of two kinds of dosimetry systems: one, with electronic readout, would
be a fast (timing resolution in the microsecond range), moderate spatial
resolution (1–2mm, limited by the cost of electronics) dose imaging
detector for online scanning beam monitoring. Such a detector, with
cross-strip readout, would also be a good candidate for low-rate
applications, such as proton tomography. Another detector system, with
optical readout, would be a slower, moderately priced detector with
sub-millimeter spatial resolution suitable for the dose distribution
verification and for quality assurance measurements in hadron therapy.

Acknowledgements

The authors would like to thank Dr. V. Anferov and Dr. V.

Derenchuk for support and assistance in carrying out the measurements
with the proton beam, as well as for valuable discussions. This work
was partially supported by NIH SBIR Grant 1R43CA13791-01A1.

References

[1] J.M. Schippers, et al., J. Phys. Conf. Ser. 41 (2006) 61.

[2] M.T. Gillin, et al., Med. Phys. 37 (2010) 154.
[3] H.-J. Borchert, et al., Nowotwory J. Oncol. 58 (2008) 62e.
[4] F. Sauli, Nucl. Instr. and Meth. A 386 (1997) 531.
[5] E. Seravalli, et al., Phys. Med. Biol. 54 (2009) 3755.
[6] E. Seravalli, et al., IEEE Trans. Nucl. Sci. NS-54 (2007) 1271.
[7] E. Seravalli, et al., Phys. Med. Biol. 53 (2008) 4651.
[8] Tech-Etch Corp, 45 Aldrin Rd, Plymouth, MA 02360.
[9] C. Altunbas, et al., Nucl. Instr. and Meth. A 490 (2002) 177.
[10] Santa Barbara Instrument Group, 147-A Castilian Drive, Santa Barbara, CA 93117.
[11] M.M.F.R. Fraga, et al., Nucl. Instr. and Meth. A 504 (2003) 88.
[12] W.S. Rasband, ImageJ, U.S. National Institutes of Health, Bethesda, Maryland, USA
/http://rsb.info.nih.gov/ij/S, 1997–2008.