Assessment of the strength of individual 192 Ir seeds in ribbonsBruce Thomadsen, Larry DeWerd, Todd McNutt, Scott DeWerd, and Dan Schmidt Citation: Medical Physics 26, 2471 (1999); doi: 10.1118/1.598766 View online: http://dx.doi.org/10.1118/1.598766 View Table of Contents: http://scitation.aip.org/content/aapm/journal/medphys/26/11?ver=pdfcov Published by the American Association of Physicists in Medicine Articles you may be interested in Assaying 192 Ir line sources using a standard length well chamber Med. Phys. 29, 2692 (2002); 10.1118/1.1517046 Calibration of new high dose rate 192 Ir sources Med. Phys. 29, 1483 (2002); 10.1118/1.1487860 Experimental derivation of wall correction factors for ionization chambers used in high dose rate 192 Ir sourcecalibration Med. Phys. 29, 1 (2002); 10.1118/1.1427081 Spectra and air-kerma strength for encapsulated 192 Ir sources Med. Phys. 26, 2441 (1999); 10.1118/1.598763 A parallel plate chamber for calibration of 192 Ir LDR and HDR sources Med. Phys. 26, 2438 (1999); 10.1118/1.598762

Assessment of the strength of individual 192Ir seeds in ribbonsBruce Thomadsena)

Departments of Medical Physics and Human Oncology, University of Wisconsin, Madison, Wisconsin 53706

Larry DeWerdDepartment of Medical Physics, University of Wisconsin, Madison, Wisconsin 53706

Todd McNuttb)

Departments of Medical Physics and Human Oncology, University of Wisconsin, Madison, Wisconsin 53706

Scott DeWerd and Dan SchmidtStandard Imaging, Middleton, Wisconsin 53526

~Received 27 April 1999; accepted for publication 25 August 1999!

Assessing the strength of individual seed-type sources in ribbon assembles remains a challenge inbrachytherapy quality assurance. Geometries to measure a single source in the ribbon usually failbecause of low signals if using very thick shielding to block the radiation from the other sources, orcontributions from all the other sources if they are not shielded well. A normal well-type chamberwith partial lead shielding forming a small slot provides a differential response along the chamberaxis that, through a deconvolution/simultaneous-equations technique, sorts the contributions fromeach source, allowing the derivation of each source’s strength. ©1999 American Association ofPhysicists in Medicine.@S0094-2405~99!03311-8#

Key words: brachytherapy, quality assurance,192Ir, calibration, source strength

I. INTRODUCTION

Assaying sources used in brachytherapy form a standard ofpractice. The report of Task Group 40 of the American As-sociation of Physicists in Medicine recommends assay of10% of a batch of permanently implanted sources, and 10%or two ribbons for sources so configured.1

The simplicity of assay for a single source contrasts withthe difficulty of determining the strength of individualsources comprising a source train. A common approach mea-sures a signal for the entire ribbon wrapped in a loop in acalibrated well chamber, and simply checks the total strengthfor all the sources combined. Even determining the totalstrength in that manner requires addressing the geometricresponse of the chamber to sources positioned in locationsother than that used to calibrate the chamber. Self-shieldingand source orientation effects also increase the uncertainty inthe strength measurement. While such an approach may givesome assurance of source accuracy with a uniformly loadedribbon, modern implant techniques frequently use differen-tial loading. For nonuniform source-strength distributions,the quality assurance evaluation should verify not only theaccuracy of the source strength, but that the correct strengthsoccupy the correct locations as well.

Bernard and Dutreix describe the use of a lead block witha channel for a192Ir wire below an orthogonal hole runningbetween the source train and a detector. The latter holeserves as a collimator to limit the view of the detector to asmall section of the wire at a time to assess the wire’s linearactivity.2 While this approach could be applied directly to atrain of 192Ir sources in a ribbon, our experience with thisapproach encountered very poor correlation between themeasured signal and the individual source strengths. Lead

thick enough to reduce sufficiently the contribution fromsources other than the one aligned with the measurementhole reduces the sensitivity of the detector and compromisesthe precision of the measurement.

Modern practice of temporary implants dictates differen-tial loading. Current conceptions of quality assurance sug-gest that all aspects of treatments that could affect dosesshould undergo verification. A method to assure correctsource strength for individual radioactive seeds in ribbon re-mains elusive. Simple, straightforward techniques have notworked sufficiently well. The approach of Bernard holdspromise but requires two modifications to overcome the dif-ficulties previously encountered:~1! Increasing the signal forthe source aligned with the measurement channel, and~2!.Accounting for the contribution of the shielded sources to themeasurement. The work described in this report addressesthese issues.

II. MATERIALS AND METHODS

This investigation used a well-type ionization chamber~reentrant chamber—modified model HDR-1000, StandardImaging, Middleton, WI! as the detector system. Lead cylin-ders fill the well above and below an air-filled aperture. A3.2-mm diameter tunnel through the lead that also penetratedthe base of the unit allowed iridium ribbons to pass throughthe chamber. Figure 1 shows the chamber schematically. Theutility of a similar device for assessment of positioning forhigh dose-rate brachytherapy sources suggested that theequipment might satisfy the modification requirements toBernard’s approach mentioned above. Without the aperturein place, the well chamber exhibits a typical response tosource position along its length, with a broad plateau in the

2471 2471Med. Phys. 26 „11…, November 1999 0094-2405/99/26 „11…/2471/5/$15.00 © 1999 Am. Assoc. Phys. Med.

middle, and decreased sensitivities toward the ends. The leadshielding and aperture produce an additional differential re-sponse in the chamber superimposed upon the chamber’susual response as a function of position along the axis of thechamber. This configuration is similar to the HDR qualityassurance test tool for the chamber.3 The selection of a3.9-mm aperture maximized the signal from a 3-mm-long192Ir source while excluding most of any radiation fromneighboring sources.~Section IV addresses this issue fur-ther.! The 1.1-cm-thick lead fillers, approximately 3.5 half-value layers, reduce the signal to approximately 8% throughattenuation, although geometry and scatter produce signalsof about 30% when the edge of the source lies within 0.7 cmof the aperture.

The tests of the device used192Ir seeds in plastic ribbons.Figure 2 illustrates the configurations for the ribbons. Ribbon1, called a ‘‘standard ribbon,’’ presented a simple standardwith 12 sources spaced at centimeter intervals, center-to-center~c-c!. Ribbon 2, referred to as a ‘‘treatment ribbon,’’simulates a ribbon from a customized template implant with

a mix of source strengths and spacings. Ribbon 3 representsa ribbon for an endobronchial application, with sourcesspaced 0.5 cm c-c and differentially loaded with half-strength seeds in the center third of the source train. Twomanufacturers~Best Medical, Inc, Springfield VA, andAlpha-Omega Services, Inc., Bellflower, CA! provided iden-tical sets of source ribbons.

For the initial tests, the source ribbons were placed in avery thin polyimide tube marked in millimeter increments.This tube was extended through the tunnel, and then with-drawn by hand, stopping at each millimeter for a reading.The measurements used an electrometer~model 602, Kei-thley Instruments, Cleveland, OH! on the 10211A scale, readwith a digital multimeter~Fluke 8060 A! connected to theanalog outputs in the back of the electrometer. A pause aftereach millimeter displacement of the source tube allowed thesignal to settle before recording. All measurements excludethe background reading with no source in the chamber.

After making measurements for a ribbon, the final seed inthe ribbon was removed and passed through the test chamberin the same manner as the entire ribbon to establish thechamber response as a function of source position in thetunnel.

After completing the measurements on a test ribbon, thesources were cut apart and individually calibrated. This pro-cedure used a calibrated well chamber~model WC-2, Preci-sion Radiation Measurements, Nashville, TN!. The calibra-tion on the chamber was based on measurements using a192Irsource calibrated at the National Institute of Standards andTechnology, and the same electrometer used for this study. Astandard137Cs seed provided standardization for the wellchamber’s sensitivity between the time of the chamber’s cali-bration and that of the test seeds.

For a ribbon in a given position in the chamber, eachsource in a ribbon should contribute to the signal as per itsposition along the chamber according to the single-sourceresponse curve, and in proportion to its strength. As such, thesignal becomes the convolution of the ribbon configurationand the single-source response, with the latter serving as theconvolution kernel. Reversing the process, from the mea-sured signal as a ribbon passes through the chamber, decon-volution using the kernel should produce the ribbon configu-ration.

The kernel profile also carries the information on absolutestrength by normalizing the signal to the assayed strength ofthe seed used for the determination. Thus, the deconvolutionresults in a mapping of the position and strength of eachsource in the tested ribbon.

III. RESULTS

The broader line in Fig. 3 shows the signal for one of theribbon 1 configurations as the ribbon passed through thechamber. The general shape peaks in the middle indicatingthat seeds behind the lead still contribute to the signal as-signed to the seed aligned with the spacer.

Figure 4 displays the response of a single source passing

FIG. 1. The chamber designed for assay of seed-type sources in ribbons~an‘‘eclipse’’ chamber!. The lead cylinders and spacer provide a differentialresponse along the chamber axis.

FIG. 2. The source ribbons used for testing the chamber shown in Fig. 1.

2472 Thomadsen et al. : Strength of Individual 192Ir Seeds 2472

Medical Physics, Vol. 26, No. 11, November 1999

through the chamber. The asymmetry of the signal arisesfrom a displacement of some of the sensitive volume of thechamber by electronics and some additional extracameralvolume on one end. The contribution to the signal when thesource falls entirely behind the lead shows clearly in thisfunction. The single-source response curve was found to bereproducible between seeds to within63%. Comparisons ofthe kernels from the two manufacturers, representing twodifferent source constructions, showed no significant differ-

ence. The reproducibility for repeated trials with the samesource was better than62% through the range of the re-sponse curve.

An iterative deconvolution technique provided the actualpositions of the sources to within 0.5 mm. The true positionsdiffered from the position of the signal peaks due to thecontributions of other sources and the asymmetry of thechamber response. However, the deconvolution failed toyield the true source strengths to within65%. An alternativeapproach used the positions identified by the deconvolutionto designate locations on the raw data curve. At these loca-tions, the signals would be the sum of the products of eachsource’s strength times its kernel value. Thus, for a ribbon ofn seeds, the reading,Rk , when sourcek falls in the center ofthe space, becomes

Rk5(i

n

si3kk,i , ~1!

wheresi5the strength of the source at position I along theribbon, andkk,i5the contribution to the signal from a sourceat position k, as indicated by the kernel. The equations foreach of then sources establish a system of simultaneouslinear equations, the solution of which yields the sourcestrengths.

Table I compares the values for the source strengths de-rived from this procedure with that from the measurement inthe calibrated well chamber. The columns labeled ‘‘% Dif-ference absolute’’ simply give the difference between thesource strengths measured by the two methods as

%Difference absolute5Se2Sc

Sc, ~2!

whereSe5the strength determined with the eclipse chamberandSc5the strength measured individually in the calibratedchamber.

The columns labeled ‘‘% Difference relative’’ normalizeeach reading by the average of the sources in the given trainas determined by the given measurement technique. This re-moves any bias for the calibration to allow evaluation of therelative strengths of the sources in the train. Thus,

%Difference relative5Se /S̄e2Sc /S̄c

Sc /S̄c

, ~3!

where the bar over the quantities indicates the average of allstrengths measured with the given technique for the givenribbon.

In all cases, the row indicated as the average displays theabsolute value of the differences in the column above. Thenonabsolute value for the averages of the relative columns,by definition, would be zeros. The average discrepancy in thesource assay was less than63%, with no discrepancy ex-ceeding 6.5%.

The table shows the results for only one of the treatmentribbons. The measurements of the other ribbon used a differ-ent, computer-controlled electrometer that provides an inter-face to transfer measurements digitally to a computer. This

FIG. 3. Sample signal measured for a ribbon of pattern 1 of Fig. 2~uniformstrength seeds separated by centimeter intervals! shown as a broad line. Thenarrow line displays the signal after deconvolution using the kernel in Fig.4.

FIG. 4. The deconvolution kernel based on measuring the chamber responseto a single source.

2473 Thomadsen et al. : Strength of Individual 192Ir Seeds 2473

Medical Physics, Vol. 26, No. 11, November 1999

electrometer automatically resets the background readingwhenever the signal decreased below a built-in threshold,and subtracts the determined background from each signalmeasured. During the measurement run, the background re-set several times, reducing the central readings more thanthose for the peripheral sources. It was not until after sepa-rating the sources for the individual calibrations that thisshifting of background was noted in the data. Thus, the mea-surements for this ribbon yield no useful information—onlya lesson.

The signals for the endobronchial ribbons, with 5 mmbetween centers and 2 mm spaces between source ends,showed no discernable peaks after deconvolution~Fig. 5!.This finding reflects the Nyquist frequency,f n , associatedwith the aperture size. The 3.9-mm spacer produces a kernelwith a full width at half maximum~FWHM! of approxi-mately 4 mm when measured using a 3-mm-long seed. Inthis setting,

f n5~2 FWHM!2150.12 mm21. ~4!

This spatial frequency corresponds to a spacing of(1/0.12 mm21)58.3 mm between source centers. The spatialfrequency of the sources in the endobronchial ribbon be-comes

f endo5~period55 mm center-to-center!21

50.2 mm21.

This frequency of 0.2 mm21 exceedsf n and, thus, becomesunresolvable~i.e., there are more cycles per millimeter thanthe system can distinguish!.

IV. DISCUSSION

The experiment demonstrates that a chamber and mea-surement procedure such as described could provide assaysfor individual source strengths for192Ir seeds in ribbons for

most applications where the source centers fall about 8 mmor more apart. Such a test allows verification of differentialloading, uniformity of source strength, and a measure of cor-rect spacing.

The manual movement of the ribbons through the detectorproves excessively time consuming for the number of rib-bons commonly used in a realistic implant. As a test, mea-surements on one ribbon made use of a automated scanningsystem~model RFA 300, Therados, Sweden!. Replacing the

TABLE I. Source strengths measured with the Eclipse chamber compared to that measured individually in acalibrated well chamber. The columns marked ‘‘Relative’’ relate the discrepancy relative to the average of allstrengths, while the columns marked ‘‘Absolute’’ compare the calculated absolute strength of the source to thatassayed.

Standard ribbons

Ribbon A Ribbon B Treatment ribbon

Seednumber

% Differencerelative

% Differenceabsolute

% Differencerelative

% Differenceabsolute

% Differencerelative

% Differenceabsolute

1 0.1 0.0 0.3 21.6 22.0 23.12 3.5 3.4 1.1 20.7 1.8 0.83 6.1 6.1 21.4 23.3 25.4 26.54 1.2 1.2 21.0 22.9 22.5 23.65 26.3 26.4 2.5 0.7 2.8 1.86 21.1 21.1 1.3 20.6 20.6 21.77 2.5 2.4 22.2 24.1 5.5 4.58 25.5 25.5 3.1 1.3 1.5 0.49 22.5 22.5 21.5 23.4 20.9 22.010 22.1 22.1 22.1 24.0 0.0 0.011 3.6 3.6

Average 3.1 3.1 1.7 2.3 2.4 2.3

FIG. 5. The measurement results for an ribbon of pattern 3 in Fig. 2 simu-lating an endobronchial configuration. The measurement shows no discern-able peaks representing the sources. The maximum value simply falls nearthe center of the ribbon, and the smaller peak to the left is due to theasymmetric chamber response seen in Fig. 4.

2474 Thomadsen et al. : Strength of Individual 192Ir Seeds 2474

Medical Physics, Vol. 26, No. 11, November 1999

glass plate used with the optical densitometer with a board tosupport the chamber above the phantom tank, clamps at-tached to the densitometer’s scanning arm stretched the rib-bon through the chamber and automatically moved it 0.5 mmat a step. The system gated the signal from the chamber intothe scanning system’s computer at each step, creating thedata file of signal vs position. While the automated scan stilltook about 30 min for a ribbon, this consumed no human’stime. Such an automated system, either as a stand-alone de-vice or as an adaptation for automated scanning systems,could provide a feasible method for assuring the accuracy ofsource strength for interstitial implants.

Resolving the sources in the endobronchial ribbons posestwo problems. Differentiating the sources requires resolvingthe spaces between the seeds. For the test ribbons, this spac-ing was 2 mm, for a spatial frequency of 0.25 mm21. Re-solving this frequency requires a FWHM, and assumedly anaperture width also of 2 mm. Concentrating on resolving thespace between seeds, then, would limit the measurement ofthe seed strength to sampling parts of the source rather thanincluding all of the source at once, rendering the determina-tion sensitive to variations in activity along the source lengthand to partial volume artifacts in the calculated total strength.Additionally, many applications like the endobronchial usesources closer than the separation tested, approaching end-to-end. To resolve individual sources in these cases requiressmaller apertures still. As the aperture narrows, the signal-to-noise ratio decreases, compromising the accuracy of theactual source assay beyond the sampling problems notedabove. These considerations establish the limits for the ap-plication of this device.

The signal for these test ribbons peaked at a factor ofabout 600 times the background signal for sources withstrengths of approximately 5mGy m2h21 ~0.7 mgRaed.!.Some implants use sources one-sixth of that strength, whichstill allows for a signal-to-noise ratio of 100. Such would still

allow for assessment of source strength; however, improvedsignal-to-noise ratio would still be beneficial.

Overall, the approach described worked well for verifica-tion of source strength for seeds in ribbons with separationsof 8 mm or greater between centers~about 5 mm from end toend!. Such configurations account for most of the sourcetrains used clinically. More development remains necessaryfor verification for source trains with closer spacing. Thetechnique lends itself well to automation, having beenadapted to a common computerized beam scanner.

V. CONCLUSION

The well chamber with differential shielding using adeconvolution/simultaneous-equation technique allows assayof individual seed-type sources in ribbons to within an aver-age accuracy of 3%, with no errors greater than 6.5%. Themethod works for ribbons with seed spacing greater than 8mm center-to-center, and for uniformly and differentiallyloaded ribbons. The response of the system remained thesame regardless of the manufacturer of the sources. Whilethe method does not work directly for ribbons with spacingof less than 8 mm between source centers, this could serve asa valuable quality assurance tool for most192Ir interstitialimplant source trains.

a!Electronic mail: brthomad@facstaff.wisc.edub!Currently at ADAC/Geometrics, Madison, WI, CA.1G. J. Kutcher, L. Coia, M. Gillin, W. F. Hanson, S. Leibel, R. J. Morton,J. R. Palta, J. A. Purdy, L. E. Reinstein, G. K. Svensson, M. Weller, andL. Wingfield, ‘‘Comprehensive QA for radiation oncology: report ofAAPM Radiation Therapy Committee Task Group 40,’’ Med. Phys.21,581–618~1994!.

2M. Bernard and A. Dutreix, ‘‘Dispositif de mesure de l’activite li-neique,’’ J. Radiol. Electrol. Med. Nucl.49, 534–538~1968!.

3L. A. DeWerd, P. Jursinic, R. Kitchen, and B. R. Thomadsen, ‘‘Qualityassurance tool for high dose rate brachytherapy,’’ Med. Phys.22, 435–440 ~1995!.

2475 Thomadsen et al. : Strength of Individual 192Ir Seeds 2475

Medical Physics, Vol. 26, No. 11, November 1999