Ionization chamber volume averaging effects in dynamic intensity modulated radiationtherapy beamsDaniel A. Low, Parag Parikh, James F. Dempsey, Sasha Wahab, and Saiful Huq Citation: Medical Physics 30, 1706 (2003); doi: 10.1118/1.1582558 View online: http://dx.doi.org/10.1118/1.1582558 View Table of Contents: http://scitation.aip.org/content/aapm/journal/medphys/30/7?ver=pdfcov Published by the American Association of Physicists in Medicine Articles you may be interested in Dosimetric investigation of breath-hold intensity-modulated radiotherapy for pancreatic cancer Med. Phys. 39, 48 (2012); 10.1118/1.3668314 Development and commissioning of a multileaf collimator model in Monte Carlo dose calculations for intensity-modulated radiation therapy Med. Phys. 33, 770 (2006); 10.1118/1.2170598 Ionization chamber-based reference dosimetry of intensity modulated radiation beams Med. Phys. 31, 2454 (2004); 10.1118/1.1781333 Dose linearity and uniformity of a linear accelerator designed for implementation of multileaf collimation system-based intensity modulated radiation therapy Med. Phys. 30, 2253 (2003); 10.1118/1.1592640 MAGIC-type polymer gel for three-dimensional dosimetry: Intensity-modulated radiation therapy verification Med. Phys. 30, 1264 (2003); 10.1118/1.1576392

Ionization chamber volume averaging effects in dynamic intensitymodulated radiation therapy beams

Daniel A. Lowa) and Parag ParikhDepartment of Radiation Oncology, Mallinckrodt Institute of Radiology, Washington University School ofMedicine, St. Louis, Missouri 63110

James F. DempseyDepartment of Radiation Oncology, University of Florida, Gainesville, Florida

Sasha WahabDepartment of Radiation Oncology, Mallinckrodt Institute of Radiology, Washington University School ofMedicine, St. Louis, Missouri 63110

Saiful HuqThomas Jefferson University, Philadelphia, Pennsylvania

~Received 25 November 2002; accepted for publication 25 April 2003; published 24 June 2003!

The commercial cylindrical ionization chamber ionization integration accuracy of dynamicallymoving fields was evaluated. The ionization chambers were exposed to long~14 cm!, narrow~0.6,1.0, 2.0, and 4.0 cm! 6 MV and 18 MV fields. Rather than rely on the linear accelerator toreproducibly scan across the chamber, the chambers were scanned beneath fixed portals. A water-equivalent phantom was constructed with cavities that matched the chambers and placed on acomputer-controlled one-dimensional table. Computer-controlled electrometers were utilized incontinuous charge integrate mode, with 10 samples of the charge, along with time stamps, acquiredfor each chamber location. A reference chamber was placed just beneath the linear accelerator jawsto adjust for variations in linear accelerator dose rate. The scan spatial resolution was selected toadequately sample regions of steep dose gradient and second spatial derivative~curvature!. A fixedmeasurement in a 10310 cm2 field was used to normalize the profiles to absolute chamber re-sponse. Three ionization chambers were tested, a microchamber (0.009 cm3), a Farmer chamber(0.6 cm3) and a waterproof scanning chamber (0.125 cm3). The larger chambers exhibited severeunder-response at the small field’s centers, but all of the chambers, independent of orientation,accurately integrated the ionization across the scanned portal. This indicates that the tested ioniza-tion chambers provide accurate integrated charges in regions of homogeneous dose regions. Partialintegration~less than the field width plus the chamber length plus 2 cm!, yielded integration errorsof greater than 1% and 2% for 6 MV and 18 MV, respectively, with errors for the Farmer chamberof greater than 10% even for the 4 cm wide field. ©2003 American Association of Physicists inMedicine. @DOI: 10.1118/1.1582558#

Key words: radiation therapy, intensity modulated radiation therapy, ionization chamber, dynamicmultileaf collimation

INTRODUCTION

The complex fluence distributions required for intensitymodulated radiation therapy~IMRT! typically use large num-bers of small radiation portals, either delivered as a sequenceof fixed subportals, or as a dynamic moving sequence ofcontinuously changing portal shapes. Unlike conventionalconformal therapy, IMRT validation currently relies on directmeasurements and subsequent comparison with calculateddose distributions. None of the dosimeters in clinical use areideal for this validation; each has limitations that must beaddressed when the measured doses are used.

The reason that IMRT is so attractive is that it can pro-duce dose distributions that conform to tumors and criticalstructures in three dimensions. The distribution shapes arecomplex so adequately characterizing them requires multidi-mensional dosimeters. Radiographic film1–6 has become the

de factostandard for two-dimensional dose measurements,but it suffers from poorly characterized beam spectral effectsthat may affect the dose response as a function of depth andfield size. Quantitative use of radiographic film also requiresthe acquisition of a sensitometric curve, and accuratedensitometry.7 Even when the proper techniques are utilized,dose distribution measurement accuracy has not been shownto be better than 5%, and our institutional experience hasbeen that absolute dose measurements using radiographicfilm can disagree with ionization chamber measurements byup to 10% even in low dose gradient regions. Independentmeasurements using other dosimeters~e.g., radiochromicfilm and thermoluminescent dosimeters! have agreed withthe ionization chamber results, indicating that the radio-graphic film data is suspect.

Because of radiographic film’s limitations, quantitative,absolute dose measurements are required to provide an ab-

1706 1706Med. Phys. 30 „7…, July 2003 0094-2405 Õ2003Õ30„7…Õ1706Õ6Õ$20.00 © 2003 Am. Assoc. Phys. Med.

solute dose normalization. Ionization chambers are bench-mark dosimeters that can be calibrated using NIST-traceablestandards.8 The use of ionization chambers for IMRT mea-surements necessitates that the chamber be placed in a singlelocation while the entire IMRT dose distribution is delivered,taking up to 10 minutes. This limitation means that measure-ments using ionization chambers are made at a few points atmost, although simultaneous use of multiple chambers canincrease the efficiency and therefore the number of pointsmeasured per dose distribution. Benchmark measurementsusing ionization chambers in phantoms are principally madeusing cylindrical ionization chambers.

One advantage of cylindrical ionization chambers is therelative independence of response as a function of incidentbeam angle when the beam direction is orthogonal to thechamber longitudinal axis. This response independencemakes cylindrical ionization chambers very convenient forverifying doses delivered using coplanar beam geometries.

One difficulty with ionization chambers is that the mea-sured dose can be perturbed by volume averaging over theirrelatively large active volume. As has been demonstrated instereotactic radiosurgery dosimetry measurements,9–11 sig-nificant dose measurement errors will occur when measuringdose for small fields. This is due, in part, to the relative sizeof the ionization chamber cross-sectional dimension with re-spect to the radiation portal size. Other effects, such as varia-tions in electron stopping powers and the electron fluenceperturbation across the field can affect the measured dose inthe center of a stereotactic beam.12 Unlike stereotactic radio-surgery, IMRT dose measurements are concerned with time-integral dynamically delivered doses, typically measured inregions for which the total delivered dose distribution is rela-tively homogeneous. We conducted experiments to deter-mine if ionization chambers yield the correct integral ioniza-tion for sliding-window portals as a function of the chambersize, orientation and radiation field size.

METHODS AND MATERIALS

IMRT fluence distributions are typically delivered using aseries of small radiation portals, either with multileaf colli-mators~MLC! moving during radiation delivery~dynamic,DMLC! or stationary between fixed portals~static, SMLC!.Both SMLC and DMLC field have the potential for causingvolume-average dose artifacts. We studied beam geometriessimilar to DMLC field delivery. Rather than relying on thedynamic accelerator movement to consistently deliver thedose, the IMRT dose delivery was simulated by scanning theionization chamber beneath a fixed portal, created using thesecondary linear accelerator collimators. The use of a fixedportal shape allowed a consistent portal size to be scannedusing a wide variety of ionization chambers, orientations,and beam energies. An in-house fabricated one-dimensionalcomputer controlled ionization chamber scanner, utilizing acustom-fabricated water-equivalent plastic~Solid Water,Gammex RMI, Middleton, WI! phantom, was developed forthis study. Custom water-equivalent inserts were fabricatedfor each ionization chamber that accurately matched the ex-

ternal chamber dimensions. Figure 1 shows the phantom andscanning table. The ionization chambers were placed at100.5 cm from the source~based on phantom geometry con-straints! and at a depth of 7.4 cm in water-equivalent plastic.

The ionization chambers were scanned using a digitallycontrolled stepping motor~0.01 mm/step! across 14 cm longfields of 4 cm, 2 cm, 1 cm, and 0.6 cm nominal widths andmeasurements were obtained at fixed positions across thefields. The chambers were scanned both longitudinally~ion-ization chamber longitudinal axis perpendicular to long axisof the field! and transversely~ionization chamber longitudi-nal axis parallel to long axis of the field!. Commercial dataacquisition software~Labview, National Instruments, PaloAlto, CA! and data acquisition hardware were used to simul-taneously gate identical electrometers~6514, Keithley,Cleveland, OH! to read and record the charge at each loca-tion across the beam portal. The linear accelerator was oper-ated at 600 MU min21. The electrometers were set to ac-quire a charge measurement each 60 ms without rezeroingthe electrometer. Therefore, the measured quantities were aseries of increasing collected charge and time-stamps, fromwhich the ionization chamber current was calculated. To im-prove measurement accuracy, especially in low-dose regions,10 charge measurements were obtained at each chamber po-sition. The collected charges were fit to linear functions intime and the first-order time derivative used to determine theaverage ionization current.

Three ionization chambers were investigated~Fig. 2,Table I!, a 0.009 cm3 microchamber~A14, Exradin, Lisle,IL !, a 0.65 cm3 Farmer chamber~A12, Exradin, Lisle, IL!,and the 0.125 cm3 ionization chamber we typically use forIMRT dose measurements~N31005, PTW-Freiburg,Freiburg, Germany!. Measurements of continuous beam pro-files were approximated by acquiring the ionization currentat closely spaced points. The measurement point spacing wasreduced at regions of large profile curvature~second-orderspatial derivative!. These regions lay on either side of thesteep gradient regions. For simplicity, the entire profile be-tween the penumbra tails was scanned at the higher spatialdensity. In the regions of high profile curvature or highdoses, the point spacing was 0.5 mm~2 and 4 cm fields! or0.25 mm~0.6 and 1 cm fields! and in the low-dose regions

FIG. 1. Ionization chamber phantom and computer-controlled stepping mo-tor scanner.

1707 Low et al. : Ionization chamber response to IMRT 1707

Medical Physics, Vol. 30, No. 7, July 2003

the spacing was 2 mm. The Farmer chamber was scannedacross a minimum of 12 cm and the other chambers werescanned across a minimum of 10 cm. The electrometer read-ing collection, and the phantom position was controlled us-ing commercial software~LabView, National Instruments,Palo Alto, CA!.

A reference ionization chamber~N31005, PTW-Freiburg,Freiburg, Germany! with a 5 mm brass buildup cap was

placed in the radiation beam just downstream of the collima-tors and its ionization current used to correct for variations inthe linear accelerator dose rate by dividing the scanningchamber current by the reference chamber current. The ref-erence chamber was placed in the rectangular field such thatit was at least 5 cm from the center of the portal to avoidperturbing the profile measurements. The ratio of ionizationcurrents yielded a profile across the portal with an arbitrarynormalization. The profile quantity that was desired was theionization per delivered monitor unit, so the profile was nor-malized at the field center by placing the scanning ionizationchamber at the center of the field and integrating its chargefor a fixed number of monitor units.

Even the microchamber had some volume averaging. Todetermine the radiation profile in the absence of volume av-eraging, the volume-averaging spread function of thetransverse-mounted microchamber, which had been previ-ously described by Dempseyet al.13 as a Gaussian distribu-tion with a full-width at half maximum of 1.9 mm, was de-convolved from the transverse-mounted microchamber-measured profiles. The deconvolved profile was used insubsequent comparisons and is labeled here as the ‘‘actual’’profile.

The magnitudes of the monitor-unit normalized profileswere proportional to the ionization chamber sensitivity. Tostudy the effects of volume averaging on the profile measure-ments, the sensitivity differences caused by the varied ion-ization chamber volumes had to be removed. The chamberswere irradiated at the center of a relatively large radiationfield (10310 cm2) and the collected ionization per monitorunit used to normalize the measured profiles. Variations inthe resulting profiles were assumed to be due entirely to vol-ume averaging, allowing the profiles to be directly com-pared. Effects of different leakage currents were not ad-dressed, but a recent manuscript by Leybovichet al.14 showsthat, even for small ionization chambers, the leakage contri-bution is a strong function of the quality of the electrometer.We tested the electrometer and chamber leakage by integrat-ing charge with the radiation beam off and the measuredleakage was negligible. Polarity effect was not investigated.

Integrating the normalized profiles showed the extent ofvolume averaging on the distance-integrated profile. To de-termine if the use of relatively large ionization chambersintroduced an error in the integrated charge measurements,the normalized profiles were integrated across the scanned

FIG. 2. Ionization chambers used for measurements.~a! PTW waterproofscanning ionization chamber.~b! Exradin model A14 microchamber.~c!Exradin Farmer chamber.

TABLE I. Characteristics of tested ionization chambers.

Name PTW–N31005 Exradin A12 Exradin A14

Type Cylindrical ion chamber Farmer chamber MicrochamberCollecting volume 0.125 cm3 0.65 cm3 0.009 cm3

Distance to centroid ofcollecting volume

7.25 mm 12.9 mm 2 mm

Outside diametersensitive region

5.5 mm 7.1 mm 6.0 mm

Effective length~per company!

6.5 mm 21.1 mm 0 mm

1708 Low et al. : Ionization chamber response to IMRT 1708

Medical Physics, Vol. 30, No. 7, July 2003

distance for each ionization chamber and orientation. Be-cause the complete normalized profiles were collected, wealso simulated the effects of variable scanning lengths on thecharge integration accuracy. This was equivalent to investi-gating the chamber response to relatively small homoge-neous dose regions delivered using narrow-window DMLC.

RESULTS

Examples of the measured profiles for the 1 cm widefield, are shown in Figs. 3~a! and 3~b! for 6 MV and 18 MV,respectively. The profiles include the microchamber~trans-verse orientation!, Farmer chamber~longitudinal orientation!and PTW chamber~longitudinal orientation!. In each case,the normalization is relative to the response for the 10310 cm2 field, so the values at the center of the scans~off-axis distance50) are the measured relative output factors.Variations in the measurements are assumed to be due tovolume averaging. The profiles are profoundly affected byvolume averaging. The profile obtained from the Farmerchamber is wider and has a smaller peak magnitude than theother ionization chambers. As expected, these data show that

use of the Farmer chamber in these beam geometries forstatic measurements would yield significant dose errors.

The ratio of measured to actual relative output factors areshown in Fig. 4. As described above, the actual profile wasobtained by deconvolving the transverse microchamber pro-file with the previously determined response function. Theresulting profile at the field center was assumed to be theactual output factor. A ratio of measured to actual outputfactors of 1.0 indicated that the output factor was accuratelymeasured by the ionization chamber. As shown in Fig. 4,most chambers yielded significant output factor measure-ment errors for the smaller field sizes, but all were accuratefor the 4 cm wide field. These values are consistent with thevalues reported by Serago.10

The integrated profiles are shown in Fig. 5 as a functionof integration length. These curves show the normalized pro-files integrated for a selected distance, placed symmetricallywith respect to the field center. For example, a 40 mm scan-ning distance corresponds to integrating the profile from220 mm to120 mm. As the scanning distance goes to 0, theintegral approaches the measured to actual output factor ratioshown in Fig. 4. As the scanning distance approaches themaximum measured distance, the entire radiation field is as-sumed to have been scanned. Even though there is still col-limator leakage and scattered dose well beyond the portal,because we are evaluating the integral of the ratio of mea-sured to actual profiles, the integrals approach an asymptoticvalue.

For sufficiently large scan distances, all chambers respondaccurately~average 1.004, standard deviation 0.008, maxi-mum 1.027, minimum 0.992!. When smaller integration dis-tances are used, the penumbra is incompletely integrated,and the measured integral is smaller than the actual integral.As shown in Fig. 3, volume averaging underestimates theactual profile near the field center, but overestimates beyondthe field edge. Because in this study these integrals are con-ducted from the field center outward, they always include the

FIG. 3. Example profiles for~a! 6 MV and ~b! 18 MV for the 1 cm wideportal.

FIG. 4. Central axis measured output factors for the rectangular 6 MV fieldsas a function of field width~each field was 14 cm long! for each testedionization chamber and orientation. The lines are guides to the eye.

1709 Low et al. : Ionization chamber response to IMRT 1709

Medical Physics, Vol. 30, No. 7, July 2003

underestimated portion of the profile, and are therefore lessthan 1.0 within experimental error. As the integration dis-tance extends to the field edges, the underestimation is maxi-mized. As the integration distance goes beyond the fieldedges, the measured profile is overestimating the actual pro-

file, and this begins to compensate for the underestimationwithin the field boundary. As the integration distance extendssufficiently far, the over-response cancels the under-responseand the integral is correctly measured.

As a rule of thumb, the integrated ionization measurement

FIG. 5. Relative ionization integratedacross rectangular fields of 0.6, 1.0,2.0, and 4.0 cm width for the testedionization chambers and scanning ori-entations. The results are presented asa function of total scanning lengthacross the field~symmetrically posi-tioned relative to the field center!. ~a!6 MV, ~b! 18 MV.

1710 Low et al. : Ionization chamber response to IMRT 1710

Medical Physics, Vol. 30, No. 7, July 2003

error was uniformly less than 1% and 2% if the scan distancewas greater than the sum of the scanning field size and cham-ber length by 2 cm~roughly the extent of the penumbrabeyond the field edge! for 6 MV and 18 MV, respectively.The maximum integration error was experienced when thechamber was scanned the distance corresponding to the fieldwidth, with errors of at least 3% for the 0.125 cm3 andFarmer chamber, even for the 4 cm wide field.

CONCLUSIONS

The influence of volume averaging on dynamically deliv-ered dose distributions is profound and can yield significantdose measurement errors. These errors can be managed byassuring that the dose measurement points are in regions ofhomogeneous integrated dose. This study measured the ion-ization chamber response to rectangular scanning portals tomimic dynamically delivered IMRT portals. Instead of creat-ing a region of homogeneous dose by scanning a rectangularportal across a large area and measuring the dose with a fixedionization chamber, we used a fixed portal and scanned thechamber. Assuming that creating a region of homogeneousdose using a rectangular portal is an appropriate model for aclinical IMRT treatment plan with a large homogeneous doseregion, the results of these measurements can be extrapolatedto clinical IMRT treatment conditions.

The results of this study indicate that accurate ionizationintegration will take place if the homogeneous dose region issufficiently large. As a rule of thumb, the width of the ho-mogeneous dose region should be larger than the ionizationchamber cross-sectional length. Therefore, using cylindricalionization chambers for coplanar IMRT dose distributionvalidation will provide accurate results if the homogeneousdose region is sufficiently large. Volume-averaging errorsmay be significant for smaller homogeneous dose regions orregions with heterogeneous dose distributions.

The results were valid only for the ionization chambersand the electrometers we tested. We did not detect significantstem effects which would have appeared as asymmetries inthe measured longitudinal profiles, nor was there significantleakage current for the chamber-electrometer combinationswe tested. However, this may not be true for other chambersand electrometers. For example, Leybovichet al.14 recentlypublished an examination of ionization chamber responses toIMRT fields and noted that some electrometers can haveleakage values that cause significant dose measurement er-rors. While they showed that these can be corrected, we feelthat based on the availability of accurate electrometers, themore appropriate technique is to use a suitable electrometer.

This study did not examine the integration accuracy fornon-normal beam incidence angles, so the conclusions arerelevant for coplanar beam delivery when the ionizationchamber axis is normal to the beam delivery plane. We areexpanding the study to investigate noncoplanar ionizationchamber response.

ACKNOWLEDGMENTS

This work was supported in part by a corporate grant fromComputerized Medical Systems and by NIH R01 84409.

a!Electronic mail: low@castor.wustl.edu1J. F. Williamson, F. M. Khan, and S. C. Sharma, ‘‘Film dosimetry ofmegavoltage photon beams: a practical method of isodensity-to-isodosecurve conversion,’’ Med. Phys.8, 94–98~1981!.

2J. I. Hale, A. T. Kerr, and P. C. Shragge, ‘‘Calibration of film for accuratemegavoltage photon dosimetry,’’ Med. Phys.19, 43–46~1994!.

3S. E. Burch, K. J. Kearfott, J. H. Trueblood, W. C. Sheils, J. I. Yeo, andC. K. Wang, ‘‘A new approach to film dosimetry for high energy photonbeams: lateral scatter filtering,’’ Med. Phys.24, 775–783~1997!.

4J. L. Robar and B. G. Clark, ‘‘The use of radiographic film for linearaccelerator stereotactic radiosurgical dosimetry,’’ Med. Phys.26, 2144–2150 ~1999!.

5C. Danciu, B. S. Proimos, J. C. Rosenwald, and B. J. Mijnheer, ‘‘Varia-tion of sensitometric curves of radiographic films in high energy photonbeams,’’ Med. Phys.28, 966–974~2001!.

6N. Suchowerska, P. Hoban, A. Davison, and P. Metcalfe, ‘‘Perturbation ofradiotherapy beams by radiographic film: measurements and Monte Carlosimulations,’’ Phys. Med. Biol.44, 1755–1765~1999!.

7J. F. Dempsey, D. A. Low, A. S. Kirov, and J. F. Williamson, ‘‘Quantita-tive optical densitometry with scanning-laser film digitizers,’’ Med. Phys.26, 1721–1731~1999!.

8C. Martens, ‘‘The value of the PinPoint ion chamber for characterizationof small field segments used in intensity-modulated radiotherapy,’’ Phys.Med. Biol. 45, 2519–2530~2000!.

9C. McKerracher, ‘‘Assessment of new small-field detectors againststandard-field detectors for practical stereotactic beam data acquisition,’’Phys. Med. Biol.44, 2143–2160~1999!.

10C. F. Serago, ‘‘Tissue maximum ratios~and other parameters! of smallcircular 4, 6, 10, 15 and 24 MV x-ray beams for radiosurgery,’’ Phys.Med. Biol. 37, 1943–1956~1992!.

11X. R. Zhu, ‘‘Total scatter factors and tissue maximum ratios for smallradiosurgery fields: Comparison of diode detectors, a parallel-plate ionchamber and radiographic film,’’ Med. Phys.27, 472–477~2000!.

12K. A. Paskalev, J. P. Seuntjens, H. J. Patrocinio, and E. B. Podgorsak,‘‘Physical aspects of dynamic stereotactic radiosurgery with very smallphoton beams~1.5 and 3 mm in diameter!,’’ Med. Phys. 30, 111–118~2003!.

13J. F. Dempsey, D. A. Low, S. Mutic, J. Markman, A. S. Kirov, G. H.Nussbaum, and J. F. Williamson, ‘‘Validation of a precision radiochromicfilm dosimetry system or quantitative two-dimensional imaging of acuteexposure dose distributions,’’ Med. Phys.27, 2462–2475~2000!.

14L. B. Leybovich, A. Sethi, and N. Dogan, ‘‘Comparison of ionizationchambers of various volumes for IMRT absolute dose verification,’’ Med.Phys.30, 119–123~2003!.

1711 Low et al. : Ionization chamber response to IMRT 1711

Medical Physics, Vol. 30, No. 7, July 2003