Effects of Quality Assurance Regulatory Enforcement on Performance of Mammography Systems: Evidence From Large- Scale Surveys in Taiwan

Transcription

1 Quality Assurance in Mammography Medical Physics and Informatics Original Research Medical Physics and Informatics Original Research Yi-Shuan Hwang 1 Hui-Yu Tsai 2 Chien-Chuan Chen 1 Pei-Kwei Tsay 3 Huay-Ben Pan 4,5 Giu-Cheng Hsu 6 Jen-Hsuan Lin 7 Chen-Shou Chui 8 Yung-Liang Wan 1,2 Ho-Ling Liu 1,2 Hwang YS, Tsai HY, Chen CC, et al. Keywords: average glandular dose, image quality, mammography, quality assurance, standards DOI: /AJR Received July 17, 2012; accepted without revision September 18, This work was supported by grants from the National Science Council and the Atomic Energy Council of Taiwan (NSC NU, E NU, E NU, and AEC10103L) and grants from the Bureau of Health Promotion of Taiwan (SC961035, C, and 9927C). 1 Department of Medical Imaging and Intervention, Chang Gung Memorial Hospital at Linkou, 5 Fushing St, Kweishan, Taoyuan 333, Taiwan. Address correspondence to H. L. Liu (hlaliu@mail.cgu.edu.tw) or Y. L. Wan (ylw0518@adm.cgmh.org.tw). 2 Department of Medical Imaging and Radiological Sciences, College of Medicine, and Healthy Aging Research Center, Chang Gung University, Taoyuan, Taiwan. 3 Department of Public Health and Center of Biostatistics, Chang Gung University, Taoyuan, Taiwan. 4 Department of Radiology, Kaohsiung Veterans General Hospital, Kaohsiung, Taiwan. 5 National Yang-Ming University, Taipei, Taiwan. 6 Department of Radiology, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan. 7 Department of Radiation Protection, Atomic Energy Council, Taipei, Taiwan. 8 Department of Medical Physics, Koo Foundation Sun Yat-Sen Cancer Center, Taipei, Taiwan. WEB This is a web exclusive article. AJR 2013; 201:W307 W X/13/2012 W307 American Roentgen Ray Society Effects of Quality Assurance Regulatory Enforcement on Performance of Mammography Systems: Evidence From Large- Scale Surveys in Taiwan OBJECTIVE. The Standards for Medical Exposure Quality Assurance in mammography systems were enacted on July 1,, in Taiwan. This study aimed to evaluate the trends in performance of mammography units before and after the regulation started on the basis of annual on-site surveys from to MATERIALS AND METHODS. On-site measurements were conducted on 215, 205, and 209 mammography units in, 2009, and 2010, respectively, which accounted for more than % of all units in Taiwan. Phantom image quality, average glandular dose (AGD), and half-value layer were evaluated on all systems. Processor conditions, compression conditions, radiation output, and computed radiography exposure indicators were assessed on units participating in mammography screening in and on all units in the later years. Evaluations of maximum compression force and automatic exposure control reproducibility were added into the protocol from 2009 onward. RESULTS. Mean phantom scores were improved significantly from to 2009 (11.63 ± 1.30 vs ± 0.94, p < 5) and remained stable for 2010 (12.35 ± 0.87). Mean AGDs were 1.48 ± 0.47, 1.38 ± 0.41, and 1.37 ± 0.42 mgy over the 3 years, with a significant reduction from to 2009 (p < 5). For film-screen mammography systems, variations of sensitometric curves were greatly reduced in 2009 and 2010 when compared with. Passing rates were increased after the regulation took effect in almost all aspects. CONCLUSION. Results from large-scale on-site surveys showed an overall improvement in performance after quality assurance in mammography was enforced in Taiwan. M ammography is regarded as the most important tool in early detection of breast cancer. For accurate diagnosis, high-quality mammographic images are needed to detect small lesions or any abnormal structures. However, radiation dose in mammography also has been a critical concern. To produce high-quality mammographic images and keep patient dose reasonably low, a quality control (QC) program in mammography, especially in successful breast cancer screening programs, must be implemented. Because of the increased attention being paid, the Mammography Quality Standards Act (MQSA) was enacted in the United States in 1992 and enforced in 1994 [1]. The mammography QC manual published by the American College of Radiology (ACR) in 1999 [2] and the vendor/model-specific QC manuals approved by U.S. Food and Drug Administration (FDA) for digital mammography provide practical references for complying with the regulation in the United States. For breast cancer screening, the European Commission established European guidelines for quality assurance (QA) in screening mammography [3]. Similarly, extensive QC programs have been implemented in mammography screening centers across other countries [4]. In Taiwan, the Bureau of Health Promotion, Ministry of Health, started the mammography screening project in 2002 [5]. Although general image quality and radiation doses were monitored on the screening units, no comprehensive QA program was required. Since July 1,, mammography systems have been included in the scope of Standards for Medical Exposure Quality Assurance by the Atomic Energy Council of Taiwan. All clinical facilities are now required by law to perform mammography QC procedures, which are similar to those required by MQSA with some modifications for local practice. To evaluate the effectiveness of QC practices and the performance of mammography systems, national or regional surveys were previously conducted in several countries [4, 6 14]. AJR:201, August 2013 W307

2 Data from the Nationwide Evaluation of X-Ray Trends (NEXT) survey and the MQSA inspection have shown continuous improvements in performance of mammography systems in the United States since 1985 [6, 7]. In the United Kingdom, all mammographic systems used in the National Health Service Breast Screening Programme (NHSBSP) are subject to regular testing following the procedures defined by the Institute of Physics and Engineering in Medicine (IPEM) [15]. When compared with previous survey data, results collected from 1998 to 1999 showed improvements in image quality without a significant increase in radiation dose [9]. Generally, medical equipment in Taiwan is considered quite modern, but the concept of QA in diagnostic imaging started relatively late compared with developed Western countries. Most previously published studies about the impact of QA on mammography system performance are from 10 years ago and are focused on film-screen mammography systems. To initiate mammography QA regulation in Taiwan and evaluate its impact, large-scale surveys with on-site measurements were conducted annually from to This study aimed to evaluate the effects of QA enforcement on the basis of the survey results, which may constitute a unique reference, especially for countries or regions that have not started similar regulations. Materials and Methods A total of 215, 205, and 209 mammography units were evaluated in, 2009, and 2010, respectively, which accounted for more than % of the total units in Taiwan. The numbers of filmscreen and digital mammography systems (with the distribution of manufacturers and models) are listed in Table 1. On-site visits and measurements were conducted by physics assistants who were trained by certificated medical imaging physicists. Phantom image quality, average glandular dose (AGD), and half-value layer (HVL) were evaluated on all systems over the 3 years. Film processor conditions, compression conditions, automatic exposure control reproducibility, radiation output, and computed radiography exposure indicator confirmations were assessed on the screening units in and on all units in 2009 and Phantom Image Quality An ACR accreditation phantom (Victoreen model N , Fluke Biomedical) was exposed with the clinical technique for a 4.2-cm compressed breast of average density at each surveyed site [4]. Each phantom image was scored by two of four qualified medical physicists according to the ACR scoring protocols, and the average scores were taken under optimal viewing condition for each system. For film-screen mammography systems, background density and density difference were measured on the phantom imaging test films. Passing criteria were given for the visibility of four fibers, three specks, three masses, and a background density no less than 1.20 for film-screen mammography units. For digital systems, passing criteria were based on manufacturer recommendations (quality control manuals for Senographe 2000 D and Senographe Essential, GE Healthcare; Lorad Selenia, Hologic; Giotto Image 3D, IMS; Navation DR and Inspiration, Siemens Healthcare; FCRm, Fujifilm Medical Systems; and Regius Contact, Konica Minolta Medical Imaging). Phantom total scores were calculated as the summation of the scores of fibers, specks, and masses. Automatic Exposure Control Reproducibility and Average Glandular Dose Breast entrance exposures were measured using a 6-cm 3 ion chamber and an electrometer (models M-3 and 10, Radcal Corporation) calibrated at mammographic x-ray beam energies, with the setup according to the ACR QC manual [4]. Four exposures were made with the clinical technique for a 4.2-cm average breast at each site. The passing criterion for the automatic exposure control reproducibility was a coefficient of variation (CV) of the four measurements no greater than 5. The mean entrance exposure was used to estimate the AGD by applying the exposure to the AGD conversion factor from Wu et al. [16, 17] or Dance et al. [18, 19], depending on the mammographic system. The upper limit of the AGD was 3 mgy. Fig. 1 Chart shows mean phantom scores and average glandular dose (AGD) values obtained from to 2010 in Taiwan. Asterisk denotes significant differences (p < 5) compared with previous year. Error bars indicate standard error of mean. Phantom Score Phantom score AGD Film Processor Conditions Film processing tests were performed at 115, 73, and 51 sites using film-screen mammography in, 2009, and 2010, respectively. A film of the surveyed site was exposed by a calibrated sensitometer (Victoreen model , Fluke Biomedical) and developed in the local processor. A sensitometric curve was produced for each processor by measuring the optical densities using a calibrated densitometer (Victoreen model , Fluke Biomedical). Base plus fog, mid density, and density difference were obtained according to the ACR QC manual [4]. Other Measurements HVL, radiation output, and compression conditions were evaluated as described in the ACR QC manual [4], with the same ion chamber and electrometer used for the breast entrance exposure measurements. In addition, exposure indicator accuracy was investigated on computed radiography systems according to protocols recommended by the manufacturers. Type 1145 aluminum alloy plates (model 8220, Radcal Corporation) were used for HVL measurements and the passing criterion was no less than kvp / + 3 mm aluminum, with the compression paddle in place. For radiation output, the passing criteria were 7.0 mgy/s for film-screen mammography and according to manufacturers recommendations for digital systems. For computed radiography exposure indicator confirmation, the passing criterion was within 20% of the target values. For compression conditions, measurements included the compression thickness accuracy and reproducibility, maximum compression force, and alignment of the chest wall edge of the compression paddle. The ACR phantom was adopted for the compression thickness check, with passing criteria of ± cm for accuracy and ± 0.2 cm for reproducibility. Maximum compression force was measured with a mammography compression test device (model 163, RMI), and the result had to lie between 11.4 and 20.4 kg (25 and 45 pounds) in the initial power drive mode. Passing criteria for the compression paddle alignment required the chest wall edge of the paddle to be invisible in the image and not extend beyond the image receptor by more than 1% of the source-to-image distance. Statistical Analysis Quantitative data were analyzed using analysis of variance to test different systems and years. If the test was significant, we used the Bonferroni procedure Year AGD (mgy) W308 AJR:201, August 2013

3 Quality Assurance in Mammography for multiple comparisons. All statistical analyses were performed using SPSS, version 17.0, and a p value < 5 was considered statistically significant. Optical Density A Optical Density C Fig. 2 Scatterplot shows mean phantom score versus average glandular dose (AGD) for film-screen mammography ( ), computed radiography ( ), digital radiography ( ), and all systems combined ( ) obtained from to Phantom Score Results Phantom Image Quality and Average Glandular Dose From the on-site measurements in, 2009, and 2010, the mean phantom total scores were ± 1.30, ± 0.94, and ± 0.87, respectively, and the corresponding mean AGDs were 1.48 ± 0.47, 1.38 ± 0.41, and 1.37 ± 0.42 mgy. Figure 1 illustrates the trends of these two indexes, showing a significant increase in phantom scores and a significant decrease in AGDs of all systems over the first 2 years (p < 5). No significant changes in these two measurements were observed from 2009 to The results were further categorized into film-screen mammography, computed radiography, and digital radiography for comparison. All systems showed increasing phantom scores from to 2009, with significance seen especially in film-screen mammography (p < 5). Phantom scores for both computed radiography and digital radiography were significantly higher (p < 5) than filmscreen mammography units, and digital radiography systems had significantly higher scores than computed radiography for all 3 years. On other hand, all systems showed a slight decrease in AGD over the 3-year period, with film-screen mammography and digital radiography having significantly lower doses compared with computed radiography systems (p < 5) in all 3 years. Scatterplots of mean phantom scores versus mean AGD of the combined data and the Optical Density B Coefficient of Variation D AGD (mgy) 2009 three categories are illustrated in Figure 2. An ideal system would fall in the upper left quadrant, having both high image quality and low radiation dose. Thus, digital radiography systems were considered better overall. More specifically, a clear trend of improvement could be observed for all systems between and 2009, and this advancement continued for computed radiography but not for film-screen mammography and digital radiography toward the third year Fig. 3 Sensitometric curves analysis in ( ), 2009 ( ), and 2010 ( ). A D, Analyses of film sensitometric curves obtained from surveyed processors show all curves obtained in (A) and 2009 (B) and mean curves (C) and coefficient of variation (CV) (D) for all 3 years. AJR:201, August 2013 W309

4 Film Processor Conditions Figure 3A shows large variations in film characteristic curves from. These variations were largely reduced after QA enforcement, as seen in data from 2009 (Fig. 3B). In Figure 3C, the curve had a slower speed and a lower contrast when compared with the two curves from the following years. The large variation of the processor conditions in is also clearly illustrated in Figure 3D, with greater excessive CVs in the darker regions. From, the average mid density values increased from 0 to 3 6 over 3 years, and the respective average density difference values were 1.65, 1.72, and 1.75, respectively. However, the differences of both mid density and density difference did not reach statistical significance. Passing Rates Passing rates of phantom image quality and AGD are shown in Figure 4A. Although all AGDs, except those from one unit in, were below 3 mgy, the passing rates of phantom image quality improved from to 2010 (83%, %, and 96%, respectively). The relatively low passing rate in all stemmed A 85 AGD from film-screen mammography systems, either due to insufficient background density (14 units) or lower phantom scores (9 units) or both. Figure 4B shows changes in passing rates of items related to compression conditions. Except for the compression thickness reproducibility, which was % for all years, trends of improvement were found with all other items. In, paddle edges were not visible for all systems because compression conditions were investigated only for the screening systems in that year. As for 2010, all compression-related evaluations reached passing rates greater than 98%. Passing rates of HVL, automatic exposure control reproducibility, and radiation output are summarized in Figure 4C. Each year, no more than two units failed these tests. Figure 4D shows the results from computed radiography exposure indicator confirmations. In, only 10 computed radiography systems accredited for screening were included. As for 2009, all the computed radiography systems were evaluated, which yielded a relatively low passing rate of 75% (24 of 32) compared with. In 2010, the passing rate improved to 93% (41 of 44). Phantom Image Quality B Paddle Edge Not Visible Discussion The performance evaluations of more than % of the mammographic units in Taiwan were conducted annually before and after the enforcement of mammography QA, and the results showed significantly improved phantom image quality, reduced AGD, and overall increased passing rates after the regulation took effect. The overall improvements found in this study were in agreement with previous analyses of the influences of MQSA in the United States [10, 11]. An important factor to consider when interpreting our data is the rapid switch from film-screen mammography to digital mammography systems during this period, which might be naturally occurring or partially attributed to the regulation. The proportion of digital systems increased from 41% (87 of 214) of the total units in to 60% (124 of 205) and 75% (156 of 209) in 2009 and 2010, respectively. The survey results included major digital manufacturers and models that are presently marketed in the world and could serve as a reference for performance comparisons. The improvements in phantom image quality from our survey agreed with the results of Compressed Thickness Accuracy Compressed Thickness Reproducibility Paddle Alignment Maximum Compression Force C HVL AEC Reproducibility Radiation Output D 70 Exposure Indicator Fig. 4 Passing rates of surveys from (blue bars), 2009 (red bars), and 2010 (green bars). A D, Graphs show passing rates of phantom image quality and average glandular dose (AGD) (A), compression and paddle related tests (B), half-value layer (HVL), automatic exposure control (AEC) reproducibility and radiation output (C), and computed radiography exposure indicator (D) obtained from to W310 AJR:201, August 2013

5 the FDA inspection between 19 and 2006 in which trends toward higher background density and phantom scores were observed [12, 13]. However, increased phantom scores according to the FDA corresponded with an increase in mean AGD from 0 to 1.76 mgy, whereas in our survey a reduction from 1.48 to 1.39 mgy for film-screen mammography was found. This may be explained by the fact that major improvements in phantom image quality of our film-screen mammography systems came from the film processor rather than exposure techniques. As for the comparison between film-screen mammography and digital systems, the FDA inspection data showed a significantly higher mean phantom score of 1 for digital systems when compared with a score of 12.3 for film-screen mammography systems [8]. The same trend was observed from our survey, with a mean score of 12.8 for digital radiography and 11.6 for film-screen mammography systems in In our AGD measurements, computed radiography resulted in significantly higher AGDs than film-screen mammography and digital radiography systems. AGDs were also further reduced with new systems using a tungsten target, which were introduced in Taiwan after These AGD trends for digital systems were similar to the NHSBSP equipment review report [20]. The sensitometric data obtained from surveyed processors showed significant variations in film processing conditions among facilities, especially before QA enforcement. After the regulation took effect, the variations decreased, both due to improvements of film processing and retirement of processors that were not working properly. Many filmscreen mammography systems were replaced with digital systems during this period, reflecting the rapid reduction in the number of processors from to 2010 (Table 1). In our surveyed data, passing rates for phantom image quality were slightly lower in than the following 2 years, which agreed with the trends found in the United Kingdom and the United States [7 13]. Most of the failures in our data appeared in the film-screen mammography systems, either due to inappropriately selected exposure techniques or defective processor conditions. In contrast, passing rates for digital systems were greater than 97% in all 3 years, even with stricter criteria for some manufacturers systems. Our results agreed with the results in the Digital Mammography Imaging Screening Trial QC survey in the United States [21, 22], which suggested that a phantom that was more discriminative Quality Assurance in Mammography TABLE 1: Film-Screen and Digital Systems and Distribution of Digital Manufacturers and Models in the Survey Between and 2010 in Taiwan Year Manufacturer and Model Film-screen system Digital system (total) GE Healthcare Senographe 2000 D Senographe DS Senographe Essential Hologic Lorad Selenia Siemens Healthcare Mammomat Novation DR Mammomat Inspiration IMS Giotto Image 3D Computed radiography Fujifilm Medical Systems Konica Minolta Medical Imaging Kodak Medical Imaging of image quality of digital systems should be used in the future. In this study, mammography QA regulation and inspection appeared to have a great impact on system performance that could be adjusted or calibrated by service engineers. One example was related to the compression conditions. Relatively poor paddle alignment and compression force results were initially noticed, of which 19% (23 of 124) failed, but these errors were easily corrected after the regulation took effect. In another case, the computed radiography exposure indicator accuracy served as an important medium for monitoring both image quality and patient dose. When one of the computed radiography models was first introduced to Taiwan for mammography use in 2009, all five installations failed this test in the same year. However, by 2010, only one of 10 installations failed, indicating that calibrations could be done properly and that regulation and inspection made a positive impact. Two fundamental differences between this current study and previous ones in Western developed countries were that very few mammography facilities implemented the QA program before the regulatory requirement and only a few certified diagnostic medical physicists were in Taiwan. The mammography task group of the Chinese Society of Medical Physics, Taipei, recommended the QA procedures [23] and started to train QA personnel just before regulation enforcement. Thus, on-site measurements at all mammography facilities were meaningful as initial confirmations of quality. In fact, examples of continued improvement in compression conditions and computed radiography exposure indicator accuracy from 2009 to 2010 (Fig. 4) showed the importance of on-site measurements during the inspection in Taiwan. Our unique experience may provide a valuable reference for regions in which mammography QA is not yet common practice and diagnostic medical physics professionals have not yet been established. In summary, this study analyzed data from large-scale on-site surveys of mammography systems in Taiwan from to During this period, mammography QA was enforced under the Standards for Medical Exposure Quality Assurance and use of digital mammography systems increased in Taiwan. Compared with data surveyed in, higher phantom image quality, reduced AGD, and overall increased passing rates were found in the 2 years after the regulation took effect. The positive outcomes were attributed to facilities compliance with the regulation. The results presented in this study could provide a unique reference for countries and regions where AJR:201, August 2013 W311

6 comprehensive mammography QA programs are not yet implemented in the digital era. monds GR, Houn F. Mammography in the 19s: the United States and Canada. Radiology 1999; Australas Phys Eng Sci Med 2000; 23: Institute of Physics and Engineering in Medicine Acknowledgments We thank all the staff members of the mammography facilities who participated in the surveys for their help and cooperation. References 1. Department of Health and Human Services, Food and Drug Administration web site. Mammography Quality Standards Act Regulations. 21 CFR Chapter I, Subchapter B, cfm?cfrpart=0. Updated April 1, Accessed April 11, American College of Radiology. American College of Radiology (ACR) mammography quality control manual. Reston, VA: American College of Radiology, European Commission. European guidelines for quality assurance in breast cancer screening and diagnosis, 4th ed. Luxembourg: European Commission, Hendrick RE, Klabunde C, Grivegnee A, Pou G, Ballard-Barbash R. Technical quality control practices in mammography screening programs in 22 countries. Int J Qual Health Care 2002; 14: Pan HB, Hsu GC, Liang HL, et al. Assessing an emerging nationwide population-based mammography screening program in Taiwan. J Radiol Sci 2011; 36: Conway BJ, Suleiman OH, Rueter FG, Antonsen RG, Slayton RJ. National survey of mammographic facilities in 1985, 1988, and Radiology 1994; 191: Suleiman OH, Spelic DC, McCrohan JL, Sy- 210: Spelic DC, Kaczmarek RV, Hilohi M, Belella S. United States radiological health activities: inspection results of mammography facilities. Biomed Imaging Interv J 2007; 3:e35 9. Young KC, Ramsdale ML, Rust A. Performance of mammographic equipment in the UK Breast Screening Programme in 1998/99. NHSBSP Publication No 45. Sheffield, UK: NHS Cancer Screening Programme, Hendrick RE, Chrvala CA, Plott CM, Cutter GR, Jessop NW, Wilcox-Buchalla P. Improvement in mammography quality control: Radiology 1998; 207: Pisano ED, Schell M, Rollings J, et al. Has the Mammography Quality Standards Act affected the mammography quality in North Carolina? AJR 2000; 174: Spelic DC. Dose and image quality in mammography: trends during the first decade of MQSA. US Food and Drug Administration web site. MammographyQualityStandardsActandProgram/FacilityScorecard/ucm htm. Published September 5, Updated May 5, Accessed November 11, Spelic DC. Updated trends in mammography dose and image quality. US Food and Drug Administration web site. Products/MammographyQualityStandardsActand- Program/FacilityScorecard/ucm htm. Published June 16, Updated November 17, Accessed November 11, Ng KH, DeWerd LA, Schmidt RC. Mammographic image quality and exposure in South East Asia. (IPEM). Commissioning and routine testing of mammographic X-ray systems. IPEM Report No 89. York, UK: IPEM, Wu X, Barnes GT, Tucker DM. Spectral dependence of glandular tissue dose in screen-film mammography. Radiology 1991; 179: Wu X, Gingold EL, Barnes GT, Tucker DM. Normalized average glandular dose in molybdenum target-rhodium filter and rhodium target-rhodium filter mammography. Radiology 1994; 193: Dance DR, Skinner CL, Young KC, et al. Additional factors for the estimation of mean glandular breast dose using the UK mammography dosimetry protocol. Phys Med Biol 2000; 45: Dance DR, Young KC, van Engen RE. Further factors for the estimation of mean glandular dose using the United Kingdom, European and IAEA breast dosimetry protocols. Phys Med Biol 2009; 54: Oduko JM, Young KC, Burch A, et al. Review of measurements on full field digital mammography systems. NHSBSP equipment report 01. Sheffield, UK: NHS Cancer Screening Programme, Bloomquist AK, Yaffe MJ, Pisano ED, et al. Quality control for digital mammography in the ACRIN DMIST trial: Part I. Med Phys 2006; 33: Yaffe MJ, Bloomquist AK, Mawdsley GE, et al. Quality control for digital mammography. Part II. Recommendations from ACRIN DMIST trial. Med Phys 2006; 33: Hwang YS, Wang HC, Chen CC, Liu HL, Tsai HY. Review and recommendations for physics testing of conventional and digital mammography: CSMPT Mammography Task Group Report Chin J Radiol ; 33: [in Chinese] W312 AJR:201, August 2013