British Journal of Radiology (2005) 78, 328-334
© 2005 British Institute of Radiology
doi: 10.1259/bjr/22554286
In vivo absorbed dose measurements in mammography using a new real-time luminescence technique
M C Aznar, MSc1,
B Hemdal, BSc2,
J Medin, PhD2,
C J Marckmann, PhD1,
C E Andersen, PhD1,
L Bøtter-Jensen, DSc1,
I Andersson, MD, PhD3 and
S Mattsson, PhD2
1 Radiation Research Department, Risø National Laboratory, DK-4000 Roskilde, Denmark, 2 Department of Medical Radiation Physics, Lund University, Malmö University Hospital, SE 205 02 Malmö, Sweden and 3 Department of Diagnostic Radiology, Lund University, Malmö University Hospital, SE 205 02 Malmö, Sweden
 |
Abstract
|
|---|
A dosimetry system based on radioluminescence (RL) and optically stimulated luminescence (OSL) from carbon doped aluminium oxide (Al2O3:C) crystals was developed for in vivo absorbed dose measurements in mammography. A small cylindrical crystal of Al2O3:C (diameter 0.48 mm and length 2 mm) was coupled to the end of a 1 mm diameter optical fibre cable. Owing to their small size and characteristic shape, these probes can be placed on the body surface in the field of view during the examination, without compromising the reading of the mammogram. Our new technique was tested with a mammography unit (Siemens Mammomat 3000) and screen–film technique over a range of clinically relevant X-ray energies. The results were compared with those obtained from an ionization chamber usually used for the determination of absorbed dose in mammography. The reproducibility of measurements was around 3% (1 standard deviation) at 4.5 mGy for both RL and OSL data. The dose response was found to be linear between 4.5 mGy and 30 mGy. The energy dependence of the system is around 18% between 23 kV and 35 kV. In vivo measurements were performed during three patient examinations. It was shown that entrance and exit doses could be measured. The presence of the small probes did not significantly interfere with the diagnostic quality of the images. Entrance doses estimated by RL/OSL results agreed within 3% with entrance surface dose values calculated from the ionization chamber measurements. These results indicate a considerable potential for use in routine control and in vivo dose measurements in mammography.
|
Introduction
|
|---|
Mammography is the most effective method to screen asymptomatic women in selected age groups in order to detect early stages of breast cancer. It is however estimated that when saving 20 women (age 40–50 years) through regular mammography screening, one lethal cancer may be induced by the radiation [1]. It is therefore necessary to monitor the radiation dose in mammography. The risk from radiation is currently described by the average absorbed dose to the glandular tissue. This quantity is called the average glandular dose (AGD) and is normally derived from measurement of the entrance surface air kerma (ESAK), the half-value layer (HVL) and the compressed breast thickness. Published conversion factors [2, 3] are used to calculate the AGD for different breast thicknesses from the ESAK value with the assumption that the breast composition is 50% fat and 50% glandular tissue. In recent years, these conversion factors have been extended [4, 5] to a larger interval of breast thicknesses and also to account for other breast compositions and X-ray spectra.
Accordingly, there are special demands on quality control in mammography, both internationally [3, 6, 7] and nationally [8]. In Sweden, there is a legal need [9] to make direct patient measurements in mammography screening for at least 20 patients each year using the European protocol on dosimetry in mammography [3].
Usually an ionization chamber is used to measure ESAK and HVL without any patient present. The tube output, i.e. the air kerma (mGy) relative to the tube loading (mAs) is measured behind the compression plate and should be specified at a reference point 45 mm above the breast support (cassette table) and laterally centred 60 mm from the chest wall side of the table edge. For each patient exposure, the mAs-value and compressed breast thickness are recorded (as well as the anode/filter combination and tube potential, if those parameters could be changed). Then the ESAK value can be calculated for that particular exposure. As an alternative, thermoluminescence dosemeters (TLDs) on the breast can be used as a simple method of determining ESAK [3]. In this case the measured entrance surface dose (ESD), defined in terms of absorbed dose in air at the skin surface including the contribution from backscatter, has to be divided by an appropriate backscatter factor (BSF), which is tabulated as a function of HVL (BSF is 1.07 for HVL=0.25 mm Al and 1.13 for 0.65 mm Al).
The aim of this work was to develop a small size, high-sensitivity optical fibre dosemeter capable of measuring real-time entrance as well as exit absorbed doses and dose rates in mammography, and to test the dosemeter with respect to reproducibility, linearity of the signal with absorbed dose, energy dependence, backscatter and breast thickness simulated with a phantom. Finally we wanted to use the dosemeter for in vivo entrance and exit measurements in real mammography examinations in order to see whether the disturbances from the dosemeters on the image are acceptable in routine mammography screening and whether this technique has a potential for use in a quality control programme for mammography.
|
Materials and methods
|
|---|
The RL/OSL dosemeter system
Most crystalline materials produce luminescence when exposed to ionizing radiation owing to the presence of impurities or defects in the lattice. One has to distinguish between two different signals: radioluminescence (RL), which is emitted promptly during irradiation; and luminescence that is emitted when the crystal is stimulated either by heat, referred to as thermoluminescence (TL) or, as in our case, by laser light referred to as optically stimulated luminescence (OSL). The instantaneous RL signal is proportional to the dose rate while the OSL is proportional to the absorbed dose in the crystalline detector. Figure 1
shows an example of RL and OSL signals obtained for a single exposure from the mammography unit corresponding to an air kerma of approximately 10 mGy, which can be considered equal to the absorbed dose in air for the photon beam qualities included in the study.
View larger version (12K):
[in this window]
[in a new window]
|
Figure 1. Schematic representation of radioluminescence (RL) and optically stimulated luminescence (OSL) signals from a typical exposure (the absorbed dose is approximately 10 mGy). The exposure starts at time zero and the RL is collected during the exposure. The OSL is shown collected during a time which starts approximately 20 s after the end of the exposures.
|
|
The RL/OSL dosimetry system presented in this study uses carbon-doped aluminium oxide (Al2O3:C) as a sensor crystal. Al2O3:C was chosen among various potential luminescent materials in view of its excellent OSL properties [10, 11]. The effective atomic number of Al2O3:C is 10.2 and its density is 3.97 g.cm–3. Al2O3:C emits both RL and OSL in the blue region (
420 nm). The Al2O3:C detector (2 mm long, 0.48 mm in diameter; provided by Landauer Inc., Stillwater Crystal Growth Division, USA) is connected to the optical system via a 10 m long optical fibre cable made of a 0.48 mm polymethylmethacrylate (PMMA) core (Super Eska GK-20, Mitsubishi Rayon Co., Ltd., Japan). The total diameter of the fibre cable, including cladding and jacket, is 1 mm. To produce OSL, a green laser beam (532 nm, 20 mW) is focused through a dichroic colour beamsplitter and transported via the optical fibre into the Al2O3:C crystal. The stimulated OSL signal is carried back from the crystal in the same fibre and reflected by the beamsplitter into a photomuliplier (PM) tube. A set of filters in front of the PM tube minimizes the scattered green light from the laser. The dosemeter system used in this study consists of two optical units, which are controlled using a standard laptop computer, equipped with a data acquisition card and a Labview interface (both products of National Instruments, Austin, USA). This configuration enables the system to be compact and uses only one optical fibre per dosemeter. A schematic drawing of the RL/OSL dosimetry system is provided in Figure 2.
View larger version (25K):
[in this window]
[in a new window]
|
Figure 2. Schematic diagram of the radioluminescence/optically stimulated luminescence (RL/OSL) dosimetry system.
|
|
Two probes were used, and are referred to as "fibre #48" and "fibre #49". The sensitivity of the probes is dependent on the manufacturing process of Al2O3:C, the size of the crystal and the optical coupling. As a consequence, fibre #48 is more sensitive than fibre #49, and was therefore used for exit dose measurements. Apart from the difference in sensitivity, the two probes exhibit the same behaviour. Hence, unless otherwise specified, only results from fibre #49 are presented in the following sections.
Protocol and data analysis
In all experiments performed in the present work except for the in vivo measurements (see below), the RL and OSL signals were acquired according to the schematic arrangement shown in Figure 1
, i.e. the RL is collected during the exposure, and the laser light is switched on shortly after the end of the exposure to collect the OSL signal. Because the RL signal is collected during irradiation, it is potentially affected by the signal arising from fluorescence generated in the optical fibre (referred to as "stem effect"). However, for photon beam qualities used in mammography the signal generated in the light guide was found to be insignificant compared with the signal coming from the crystal (<1%). The stem effect is not present when the OSL signal is acquired, since it is not read out during the irradiation. Independent absorbed dose estimates were obtained by integrating either the complete RL signal, or the OSL signal for a fixed duration. For both signals (RL and OSL), a background correction is performed, and the raw signal from each probe is corrected to account for sensitivity differences. It was sufficient to integrate the OSL signal for about 100 s to provide a reliable estimate of the relatively small absorbed doses in mammography. However, in the following results, the RL signal generates a higher number of counts per reading for a given absorbed dose than the OSL signal and is consequently considered more accurate.
As mentioned in a previous work [12], the RL signal is subject to sensitivity changes with accumulated absorbed dose. As a consequence, for an irradiation at a constant dose rate, the RL signal will increase. In the case of mammography however, the resulting absorbed doses are too small for any significant changes in RL sensitivity to occur (about 0.2% for an absorbed dose of 10 mGy).
Description of the mammography system and the conventional test equipment
All experiments were performed with a mammography unit (Siemens Mammomat 3000; Siemens, Erlangen, Germany) using a Mo/Mo anode/filter combination. The tube potential ranged from 23 kV to 35 kV. A screen–film system (Kodak Min-R 2190/Min-R 2000, 36°C, 90 s; Eastman Kodak, Rochester, NY) was used. Whenever relevant, the signals from the RL/OSL dosemeter were compared with the reading of a flat ionization chamber (type TB23344 from PTW Freiburg, Germany) connected to an electrometer (Solidos 300, RTI, Mölndal, Sweden).
A PMMA phantom (area 240 mm x 150 mm) was combined to various thicknesses and was used to simulate the presence of breast tissue. The HVL measurements were performed using aluminium foils with a purity of at least 99.99% and varying thicknesses of 0.208 mm, 0.094 mm, 0.091 mm, 0.046 mm and 0.026 mm (Goodfellow, Cambridge, UK) as well as a lead diaphragm (1 mm thickness, 16 mm x 15 mm).
In vivo measurements
Before any in vivo measurements were performed, images of the probes positioned on a PMMA phantom, a common quality control phantom for mammography (RMI 156, Gammex RMI, Middleton, WI) and a phantom with human breast tissue in formalin surrounded by polyethylene [13] were acquired. Three experienced radiologists evaluated those images, and gave their approval for the use of the probes during patient examination. In vivo measurements were scheduled on three women coming to an ordinary screening examination (selected to have different breast sizes and densities).
Fibre #49 was attached to the compression paddle in such a way that it was compressed between the compression paddle and the upper part of the breast during the examination. Fibre #48 was positioned on the table of the mammography unit, and was then in contact with the inferior surface of the breast (see Figure 2). The fibres were positioned with each crystal laterally centred 60 mm from the chest wall side of the breast support so that they would not be superimposed on the mammograms. Craniocaudal (CC) and mediolateral oblique (MLO) projections were used. Lead markers were attached to the compression paddle on its upper side in order to obtain a more precise estimate of the compressed breast thickness [14] 60 mm from the chest wall side of the breast support for each patient image. Except for these preparations, the standard procedure for mammography screening was followed, including the use of a tube potential of 29 kV.
Each patient received four exposures (two for each breast). RL data were acquired for each individual exposure. However, OSL readings were only performed after each patient's examination was completed. Hence, for each patient, four RL measurements and one OSL measurement (representing the sum of four exposures) were acquired. This protocol was chosen in order to avoid delays due to the longer OSL reading procedure.
Finally, the radiologists evaluated the images from all three examinations to assess the interference of the probes with the diagnostic quality of the image. One expert radiologist also evaluated the screening images in order to quantify the glandularity of the breast parenchyma according to a modified Wolfe classification [15].
HVL, ESAK and AGD were evaluated according to the European protocol on dosimetry in mammography [3]. ESD was evaluated from ESAK using a backscatter factor of 1.084 corresponding to a measured HVL of 0.369 mm Al for the anode/filter combination Mo/Mo and the tube potential of 29 kV. The backscatter factor was interpolated from tabulated values of 1.08 and 1.09 at HVL 0.35 and 0.40 mm Al, respectively [3]. The same values were tabulated for breast tissue and PMMA as the scattering material.
|
Results and discussion
|
|---|
Reproducibility and linearity
The RL/OSL fibres and the ionization chamber were positioned underneath the compression plate of the mammography unit. No extra build-up or backscatter material was used. 13 exposures were made at 29 kV and 40 mAs corresponding to an air kerma of 4.5 mGy. RL and OSL data were evaluated for both fibres, and normalized to ionization chamber readings. The results are presented in Table 1 and indicate that the overall reproducibility of the system is approximately 3% at 4.5 mGy. The uncertainty in the RL data is dominated by counting statistics, while additional factors (such as laser instability) seem to increase the uncertainty in the OSL results. The dose–response of the RL/OSL system was also investigated at 29 kV, and Figure 3 shows that the output is linear with dose between 40 mAs and 250 mAs (4.5–30 mGy). The minimum detectable absorbed dose, calculated as 3 times the standard deviation of the background, is below 50 µGy for both fibres for the RL signal, and approximately about 200 µGy for the OSL signal.
View this table:
[in this window]
[in a new window]
|
Table 1. Reproducibility of the radioluminescence/optically stimulated luminescence (RL/OSL) system (expressed as the relative standard deviation of 13 measurements at 29 kV for an air kerma of 4.5 mGy)
|
|
View larger version (10K):
[in this window]
[in a new window]
|
Figure 3. Linearity of dose–response at 29 kV (R-squared >0.999). Data are normalized to their value at a 250 mAs exposure, corresponding to an air kerma of 30 mGy (radioluminescence (RL): crosses, optically stimulated luminescence (OSL): circles).
|
|
Energy dependence
The variation in response of Al2O3:C as a function of tube potential was investigated using the same experimental set-up as described above, keeping the tube loading constant at 40 mAs during all measurements. The RL and OSL data were normalized to the simultaneous reading of the ionization chamber. The energy dependence of this ionization chamber is of the order of 1% for the range of energies used in the study (PTW Calibration Certificate, calibration traceable to national standards of the German National Laboratory, PTB, Braunschweig).
Figure 4 shows the obtained results as a function of HVL. The response of the Al2O3:C crystal was found to increase by 18% when the tube potential is increased from 23 kV to 35 kV, corresponding to an HVL range of 0.3–0.41 mm Al.
View larger version (11K):
[in this window]
[in a new window]
|
Figure 4. Energy dependence of the luminescence signals compared with the ionization chamber (radioluminescence (RL): crosses, optically stimulated luminescence (OSL): circles). RL and OSL readings are divided by the simultaneous ionization chamber readings. Data are normalized to read unity at 29 kV.
|
|
For the range of energies studied, the Al2O3:C crystal can be considered as a "photon detector", i.e. the energy deposition in the detector volume can be directly related to the incident photons and not to the secondary electrons emitted in the material surrounding the detector. This is a reasonable approximation considering the size of the detector and the short range of the emitted electrons, which is less than 8 µm even for the highest maximum energy [16].
Backscatter
Backscatter factors for the RL/OSL probes were defined as:
where Sphantom is the signal at a point on the surface of the PMMA phantom and Sair is the signal free in air at the same point without the phantom.
The phantoms used ranged from 1 cm to 6 cm in thickness. The tube potential was kept at 29 kV. Three exposures were performed in each setting, leading to one RL measurement per exposure, and one OSL measurement after each set of three exposures. The results are presented in Table 2.
View this table:
[in this window]
[in a new window]
|
Table 2. Measurements of backscatter from a polymethylmethacrylate (PMMA) phantom. Three exposures were performed in each set up
|
|
The data in Table 2 suggest that full backscatter is achieved with 1 cm of PMMA. The average backscatter factor is 1.066±0.004 (n=12) for the pooled RL and OSL data (ignoring differences in uncertainties and assuming that the measurements are independent) where the indicated uncertainty (expressed as one standard deviation) only includes the observed random variation from measurement to measurement. The observed backscatter factor is slightly lower than the value of 1.084 for TLDs with PMMA or breast tissue as the scattering material [3]. This discrepancy may reflect differences in geometry or in energy response of the different materials [17].
Influence of phantom thickness (exit/entrance measurements)
The sensitivity of the RL/OSL probes is sufficient to measure exit doses in typical exposures. In the following experiment, the thickness of the PMMA phantom was varied between 2 cm and 6 cm, while the tube potential ranged from 23 kV to 35 kV. Fibre #49 was positioned at the entrance surface of the phantom, while fibre #48 was placed on the exit surface. The ratio "exit RL signal/entrance RL signal" was calculated in each case and corrected for the difference in sensitivity. As illustrated in Figure 5, the exit/entrance RL signal ratio decreases with increasing phantom thickness, and with decreasing tube potential. Uncertainty bars were determined by counting statistics. Only in the most extreme case (6 cm PMMA and 23 kV) was the sensitivity of the exit probe too low to provide a reliable RL signal. The OSL data provide similar results, but with considerably higher uncertainties for the low dose points.
View larger version (11K):
[in this window]
[in a new window]
|
Figure 5. Measurements with one probe (fibre #49) at the entrance surface, the other (fibre #48) at the exit surface of a polymethylmethacrylate (PMMA) phantom. (radioluminescence (RL) exit)/(RL entrance) signals are presented for 23 kV (diamonds), 29 kV (triangles) and 35 kV (circles).
|
|
In vivo measurements
Except for the first patient, the fibres were positioned in such a way that their shadows did not superimpose on the mammograms. The compression paddle was not fixed in the lateral direction, which accidentally led to superposition of the fibres in one image of the first screening examination in spite of apparent separation before the breast compression. After further separation of the fibres to about 20 mm, they were not superimposed for the next two patients. Three radiologists viewed the images and stated that neither the fibre dosemeters nor the lead markers disturbed the reading of the mammograms due to their small sizes and characteristic shapes that could not be confused with structures in the breast, either benign or malignant (see Figure 6). Single calcifications could have been obscured, but this was not considered to impair significantly early breast cancer detection.
View larger version (79K):
[in this window]
[in a new window]
|
Figure 6. Impact of the radioluminescence/optically stimulated luminescence (RL/OSL) fibres on a mediolateral oblique (MLO) mammogram. Lead markers 60 mm from the chest wall side of the breast support are indicated by arrows.
|
|
The estimated compressed breast thickness, ESAK and AGD are presented in Table 3 together with estimated glandularity and breast parenchymal category, which was found to be the same for each breast of the same woman. Table 4 summarizes the results from RL/OSL measurements, compared with the calculated ESD from the ionization chamber measurements. RL and OSL results are expressed as counts, and are corrected for the difference in sensitivity of the two probes. For measurements at a fixed tube potential (here, 29 kV), the sum of the entrance RL signal seems to be a reliable estimator of the entrance surface dose, as the ratio "RL entrance/ESD" has a standard deviation of 3% (i.e. within the uncertainty associated with the system). The exit doses could also be measured in vivo during all 12 exposures, as seen from the "RL exit" signal (this signal was at least 4 times higher than the background signal in all cases). The ratio of exit to entrance doses can be obtained for each individual image via RL analysis, or for each patient via OSL analysis. These estimates agree within ±10%.
View this table:
[in this window]
[in a new window]
|
Table 3. Patient image data and entrance surface air kerma (ESAK) and average glandular dose (AGD) values determined from the ionization chamber measurements. CC, cranio-caudal; MLO, mediolateral oblique
|
|
View this table:
[in this window]
[in a new window]
|
Table 4. Entrance surface dose (ESD) determined for each breast image from the ionization chamber measurements and radioluminescence/optically stimulated luminescence (RL/OSL) in vivo data. RL and OSL counts have been corrected for sensitivity differences between fibre #48 and fibre #49
|
|
|
Conclusions
|
|---|
A new small size, high-sensitivity optical fibre dosemeter capable of measuring real-time entrance as well as exit absorbed doses and dose rates in mammography, was developed. The system has been tested in clinical mammography and it was found that the reproducibility is of the order of 3% (1 standard deviation) at an air kerma of 4.5 mGy. The response is linear with absorbed dose between at least 4.5 mGy and 30 mGy. The energy dependence for the range of beam qualities tested (23–35 kV) is around 18%. The BSF is approximately 1.07.
In vivo entrance and exit dose measurements of three women undergoing a standard mammography screening examination showed that the technique can be used for estimation of ESD and that the disturbances from the dosemeters on the image are acceptable. It is concluded that this technique has considerable potential for use in a quality control programme for mammography.
|
Acknowledgments
|
|---|
The authors are much obliged to all the staff at the Mammography section, Department of Diagnostic Radiology, Malmö University Hospital, Sweden for providing the opportunity to perform this study and especially for their help when the in vivo measurements were performed. We would also like to thank Jan B Olsen in Oslo, Norway, for providing the breast tissue phantom. The group at Risø National Laboratory gratefully acknowledges financial support from Landauer Inc., Chicago, IL, USA.
Received for publication April 19, 2004.
Revision received November 12, 2004.
Accepted for publication December 20, 2004.
|
References
|
|---|
- Andersson I, Janzon L. Reduced breast cancer mortality in women under age 50: updated results from the Malmo Mammographic Screening Program. J Natl Cancer Inst Monogr 1997;(22):63–7.
- Dance DR. Monte Carlo calculation of conversion factors for the estimation of mean glandular breast dose. Phys Med Biol 1990;35:1211–9.[CrossRef][Medline]
- Zoetelief J, Fitzgerald M, Leitz W, Säbel M. European protocol on dosimetry in mammography. EUR 16263, 1996.
- Dance DR, Skinner CL, Young KC, Beckett JR, Kotre CJ. Additional factors for the estimation of mean glandular breast dose using the UK mammography dosimetry protocol. Phys Med Biol 2000;45:3225–40.[CrossRef][Medline]
- van Engen R, Young K, Bosmans H, Thijssen M. Addendum on digital mammography, European Guidelines for Quality Assurance in Mammography Screening 2003 EUREF.
- EURATOM. The European medical exposure directive 97/43. Council Directive 97/43/Euratom of 30 June 1997 on health protection of individuals against the dangers of ionizing radiation in relation to medical exposure, 1997.
- Perry N, Broeders M, de Wolf C, Törnberg S, editors. European Guidelines for Quality Assurance in Mammography Screening (3rd edn). 2001 European Commission, EUREF.
- Socialstyrelsen. National Board of Health and Welfare. Screening with Mammography for early Detection of Breast Cancer 1998. SoS-report 1998:17 (Stockholm, in Swedish).
- SSI, Swedish Radiation Protection Authority, Regulations and general advice on diagnostic standard doses and reference levels within medical X-ray diagnostics, SSI FS 2002:2 (Stockholm, in Swedish, non-authorised translation to English is available at www.ssi.se), 2002.
- Bøtter-Jensen L, McKeever SWS. Optically stimulated luminescence dosimetry using natural and synthetic materials. Radiat Prot Dosim 1996;65:273–80.[Abstract]
- Bøtter-Jensen L, Agersnap Larsen N, Markey BG, McKeever SWS. Al2O3:C as a sensitive OSL dosemeter for rapid assessment of environmental photon dose rates. Radiat Meas 1997;27:295–8.[CrossRef]
- Aznar MC, Andersen CE, Bøtter-Jensen L, Bäck SÅJ, Mattsson S, Kjær-Kristoffersen F, et al. Real-time optical-fibre luminescence dosimetry for radiotherapy: physical characteristics and applications in photon beams. Phys Med Biol 2004;49:1655–69.[CrossRef][Medline]
- Olsen JB, Sager EM. Subjective evaluation of image quality based on images obtained with a breast tissue phantom: comparison with a conventional image quality phantom. Br J Radiol 1995;68:160–4.[Abstract]
- Burch A, Law J. A method for estimating compressed breast thickness during mammography. Br J Radiol 1995;68:394–9.[Abstract]
- Wolfe JN. Breast parenchymal patterns and their changes with age. Radiology 1976;121:545–52.[Abstract]
- ICRU Stopping powers for electrons and positrons, ICRU Report 37, 1984.
- Chan HP, Doi K. Monte Carlo simulation studies of backscatter factors in mammography. Radiology 1981;139:195–9.[Abstract/Free Full Text]
Top
Abstract
Introduction
