British Journal of Radiology (2004) 77, 426-432
© 2004 British Institute of Radiology
doi: 10.1259/bjr/91138314
Towards in vivo TLD dosimetry in mammography
H M Warren-Forward, BSc (Hons), PhD1 and
L Duggan, BSc (Hons), PhD2
1 University of Newcastle, School of Health Sciences (Medical Radiation Science), Callaghan NSW 2308 and 2 Newcastle Mater Misericordiae Hospital, Department of Radiation Oncology, Waratah NSW 2298, Australia
 |
Abstract
|
|---|
While phantoms are used for quality control assessment of the mammography unit, in vivo dose measurements are necessary to account for the variation in size and composition of the female breast. The use of thermoluminescent dosimeters (TLDs) in mammography has been limited due to TLD visibility. The aim of this current investigation was to access the suitability of a paper-thin LiF:Mg,Cu,P TLD (GR-200F) for in vivo dosimetric mammography measurements. The visibility of GR-200F has been directly compared with LiF:Mg,Cu,P TLDs (GR-200A) using a number of commercially available phantoms. The phantoms of thickness 2–5 cm were imaged over the range of tube potentials (24–28 kVp) used clinically. Both types of TLD were placed on the surface of the phantoms allowing assessment of visibility, entrance surface dose (ESD) and field homogeneity. In vivo assessment of ESD and visibility was also carried out on a volunteer undergoing a routine mammography examination. The positions of the GR-200F TLDs were not identified either on the image of the Leeds TOR(MAM) phantom or the patient mammograms. The average ESD for the Leeds phantom was 8.8 mGy, while the patient ESD was 13 mGy. It is now possible to perform in vivo measurements with the potential of increasing the accuracy of the doses measured for women that do not conform to a standard breast thickness or density.
|
Introduction
|
|---|
Mammography is considered to be the most effective method for early detection of small malignant lesions in the female breast. Detection of mammary cancer at an early stage improves the probability of survival considerably [1]. Due to this benefit, the use of mammography for the screening of predominantly healthy women has been classified as broadly justified for a certain group of women by the International Commission on Radiological Protection (ICRP) [2]. However, there is still a need to strike a compromise between high image quality to maximize cancer detection and low radiation dose to minimize cancer induction, as predominantly it is asymptomatic women who are being examined. Of the 600 000 women that have been estimated to be screened annually in Australia [3] some 3240 fatal cancers will be detected [4]. This can be compared with the estimated 24 cancers that are induced [4]. Even though the benefits clearly outweigh the risk, any use of ionizing radiation for diagnostic imaging requires careful thought so that the maximum benefit is obtained. Knowledge of the dose delivered is useful as it allows the evaluation of risk to the patient and allows comparison of different imaging techniques and equipment.
For diagnostic examinations, the entrance surface dose (ESD) as measured with thermoluminescent dosimeters (TLDs) can be used as the first step in assessing radiation risk to patients. In mammography, where it is important to achieve high contrast, a low energy X-ray spectrum is used. As a consequence, the dose within the breast decreases rapidly with increasing depth. Thus, while the ESD is easy to measure it does not take account of this dose variation, so another dose quantity has been determined for specific use in mammography. The mean glandular dose (MGD) is the dose quantity recommended by many bodies including the ICRP [5]. Since it is difficult to measure, Monte Carlo model-based conversion factors are used which relate the ESD to MGD [6, 7].
Whilst phantoms are used for quality control of mammography units it is important to measure in vivo doses. In vivo doses will take into account differences not only in the X-ray unit but differences in breast size and composition. This point is highlighted by these examples of variations from the doses measured in a phantom; DeWard and Chiu measured in vivo doses (using LiF:Mg,Ti TLDs and correcting their readings for energy response as a result of calibrating their TLDs with a caesium-137 source) and converted to MGD using factors that correct for differences in breast composition via compressed thickness [8]. Measured ESD divided by a backscatter factor (thus entrance exposure) was plotted against breast thickness and compared with results using a standard phantom (50%–50% glandular–adipose tissue). Doses were higher for in vivo measurements than for a standard phantom by up to a factor of two.
In contrast, Young et al [9] describe two alternative methods to calculate MGD to the breast used in the UK Breast Screening Programme, neither of which takes into account differences in breast composition. These methods include measurement of dose to the standard breast using varying thicknesses of polymethylmethacrylate (PMMA) material and estimation of MGD for individual women using their X-ray exposure factors. Since the methods do not take into account variations in breast composition, the former method overestimates the dose to the lower glandular content of larger breasts [10] and the latter method underestimates the doses unless the low glandularity is taken into account in the conversion factor to MGD.
TLDs of normal thickness (0.8–0.9 mm) are opaque at mammographic energies, and are not recommended for in vivo dosimetry since the detector could obscure important diagnostic information. While in vivo measurements have been performed in the USA [8, 11, 12] the number of studies have been limited. In the European Protocol [13] it was recommended that TLDs are positioned at the upper inner quadrant of the woman's breast where the risk of obscuring clinical important details is less. This may lead to uncertainty due to the inhomogeneity of the radiation field caused by the inverse square law and heel effect [14]. Ideally, the assessment of dose for in vivo measurements should be standardized in terms of position (4 cm from chest wall at mid-line) as found in criteria for phantom measurements.
The aim of this investigation was to access the suitability of a paper-thin TLD for in vivo dosimetric measurements of mammography procedures.
|
Materials and method
|
|---|
Thermoluminescent dosimeters
LiF:Mg,Cu,P has been widely reported as being the ideal TLD material for low dose diagnostic procedures as it has high sensitivity (approximately 30 times greater than the standard LiF:Mg,Ti), a linear dose–response (up to 10 Gy) and an excellent energy response [15, 16]. It was thought that with the increased sensitivity of this thermoluminescent material a small quantity (thin layer) could be used to measure adequately the relatively high ESDs found in mammography. We obtained GR-200F (LiF:Mg,Cu,P film) from Solid Dosimetric Detector and Method Laboratory (SDDML) in Beijing, to test this hypothesis. The film is 50 µm thick and has a phosphor layer on one side and a clear tape on the other side; the tape is slightly static so it adheres to the patient surface. Obviously, if we wish to re-use the TLD, it has to be cleaned between uses, as encapsulating it in another holder defeats the purpose.
TLD visibility
As the main characteristics of the LiF:Mg,Cu,P TLDs have already been assessed at diagnostic energies [17], the main focus of this investigation was the assessment of the degree of transparency at mammography energies. To allow direct comparison, GR-200A TLDs (LiF:Mg,Cu,P circular chips, 0.8 mm thick, SDDML, China) were used together with GR-200F TLDs. Examinations were performed over various breast thicknesses (2 cm to 5 cm) and voltages (24 kVp to 28 kVp) for a number of different phantoms (Table 1
).
The steps of the investigation are given in Table 2, which shows the phantom type and thickness, energy, purpose and outcome of the step. The outcome from each step led to the next step until all specific requirements of the study were met. At step 4, the GR-200F TLDs were not visible to the investigators; thus a series of 7 phantom mammograms were assessed by a panel of experts (step 5). Seven mammograms were produced with the test plate placed on top of 3 cm of PMMA at energies of 24 kVp (2 films), 26 kVp (3 films) and 28 kVp (2 films). Three images were produced at 26 kVp as this was the standard potential used at the site.
View this table:
[in this window]
[in a new window]
|
Table 2. The seven steps of the study, from the use of solid water at step 1, through to the use of the Leeds TOR(MAM) test object at steps 4, 5 and 6 and then on to the in vivo assessment in step 7
|
|
For each mammographic image, a TLD was carefully positioned onto the test plate with the use of a template containing 120 possible TLD positions. The TLD changed position between images. Seven experts, comprising 1 radiologist, 3 mammographers and 3 radiographers were asked to view each image in turn with the use of another template (altered to take into account the magnification of the image) to locate the position of the TLD. In total there were 49 (7 images x 7 experts) independent assessments of TLD position.
The definitive test for radiotransparency was to perform digital subtraction. Digital images were taken with and without the TLDs placed on the right side of the Leeds TOR(MAM) phantom and digital image subtraction was performed to see if the TLD was visible on the subtracted image. This was carried out for both the GR-200F and GR-200A TLDs using digital acquisition with an Athena Mammotome biopsy unit (Fischer Imaging, Denver, CO).
TLD characteristics
When the problem of the visibility of the TLDs was resolved, TLD characteristics were assessed. To maximize dose determination accuracy, the sensitivity factor was determined for every TLD in a batch; for this all TLDs in the batch were irradiated using a standard set-up. Before measurements were performed with a new batch, it was repeatedly annealed, irradiated and read out until consistency was attained; this was defined as a reproducibility of ±5% (1 standard deviation (SD)). The energy response of the TLD was investigated using spectral radiation from superficial/orthovoltage therapy beams at energies 60 kVp, 120 kVp and 300 kVp and a 4 MV X-ray beam from a linear accelerator. This calibration was completed with a cross comparison with a calibrated Victoreen ionization chamber (Nuclear Associates, New York, NY; external 3 cm3 mammography chamber: ONMC Model 303, Sun Nuclear Corporation, Melbourne, FL) at mammographic energies of 24 kVp, 26 kVp and 28 kVp. The dose linearity was assessed by exposing the TLDs to multiple (1, 2, 5 and 10) exposures with the mammographic unit. The unit had previously undergone a full quality assurance assessment, where the output was found to be consistent to within ±2%.
In vivo measurement
Step 7 (Table 2
) involved actual in vivo dosimetry on a volunteer (who gave informed consent) during a routine screening examination. Four images were taken — a craniocaudal and mediolateral oblique projection on each breast. One TLD was placed on the breast for each view in order to measure ESD.
Entrance surface dose
While the emphasis of this paper is on the acceptability of GR-200F for the use in in vivo mammography dosimetry, GR-200F and GR-200A were compared for accuracy in the measurement of ESDs. ESD was not converted into MGD, as the basis of this calculation is obviously the ESD. Measurements were performed on the Leeds TOR(MAX) phantom (4.5 cm thick) at 26 kVp to estimate the TLD reliability, accuracy and reproducibility when measuring ESDs. In vivo dosimetry was then performed on both breasts and projections on the volunteer.
|
Results
|
|---|
Visibility assessment
The image produced in step 1 is shown in Figure 1. The GR-200A are clearly visible, while the GR-200F were only just visible. In steps 2 and 3, even though the TLD was still just visible, it was noted that the high contrast test details were visible through the TLD shadow. This finding is similar to that of Wochos [11] who stated that microcalcifications could be seen through their TLDs which were 0.4 mm and 0.9 mm thick and hence did not hinder diagnosis. This is thought to be partially attributable to the fact that in 1978, higher potentials were used in mammography.
View larger version (45K):
[in this window]
[in a new window]
|
Figure 1. Comparison of visibility of GR-200F (4 thermoluminescent dosimeters (TLDs)) and GR-200A (5 TLDs) on 5 cm of RMI 457 solid water imaged at a potential of 26 kVp.
|
|
Of the seven members of the panel that viewed the resulting images, three could not distinguish the GR-200F TLD enough to even guess a grid position. The remaining assessors could not locate the grid correctly for any of the TLD positions.
The results of the digital subtraction are shown in Figure 2. Even though GR-200F is not obvious to the naked eye on the image (Figure 2c), it was visible on the subtracted image (Figure 2d). However, the GR-200F chip is a significant improvement over the GR-200A chip which was clearly visible on the image (Figure 2a) and therefore obvious on the subtracted image (Figure 2b). The GR200A chip is clearly seen in Figure 2a, and as such may obscure important detail behind the TLD chip, while no information was obscured or lost from the image containing the GR-200F chip.
View larger version (190K):
[in this window]
[in a new window]
|
Figure 2. Demonstration of digital subtraction of GR-200A and GR-200F. The GR-200A chip is clearly visible on both (a) the pre-and (b) the post-subtraction images. This is in contrast with the GR-200F chip which is not visible on (c) the pre-subtraction image and only just visible on (d) the post-subtraction image.
|
|
The patient mammograms (Figure 3) were reported by the radiologist, who was not informed that a TLD was positioned on the patient during the examination, and who failed to detect its presence.
TLD characteristics
The majority (92%) of the TLD readings were found to be reproducible to within ±8% (1 SD), while 65% were reproducible to within ±5% (1 SD) when exposed to standard radiation doses. The energy response of the TLDs was found to be 0.9 (±0.1) compared with the reference beams of 60 kVp and 4 MV. The energy response between mammography energies to 60 kVp (35 keV, 1.6 mm Al half value layer (HVL)) and 4 MV was determined to enable a more accurate calibration reading by using a radiotherapy beam. Energy response was measured at both 60 kVp (closest to the mammography energy range) and 4 MV as potential sources of a traceable calibration. Since radiotherapy beams are subject to more regular quality assurance, output is very consistent, beams flatter and were accessible to the investigators compared with beams nearer the mammography energy ranges.
The response of LiF:Mg,Cu,P TLDs throughout the mammography energy range has been reported by Brenier and Lisbona [18] to not vary by more than 8% from 25 kV (0.33 mm Al HVL) to 35 kV (0.39 mm Al), keeping in mind that TLD reproducibility was ±4%. Between 25 kVp and 28 kVp, variations in energy response was 
3% which was within experimental uncertainty (4%).
The output of the mammography unit was measured at 26 kVp by the Victoreen mammography ionization chamber which had a traceable calibration factor at 25 kVp. The discrepancy between the extrapolation from higher doses and the low dose (a) directly measured with the calibrated ionization chamber for multiple mammograms, and (b) interpolated from a linear regression fit to all measured low dose values, were ±3.5% and ±2.5%, respectively.
Entrance surface dose
The average ESD across the mammography field at 26 kVp, measured using GR-200F on the TOR(MAX) phantom, was 8.8±0.4 mGy (1 SD) (n=34). The average ESD measured using GR-200A was 8.8±0.3 mGy (1 SD) (n=12). The results of the four in vivo doses gave a mean dose of 13±2 mGy, with the variation in dose being explained by the variation in mAs (mean 137 mAs, coefficient of variation 11%) for the different projections (craniocaudal and mediolateral oblique). Field homogeneity was measured to be 10% in the anode–cathode direction.
|
Discussion
|
|---|
Current limitations to in vivo dosimetry and GR-200F
In vivo dosimetry should be an important part of routine quality control in mammography breast screening. The major limitation to date has been finding a detector that does not interfere with the diagnostic integrity of the procedure; GR-200F appears to be a solution to this shortfall in medical dosimetry. Even though there are substantial benefits of performing phantom measurements in mammography [19], an in vivo dosimetry study would be required for optimization of technique and filtration due to the range of breast composition and size encountered clinically.
Assessment of changing technology
An increasing number of post-menopausal women are receiving hormone replacement therapy (HRT). Due to this group's denser, more fibrous breasts, the risks of an induced cancer start to outweigh the benefits of regular breast screening. Thus dose minimization and technique optimization is of critical importance to this group of women [20].
There have been a number of advances in technology over the last few years that have been targeted at the reduction of doses to the larger, denser breast. These have included the introduction of double track targets (Mo/Rh) in 1992 [21], and the introduction of auto-selection of the tube voltage. An assessment of the dose and contrast of these units have shown that doses can be reduced for the larger breast by 19%, but the doses to the smaller breast were seen to increase by 30–40% [22]. These studies were carried out with the use of phantoms and calculation, and there needs to be a full assessment of actual doses for an extensive range of different breast size and density.
Entrance surface dose and postal dosimetry service
In vivo doses take into account differences not only in the X-ray unit but differences seen in breast size and composition. The majority of papers published on ESD and MGD are not based on in vivo dose measurements but on a calculation of entrance surface dose using X-ray exposure factors. Reference levels for ESD for patients (compressed breast thickness of 50 mm) or the standard phantom (45 mm PMMA phantom) are 10 mGy [19]. The average ESDs as assessed with the Leeds Phantom in this study were less than this value at 26 kVp (8.8 mGy), while the volunteer ESD was greater than this value (13 mGy).
The use of GR-200F should be able to provide an answer to the contradictions in ESD between phantom and in vivo studies mentioned in the introduction by measuring the ESD to patients undergoing mammography. Another important use of these TLDs might be in the use of a postal dosimetric service. The measurement of ESD accompanied with information on breast thickness and mAs allows for an accurate estimation of mean glandular tissue fraction; Monte Carlo conversion factors will finally allow a more individual assessment of MGD.
|
Conclusion
|
|---|
In vivo dosimetry is superior to any other method of dose evaluation due to measuring (rather than the calculating) of the dose delivered to the breast surface. The use of LiF:Mg,Cu,P GR-200F TLDs allows the opportunity of in vivo dose measurements across Breast Screening Centres. TLDs can now be placed on the compressed breast at a consistent position (4 cm from chest wall on mid-line) to measure ESD on women without interfering with clinical assessment. Thus, these TLDs will allow a direct measurement of radiation dose for women undergoing mammographic examinations on different types of mammography units.
|
Acknowledgments
|
|---|
The authors would like to thank the University of Newcastle for funding this project; Solid Dosimetric Detector and Method Laboratory in Beijing for providing the GR-200F TLDs and Hunter BreastScreen for the use of their facilities and staff, especially Joanne Walker and Joan Wright.
|
Footnotes
|
|---|
Current address for Dr L Duggan, St George Cancer Care Centre, Kogarah, Sydney, NSW 2217 Australia. 
Received for publication September 12, 2003.
Revision received January 5, 2004.
Accepted for publication February 11, 2004.
|
References
|
|---|
- Hankey BF, Brinton LA, Kessler LG, Abrams J. Breast. In: Miller BA, Ries LAG, Hankey BF, Kosary CL, Harras A, Devesa SS, et al, editors, SEER Cancer Statistics Review: 1973–1990. DHHS Publ. No. (NIH) 93-2789. Bethesda, MD: National Cancer Institute, 1993;IV.1–IV.24.
- International Commission on Radiological Protection. 1990 Recommendations of the International Commission on Radiological Protection. ICRP Publication 60 1991;Ann. ICRP 21(1–3). Oxford: Pergamon Press, 1991.
- National Program for the Early Detection of Breast Cancer. Monitoring and Evaluation Reference Group. National Program for the Early Detection of Breast Cancer–Evaluation of phase one: 1 July 1991–30 June 1994. Commonwealth Department of Human Services and Health, Canberra, 1994.
- Law J, Faulkner K. Cancers detected and induced, and associated risk and benefit, in a breast screening programme. Br J Radiol 2001;74:1121–7.[Abstract/Free Full Text]
- International Commission on Radiological Protection. Statement from the 1987 Como meeting of the ICRP: Reference terms for estimates of radiation dose from X-ray mammography. ICRP; Annals of the ICRP 17(4): Pergamon, Oxford, 1987.
- Wu X, Barnes GT, Tucker DM. Spectral dependence of glandular tissue dose in screen-film mammography. Radiology 1991;179:143–8.[Abstract/Free Full Text]
- Dance D. Monte Carlo calculation of conversion factors for the estimation of mean glandular breast dose. Phys Med Biol 1990;35:1211–9.[CrossRef][Medline]
- DeWerd LA, Chiu NB. The determination of radiation dose by mail for diagnostic radiological examinations with thermoluminescent dosemeters. Radiat Prot Dosim 1993;47:509–12.[Abstract]
- Young KC, Ramsdale ML, Bignell F. Review of dosimetric methods for mammography in the UK breast screening programme. Radiat Prot Dosim 1998;80:183–6.[Abstract]
- Klein R, Aichinger H, Dierker J, Jansen JTM, Joite-Barfuß S, Säbel M, et al. Determination of average glandular dose with modern mammography units for two large groups of patients. Phys Med Biol 1997;42:651–71.[CrossRef][Medline]
- Wochos JF, Fullerton GD, DeWerd LA. Mailed thermoluminescent dosimeter determination of entrance exposure and half-value layer in mammography. AJR Am J Roentgenol 1978;131:617–9.[Abstract]
- Gentry JR, DeWerd LA. TLD measurements of in vivo mammographic exposures and the calculated mean glandular dose across the United States. Med Phys 1996;23:899–903.[CrossRef][Medline]
- Zoetelief J, Fitzgerald M, Leitz W, Säbel M. European protocol on dosimetry in mammography 1996. Luxembourg: European Commission, 1996.
- Dance DR, Skinner CL, Carlsson G. Breast dosimetry. Appl Radiat Isot 1999;50:185–203.[CrossRef][Medline]
- Horowitz YS. LiF:Mg,Ti versus LiF:Mg,Cu,P: the competition heats up. Radiat Prot Dosim 1993;47:135–41.[Abstract]
- Zha Z, Wang S, Shen W, Zhu J, Cai G. Preparation and characteristics of LiF:Mg,Cu,P thermoluminescent material. Radiat Prot Dosim 1993;47:111–8.[Abstract]
- Fill UA, Regulla DF. Advanced LiF Technology for the assessment of patient exposure in diagnostic radiology. Radiat Prot Dosim 1998;80:225–9.[Abstract]
- Brenier J-P, Lisbona A. Air kerma calibration in mammography of thermoluminescence and semiconductor dosemeters against an ionisation chamber. Radiat Prot Dosim 1998;80:239–41.[Abstract]
- Zoetelief J, Fitzgerald M, Leitz W, Säbel M. Dosimetric methods for and influence of exposure parameters on the establishment of reference doses in mammography. Radiat Prot Dosim 1998;80:175–80.[Abstract]
- Miller G. Minimising radiation dose in the mammographic examination of large dense breasts. The Radiographer 1997;44:11–5.
- McLean D. Assessment of mammographic dose and image quality. In: Warren-Forward HM, editor. Patient dose levels and risk in diagnostic radiology. Australia: NMRS, 2002:149–57.
- Young KC, Ramsdale ML, Rust A, Cooke J. Effect of automatic kV selection on dose and contrast for a mammographic X-ray system. Br J Radiol 1997;70:1036–42.[Abstract]
Top
Abstract
Introduction
