H e a l t h C a r e Po l i c y a n d Q u a l i t y • O r i g i n a l R e s e a r c h Parakh et al. CT Radiation Dose Tracking in Europe Versus North America
Downloaded from www.ajronline.org by University of Auckland on 09/15/17 from IP address 130.216.158.78. Copyright ARRS. For personal use only; all rights reserved
Health Care Policy and Quality Original Research
Trans-Atlantic Comparison of CT Radiation Doses in the Era of Radiation Dose–Tracking Software Anushri Parakh1,2 Andre Euler 1,3 Zsolt Szucs-Farkas 4 Sebastian T. Schindera1,5 Parakh A, Euler A, Szucs-Farkas Z, Schindera ST
Keywords: CT, radiation dosage, reference values, registries, software DOI:10.2214/AJR.17.18087 Received February 13, 2017; accepted after revision May 4, 2017. S. T. Schindera is a consultant for Bayer Healthcare. A. Parakh has previously received personal fees from Bayer Healthcare that were unrelated to the present work. Based on a presentation at the Radiological Society of North America 2016 annual meeting, Chicago IL. Supported by funding from Bayer Healthcare. 1
Clinic of Radiology and Nuclear Medicine, D epartment of Radiology, University Hospital Basel, University of Basel, Petersgraben 4, Basel, CH 4031, Switzerland. Address correspondence to S. T. Schindera ([email protected]).
2 Department of Radiology, Massachusetts General Hospital, Boston, MA. 3 Department of Radiology, Duke University Medical Centre, Durham, NC. 4 Department of Radiology, Hospital Centre of Biel, Biel, Switzerland. 5 Department of Radiology, Kantonsspital Aarau, Aarau, Switzerland.
AJR 2017; 209:1–6 0361–803X/17/2096–1 © American Roentgen Ray Society
OBJECTIVE. The purpose of this study is to compare diagnostic reference levels from a local European CT dose registry, using radiation-tracking software from a large patient sample, with preexisting European and North American diagnostic reference levels. MATERIALS AND METHODS. Data (n = 43,761 CT scans obtained over the course of 2 years) for the European local CT dose registry were obtained from eight CT scanners at six institutions. Means, medians, and interquartile ranges of volumetric CT dose index (CTDIvol), dose-length product (DLP), size-specific dose estimate, and effective dose values for CT examinations of the head, paranasal sinuses, thorax, pulmonary angiogram, abdomen-pelvis, renal-colic, thorax-abdomen-pelvis, and thoracoabdominal angiogram were obtained using radiation-tracking software. Metrics from this registry were compared with diagnostic reference levels from Canada and California (published in 2015), the American College of Radiology (ACR) dose index registry (2015), and national diagnostic reference levels from local CT dose registries in Switzerland (2010), the United Kingdom (2011), and Portugal (2015). RESULTS. Our local registry had a lower 75th percentile CTDIvol for all protocols than did the individual internationally sourced data. Compared with our study, the ACR dose index registry had higher 75th percentile CTDIvol values by 55% for head, 240% for thorax, 28% for abdomen-pelvis, 42% for thorax-abdomen-pelvis, 128% for pulmonary angiogram, 138% for renal-colic, and 58% for paranasal sinus studies. CONCLUSION. Our local registry had lower diagnostic reference level values than did existing European and North American diagnostic reference levels. Automated radiationtracking software could be used to establish and update existing diagnostic reference levels because they are capable of analyzing large datasets meaningfully. adiation doses from medical procedures, particularly CT, have been under scrutiny in the medical and lay press recently, mainly because of the potential, but small, risk of radiation-induced cancer [1–3]. As a result, the radiology community (radiologists, technologists, medical physicists, and vendors) has been spending considerable time and monetary resources to optimize the radiation dose from CT to ensure patient safety without compromising diagnostic information. Legislative bodies in Europe and the United States have also published various directives and position papers addressing the burden of radiation dose [4–6]. In addition to multiple radiation dose awareness campaigns and referral guidelines for appropriate justification, investments are being made in innovations, such as CT scanners equipped with dose-efficient technology, as well as in new informat-
R
ics solutions, such as clinical decision support tools and radiation dose–tracking software. Radiation-tracking software was recently introduced by several different vendors. Such software tracks, analyzes, and archives dose metrics for large samples to aggregate the big data; therefore, they could facilitate the creation of dose registries that could be compared with various published diagnostic reference levels [7–9]. The application of diagnostic reference levels as an auditing tool for review of protocols and equipment for the optimization of radiation doses has been widely recommended [10, 11]. Historically, the collection of dose data from at least 10 samples for each examination to provide an estimate of exposure has been the standard practice because of the time-consuming nature of manual data collection and analysis. Even today in the United States, as part of the American College
AJR:209, December 2017 1
Downloaded from www.ajronline.org by University of Auckland on 09/15/17 from IP address 130.216.158.78. Copyright ARRS. For personal use only; all rights reserved
Parakh et al. of Radiology (ACR) accreditation program, institutions tend to submit data from small samples, which are considered as the facility’s best and do not tend to represent average or typical work performed in routine clinical practice [12]. An important caveat of using small data samples of dose metrics is that there is greater variability [13]. As practices are becoming larger and more sophisticated, the era of managing by small samples without having a clear overview of the entire practice is decidedly over. Furthermore, the institutions that have invested in doseefficient CT technologies must have the opportunity to measure the outcomes of their enduring investments and efforts. Radiationtracking software provides a pragmatic solution for meaningful radiation dose benchmarking and the monitoring of such dose optimization processes. Recently, there have been publications from academic medical centers in California and Toronto with CT dose data gathered using radiation-tracking software [7, 8]. Valuable CT dose data are also available through the ACR dose index registry [14, 15]. However, these recent CT radiation dose data are mostly from North America and, to our knowledge, there have been no recent publications with comparable data from Europe. The currently available published national diagnostic reference levels from various European countries are a few years old and therefore do not include the latest technical developments for dose reduction, such as iterative reconstruction or automatic tube voltage modulation. The assumptions of some authorities that CT doses in Europe are generally lower than those in the United States prompted us to compare the current dose data from a local European CT dose registry with recently published dose data from United States and Canada [8, 16]. In addition, we also assessed
the relevance of the various published national diagnostic reference levels from European countries. Materials and Methods Site, Scanner, and Software Specifications
The institutional review board of the University Hospital of Basel approved this retrospective observational study and waived the need to obtain informed consent because the patient data were anonymized for the analysis of the radiation dose. Between January 1, 2014, and December 31, 2015, the data from eight CT scanners at six medical institutions located in the same state of Switzerland were automatically tracked with radiation-tracking software (Radimetrics, Bayer Healthcare). These included three regional, two private (one focusing on orthopedic surgery and the other on geriatric care), and one university hospital. All the hospitals had one CT scanner, except the university hospital, which had three CT scanners installed. The geometry of the CT scanners represented a broad range with regard to detector rows (16–256 detectors) and dose reduction capabilities (e.g., five of eight had iterative reconstruction) (Table 1). All the scanners were equipped with automatic tube current modulation. With the aim of establishing a local CT dose registry, radiation-tracking software from the same manufacturer (Radimetrics, Bayer HealthCare) was installed for data from all six participating hospitals. The radiation-tracking software was connected to PACSs from four different vendors (iSite, Philips Healthcare; Impax, Agfa; JiveX, Visus; and Centricity, GE Healthcare). From the PACS, radiationtracking software collects the radiation exposure details, which are stored in the DICOM header or radiation dose structured report, and stores them in three local servers. The dose data at the site of the local servers are then anonymized before they are uploaded to a single master server, called the patient dose repository. The university hospital was identified as the nodal institution, and access
to information from the main server for further analyses was available only to a dedicated dose team from the nodal institution. The dose team consisted of a radiologist, technologist, medical physicist, and information technology specialist. A member of the dose team (the technologist) visited the different facilities to standardize the nomenclature of all of the protocols and created master protocols. The Radlex playbook, published by the Radiological Society of North America, was applied for protocol standardization [17]. This was undertaken to match the various device protocols to a homogeneous unified protocol and to facilitate protocol-based comparison. A radiologist (with 12 years of experience with CT dose optimization in clinical practice and research) confirmed the matching of all of the device protocols with the master protocols to avoid mismatching and to ensure reliable analyses of doses. Radiation-tracking software provided information about the following four dose metrics for further analyses: volumetric dose index (CTDIvol), dose-length product (DLP), size-specific dose estimate, and effective dose. CTDIvol is a measure of radiation delivered to a volume of a standardized phantom (in adults, 16 cm for head or 32 cm for body) from a concatenation of contiguous irradiations and therefore represents the radiation output of a scanner for a particular examination. It is not a direct measurement of patient dose because it does not account for patient size [18]. DLP, which is based on CTDIvol and scan length, reflects the total dose to a phantom and is therefore preferred over CTDIvol in multiphasic studies. Effective dose, which was described by the International Commission on Radiation Protection in 2007 as a protection quantity, is used as a comparative measurement to assess compliance with dose limits [19]. These three metrics were analyzed to compare radiology practice with regard to radiation exposure and not to estimate individual patient risk. Size-specific dose estimate, however, is an estimate of patient dose because it is adjusted to
TABLE 1: CT Scanner Specifications as Part of the Local CT Dose Registry CT Scanner No.
Vendor
Scanner Name
No. of Detector Rows
Iterative Reconstruction
Automatic Tube Current Modulation
Automatic Tube Voltage Modulation
1
Philips Healthcare
Brilliance iCT
256
iDose
Yes
No
2
Siemens Healthcare
Emotion
16
No
Yes
No
3
Siemens Healthcare
Somatom Definition Flash
128
SAFIRE
Yes
Yes
4
Siemens Healthcare
Sensation
16
No
Yes
No
5
Siemens Healthcare
Emotion
16
No
Yes
No
6
Siemens Healthcare
Somatom Definition AS+
128
SAFIRE
Yes
Yes
7
Siemens Healthcare
Somatom Definition Edge
128
ADMIRE
Yes
Yes
8
Siemens Healthcare
Somatom Definition Flash
128
SAFIRE
Yes
Yes
Note—SAFIRE = sinogram-affirmed iterative reconstruction, ADMIRE = advanced model iterative reconstruction.
2
AJR:209, December 2017
the size of the patient and indicates radiation dose to the soft tissues within a scan volume. It is measured for scans of the thorax and abdomen, and for its calculation, the radiation-tracking software determines the effective patient diameter at the mid scan level. CT scans as part of PET and interventions were excluded from the study. In addition, no scans from children were included because the local pediatric university hospital was not a member of the local dose registry during the study period.
Fig. 1—Graph of volumetric CT dose index (CTDIvol) for thoracic CT performed with eight different scanners. Lower and upper ends of vertical bars show 25th and 75th percentile values. Horizontal markers represent 50th percentile.
10 8 6 4 2
Protocols, Sourced Data, and Statistical Analysis
For the dose analysis, we focused on the eight most commonly performed examinations at all hospitals, which were all indication-based and single-phase protocols. These examinations were identified to be unenhanced CT of the head (e.g., to exclude cerebral hemorrhage), paranasal sinuses (e.g., to exclude sinusitis), thorax (e.g., to exclude pulmonary nodules or pneumonia), pulmonary angiogram (e.g., to exclude pulmonary embolism), unenhanced renal colic protocol (e.g., to exclude urolithiasis), enhanced abdomen-pelvis in the portal venous phase (e.g., assessment of an acute abdomen), thorax-abdomen-pelvis (e.g., for tumor staging), and angiography of thorax-abdomen-pelvis (e.g., assessment of an aneurysm or dissection). Radiation-tracking software provided the values of the dose metrics (mean, median, and interquartile range). The graphical representation of data was performed using Microsoft Excel for Mac (version 15.33). The mean, median, and interquartile range from the current study were compared with the recently published data from the University of Toronto medical centers (published in 2015) [7], the University of California medical centers (published in 2015) [8], and the ACR dose index registry (data from semiannual report for adults between July and December 2015) [14]. We also compared our data with the national diagnostic reference levels of the country in which the local CT dose registry was located (Switzerland, published in 2010 [20],
12
CTDIvol (mGy)
Downloaded from www.ajronline.org by University of Auckland on 09/15/17 from IP address 130.216.158.78. Copyright ARRS. For personal use only; all rights reserved
CT Radiation Dose Tracking in Europe Versus North America
0
1
2
the United Kingdom, published in 2011 [21], and Portugal, published in 2014 [22]) (Table 2). The sourced data were compared with our study by matching them with indication and where available for single phase.
Results The total number of CT examinations during the 24-month period was 85,187. The aforementioned eight protocols analyzed in this study were responsible for approximately 51.3% of all of the tracked CT scans (n = 43,761). The eight most common CT protocols were CT of the head (n = 15,539; 35.5% of the analyzed studies), thorax (n = 7684; 17.6% of the analyzed studies), abdomen-pelvis (n = 6132; 14.0% of the analyzed studies), thoraxabdomen-pelvis (n = 4890; 11.1% of the analyzed studies), pulmonary angiogram (n = 3531; 8.1% of the analyzed studies), urolithiasis (n = 2517; 5.8% of the analyzed studies), paranasal sinus (n = 1875; 4.3% of the analyzed studies), and angiography of the thorax-abdomen-pelvis (n = 1593; 3.6% of the analyzed studies). The interquartile ranges of different dose metrics from our local dose registry are enu-
3
4 5 6 CT Scanner Number
7
8
merated in Table 3. For all of the protocols, there was broad variation among the eight CT scanners for the different radiation dose metrics (Table 3). For example, the smallest and largest values of the 75th percentile of the CTDIvol for CT thorax were 2.3 and 11 mGy, respectively. There was a fourfold difference in the median value of CTDIvol for CT of the thorax between scanner 3 (8.4 mGy) and scanner 8 (1.8 mGy), which are the same scanners at two different institutions (Fig. 1). In addition, scanner 8 showed substantially less variability than did scanner 3. A much smaller variation in the dose metrics was observed for head CT among the eight CT scanners. For CTDIvol, the smallest value of the 75th percentile was 34 mGy and the largest value was 64 mGy. The diagnostic reference levels from the ACR dose index registry [14] and California [8] were substantially higher than our data (Fig. 2). Compared with our study, the ACR dose index registry had higher dose values for 75th percentile CTDIvol values by 55% for head, 240% for thorax, 28% for abdomen-pelvis, 42% for thorax-abdomen-pelvis, 128%
TABLE 2: CT Scanner Specifications in the Sourced Data Sourced Data University of Toronto Medical Centers [7] University of California Medical Centers [8]
American College of Radiology dose index registry [14]
Data Collection Period
No. of Scanners
Scanner Specifications
2010–2013
2
64-MDCT
2013
34
One 8-MDCT, five 16-MDCT, rest 64-MDCT or higher; all with tube current modulation; 21 with iterative reconstruction
Report from July–December 2015 Not applicable Not applicable
Switzerland [20]
2007–2009
179
16- to 64-MDCT most common (ranged from single detector to 256-MDCT)
United Kingdom [21]
2010–2011
182
16- to 64-MDCT most common (ranged from 4to 320-MDCT)
Portugal [22]
2011–2012
41
16-MDCT most common
AJR:209, December 2017 3
4 1593
Angiography thorax, abdomen, pelvis
1.5 (1.2–5.7)
9.0 (3.3–16.1) 10.2 (4–14.8)
12.3 (7.6–17.7)
3.7 (2.7–11.3) 5.9 (3.5–13)
7.5 (5.5–13.9)
2.8 (2–10) 4.9 (3.2–11.1)
4.9 (3.8–7.9)
290 (111–532) 384 (220–613)
215 (167–473) 164 (132–297) 486 822 (354–778) (454–1140)
168 (141–388) 135 (43–214) 327 (234–708)
8.3 (4.0–12.6) 17.0 (6–26.5) 9.3 (5.9–13.8)
5.2 (4.0–10.6) 11.8 (6–26.3) 5.7 (4.3–11.3)
4.0 (3.5–8.7) 7.0 (2.6–17.0) 3.9 (2.8–10.5)
CTDIvol (mGy)
243 (142–338)
124 (84–362)
84 (63–273)
7.0 (3–13.1)
3.3 (2.2–9)
15
2.3 (1.6–7.3)
CTDIvol (mGy)
10.3 (7.7–13)
14.4 16.8 (9.7–16.5) (11.4–20.4)
11 (7.8–15)
826 1051 (638–950) (863–1303)
622 (445–761)
14.1 (9.6–17.6)
11.5 (7.7–13.6)
8.6 (6.2–11.2)
5.12 (3.4–13.8)
0.3 (0.09–0.5)
2.4 (1.9–6.2)
6.6 (5.1–8.4)
15.5 (11.1–18.1)
12.5 (8–15.3)
10.2 (6.6–14.5)
544 758 (445–666) (593–977)
13.5 405 (11.7–16.2) (326–531)
1.2 (0.9–5)
7 (2.8–12.9)
3.1 (2.4–10.4)
2.4 (1.9–9.5)
10.5 (8.8–12.5)
86 (67–341)
8.2 (6.5–11.2)
65 (51–242)
5.1 (2.3–11)
2.4 (1.8–8.4)
1.7 (1.7–2.8)
200 (85–468)
25th
75th
50th
SSDE (mGy) 25th
75th
1.8 (1.3–6.5)
50th
DLP (mGy·cm)
678 715 808 (666–1118) (694–1160) (728–1212)
25th
40.3 (34–64)
75th
38.4 (34–64)
50th
CTDIvol (mGy) 37.9 (34–64)
25th
20
Note—Data are mean (range) values from the eight CT scanners.
1875
2517
3531
4890
6132
7684
15,539
No. of Scans
Paranasal sinuses
Renal colic
Pulmonary angiography
Thorax, abdomen, pelvis
Abdomen-pelvis
Thorax
Head
Anatomic Region 2.9 (1.9–4)
4 (1.4–7.9) 11 (8.7–13.8)
1.9 (1.8–2.9) 1.5 (1.2–6.5) 8.2 (6.3–9.4)
5.5 (2–10) 5.3 (2.9–9.4) 0.6 (0.4–0.9) 13.4 (7.2–19.4)
2.2 (1.6–7.3) 3.1 (2.2–7.2) 0.5 (0.3–0.8) 7.7 (5.8–16.1)
13.8 16.6 (10.8–16.2) (13.5–20.6)
75th
50th
ED (mSv)
TABLE 3: Summary of the 25th, 50th, and 75th Percentiles From Eight CT Scanners of the Local Registry for Volume CT Dose Index (CTDIvol), Dose-Length Product (DLP), Size-Specific Dose Estimate (SSDE), and Effective Dose (ED)
Downloaded from www.ajronline.org by University of Auckland on 09/15/17 from IP address 130.216.158.78. Copyright ARRS. For personal use only; all rights reserved
Parakh et al.
80
0
0
Thorax
Present Study National United Kingdom Portugal
Head
25 Present Study National United Kingdom Portugal
AbdomenPelvis
Anatomic Region
ThoraxAbdomenPelvis
Pulmonary Angiogram
Toronto ACR-DIR California
60
40
20
Paranasal Sinus Anatomic Region
A
Toronto ACR-DIR California
10
5
Renal Colic
Fig. 2—Comparison of 75th percentile of volumetric CT dose index (CTDIvol) between dose registries from North America and published European diagnostic reference levels for different anatomic regions. A and B, Graphs show data for CT of head and paranasal sinuses (A) and for CT of thorax, abdomen-pelvis, thorax-abdomen-pelvis, pulmonary angiogram, and renal colic regions (B). ACR-DIR = American College of Radiology dose index registry.
B
AJR:209, December 2017
Downloaded from www.ajronline.org by University of Auckland on 09/15/17 from IP address 130.216.158.78. Copyright ARRS. For personal use only; all rights reserved
CT Radiation Dose Tracking in Europe Versus North America for pulmonary angiogram, 138% for renal colic, and 58% for paranasal sinuses studies. The comparison with the data from California for the 75th percentile CTDIvol revealed similar differences (55% for head, 240% for thorax, 21% for abdomen-pelvis, 28% for thorax-abdomen-pelvis, and 71% for paranasal sinus studies). The largest differences were observed between our data and all of the sourced published data for CT of the thorax (up to 240% difference from California and ACR dose index registry) and for renal colic CT (up to 138% difference from ACR dose index registry), followed by pulmonary angiography (up to 128% difference from ACR dose index registry). Notably, the recently published American and Canadian institutional diagnostic reference levels were also higher than the rest of the European diagnostic reference levels for all protocols except for CT of the head. Discussion Our succinct snapshot of CT radiation dose data for eight indication-based single-phase CT protocols collected within a local European dose registry over a 24-month period using radiation-tracking software offers a basis for benchmarking dose levels across institutions for a heuristic assessment of dose practice patterns with CT. The current data could also be used to judge the relevance of previously published national diagnostic reference levels and to assess intercontinental differences in diagnostic reference levels. The 25th, 50th, and 75th percentile values in the current study could be applied as an index of benchmarking for other institutions. Facilities with exposures consistently less than the diagnostic reference levels, which are defined as the 75th percentiles, could aim for the 25th percentiles, which are defined as the target values [16, 20]. Aiming for such target values triggers an incentive for continuous optimization in CT dose reduction while maintaining an acceptable image quality. Our results show that prior national European diagnostic reference levels might be outdated, likely because of the rapid evolution of CT technology, as well as the wider acceptance of dose-optimized protocols [21]. The most divergent value was noted for the thorax, suggesting that the use of the currently available dose-reduction techniques, especially iterative reconstruction, could be applied to achieve a greater degree of dose reduction. This finding was in agreement with a recent study by Thomas et al. [23], which reported the effect of the use of iterative re-
construction in clinical routine at a national level for frequently performed CT protocols with 20–30% dose efficiency. In contrast to this finding in the thorax, the data for abdomen-pelvis and thorax-abdomen-pelvis examinations in the current study showed results very similar to those of the sourced European diagnostic reference levels. A possible reason for this outcome is that the existing scanning parameters are adequate and necessary, even with the application of iterative reconstruction techniques for acceptable image quality without compromising diagnostic confidence, especially in an anatomic area with low inherent soft-tissue contrast. Our assumption was in accordance with a phantom study, which proved that reducing radiation doses in tandem with an iterative reconstruction algorithm affected the detection of low-contrast lesions [24]. A broad range of differences was observed between our data and the published data from the California medical centers, as well as the ACR data, for various protocols. One protocol that showed this discrepancy well was the renal colic examination, which is a well-accepted example of a lowdose protocol. In addition to renal colic examinations being performed for pathologic entities with high inherent contrast, in which a higher noise level can be accepted, young patients frequently undergo such examinations. According to a scientific report using the ACR dose index registry data from 2014 with 49,903 examinations, the mean institutional DLP for this type of study was 746 mGy·cm [15]. In contrast, the mean DLP for renal colic examinations was 55% lower in our study, which reaffirms that use of doseoptimized techniques is infrequent in the United States. Reasons for the differences in practice are complex and could be explained by a gross discrepancy in knowledge or willingness to use dose-optimized protocols. Consultation of radiologic institutions by an experienced dose-optimization team consisting of a radiologist, technologist, and medical physicist could promote the process of CT dose reduction while maintaining diagnostic confidence [25]. The applied strategies include teaching practical tips, as well as effective use of state-of-the-art dose reduction techniques. Another possible explanation is the advancement in CT technology in the past few years, because the North American data were collected between 2010 and 2013. Dose management systems are analytic solutions capable of gathering and using
big data for effective clinical decision making. Their primary advantage is the ability to process large amounts of robust information about numerous dose descriptors into dashboards, which can be used for quality assurance and to assess intrainstitutional interscanner variations. Dose benchmarking will play an increasing role in CT dose optimization in the future because the initiation of an optimization process should be done in real time and should be based on facts rather than intuition. When benchmarking with others, attention to the technical environment is imperative. It would be futile to compare a single-detector CT scanner with new-generation scanners equipped with various dosereduction techniques. In the future, we intend to use these data within the institutions of the registry to optimize practice. To improve the value of radiation-tracking software in the future, it is of great importance to add objective image quality as a parameter. Addressing only radiation dose parameters while neglecting image quality does not provide a holistic approach to optimizing CT protocols in accordance with the ALARA (as low as reasonably achievable) principle. Unfortunately, the quantitative image quality parameters that are currently available, such as image noise or contrastto-noise ratio, cannot be used for such a task because they do not account for diagnostic accuracy. A promising technique for such a task is the use of model observers. Comparison of useful objective image quality parameters for radiation-tracking software is a topic of ongoing research in our institution. There are a few limitations and hurdles to this study that must be mentioned. First, we did not check the scans and their indications because the data in the master server only include dose metrics and not CT images. However, during the establishment of the dose registry, we exerted great effort in protocol matching using the Radlex playbook to avoid nonstandardization of protocol nomenclature. Furthermore, the large number of CT scans in our study diminished the effect of such cases on the average dose metrics. Second, we did not compare the image quality of the images in our study with other national and international diagnostic reference levels, which also did not provide such information. Because the data were collected from routine clinical practice, it must be assumed that image quality reflects the standard practice of the institutions, interpreted routinely by the radiolo-
AJR:209, December 2017 5
Downloaded from www.ajronline.org by University of Auckland on 09/15/17 from IP address 130.216.158.78. Copyright ARRS. For personal use only; all rights reserved
Parakh et al. gist. Third, the patient-built, protocol, scanner-type, and reconstruction techniques of sourced data were not available. Because of the unavailability of these parameters and the gaps in time, a direct comparison of all the factors was not possible; therefore, summary values have been compared. In conclusion, our local CT dose registry showed lower dose values than did the existing older published European diagnostic reference levels and recent North American diagnostic reference levels. We also found substantial trans-Atlantic differences in dose practice, with European diagnostic reference levels having consistently lower values than North American diagnostic reference levels. Programmatic radiation dose monitoring has emerged as a valuable auditing tool to understand and manage CT practice. Visualization of differences in CT practice by dose benchmarking is a promising self-assessment contrivance to call attention to an institution’s failure to participate in dose reduction and could therefore initiate an appropriate optimization process. References
1. Journy N, Rehel JL, Ducou Le Pointe H, et al. Are the studies on cancer risk from CT scans biased by indication? Elements of answer from a large-scale cohort study in France. Br J Cancer 2015; 112:185–193 2. Redberg R, Smith-Bindman R. We are giving ourselves cancer. New York Times website. www.nytimes. com/2014/01/31/opinion/we-are-giving-ourselvescancer.html. Published January 30, 2014. Accessed December 20, 2015 3. Mathews JD, Forsythe AV, Brady Z, et al. Cancer risk in 680,000 people exposed to computed tomography scans in childhood or adolescence: data linkage study of 11 million Australians. BMJ 2013; 346:f2360 4. Joint Commission. Joint Commission announces new and revised diagnostic imaging standards. Joint Commission website. www.dosemonitor.com/files/ 4413/9035/3871/JCAHO_Radiation_Dose_release_ 2.pdf.. Published December 20, 2013. Accessed August 16, 2017 5. U.S. Food and Drug Administration. White paper:
6
initiative to reduce unnecessary radiation exposure from medical imaging. U.S. Food and Drug Administration website. www.fda.gov/downloads/ Radiation-EmittingProducts/RadiationSafety/ RadiationDoseReduction/UCM200087.pdf. February 2010. Accessed April 29, 2016 6. European Commission. Council directive 2013/59/ Euratom of 5 December 2013 laying down basic safety standards for protection against the dangers arising from exposure to ionising radiation, and repealing Directives 89/618/Euratom, 90/641/Euratom, 96/29/Euratom, 97/43/Euratom and 2003/122/ Euratom. European Commission website. ec.europa.eu/energy/sites/ener/files/documents/CELEX32013L0059-EN-TXT.pdf. Published 2014. Accessed August 4, 2017 7. MacGregor K, Li I, Dowdell T, Gray BG. Identifying institutional diagnostic reference levels for CT with radiation dose index monitoring software. Radiology 2015; 276:507–517 8. Smith-Bindman R, Moghadassi M, Wilson N, et al. Radiation doses in consecutive CT examinations from five University of California medical centers. Radiology 2015; 277:134–141 9. Boos J, Meineke A, Rubbert C, et al. Cloud-based CT dose monitoring using the DICOM-structured report: fully automated analysis in regard to national diagnostic reference levels. Rofo 2016; 188:288–294 10. [No authors listed.] Radiological protection and safety in medicine: a report of the International Commission on Radiological Protection. Ann ICRP 1996; 26:1–47 [Erratum in Ann ICRP 1997; 27:61] 11. [No authors listed.] ICRP Publication 105: radiation protection in medicine. Ann ICRP 2007; 37:1–63 12. American College of Radiology. CT accreditation program requirements. Reston, VA: American College of Radiology, 2013 13. Taylor S, Van Muylem A, Gevenois PA, Tack D. Intra-institution CT dose surveys in adults: what sample size for what precision? Electronic Presentation Online System website. posterng.netkey.at/ esr/viewing/index.php?module=viewing_poster& doi=10.1594/ecr2015/B-0785. Published 2015. Accessed August 4, 2017 14. American College of Radiology. National radiology dose index registry. American College of Radiology website. www.acr.org/~/media/ACR/Documents/
PDF/QualitySafety/NRDR/DIR/DIRSampleReport. pdf. Published 2017. Accessed August 4, 2017 15. Lukasiewicz A, Bhargavan-Chatfield M, Coombs L, et al. Radiation dose index of renal colic protocol CT studies in the United States: a report from the American College of Radiology National Radiology Data Registry. Radiology 2014; 271:445–451 16. Brink JA, Miller DL. U.S. national diagnostic reference levels: closing the gap. Radiology 2015; 277:3–6 17. Langlotz CP. RadLex: a new method for indexing online educational materials. RadioGraphics 2006; 26:1595–1597 18. Strauss KJ. Dose indices: everybody wants a number. Pediatr Radiol 2014; 44(suppl 3):450–459 19. [No authors listed.] The 2007 recommendations of the International Commission on Radiological Protection: ICRP publication 103. Ann ICRP 2007; 37:1–332 20. Treier R, Aroua A, Verdun FR, Samara E, Stuessi A, Trueb PR. Patient doses in CT examinations in Switzerland: implementation of national diagnostic reference levels. Radiat Prot Dosimetry 2010; 142:244–254 21. Shrimpton P, Hillier M, Meeson S, Golding S. Doses from computed tomography (CT) examinations in the UK: 2011 review. Public Health England website. www.gov.uk/government/publications/dosesfrom-computed-tomography-ct-examinations-inthe-uk. September 1, 2014. Accessed December 23, 2015 22. Santos J, Foley S, Paulo G, McEntee MF, Rainford L. The establishment of computed tomography diagnostic reference levels in Portugal. Radiat Prot Dosimetry 2014; 158:307–317 23. Thomas P, Hayton A, Beveridge T, Marks P, Wallace A. Evidence of dose saving in routine CT practice using iterative reconstruction derived from a national diagnostic reference level survey. Br J Radiol 2015; 88:20150380 24. Schindera ST, Odedra D, Raza SA, et al. Iterative reconstruction algorithm for CT: can radiation dose be decreased while low-contrast detectability is preserved? Radiology 2013; 269:511–518 25. Schindera ST, Treier R, von Allmen G, et al. An education and training programme for radiological institutes: impact on the reduction of the CT radiation dose. Eur Radiol 2011; 21:2039–2045
AJR:209, December 2017
Downloaded from www.ajronline.org by University of Auckland on 09/15/17 from IP address 130.216.158.78. Copyright ARRS. For personal use only; all rights reserved
Health Care Policy and Quality Original Research
Trans-Atlantic Comparison of CT Radiation Doses in the Era of Radiation Dose–Tracking Software Anushri Parakh1,2 Andre Euler 1,3 Zsolt Szucs-Farkas 4 Sebastian T. Schindera1,5 Parakh A, Euler A, Szucs-Farkas Z, Schindera ST
Keywords: CT, radiation dosage, reference values, registries, software DOI:10.2214/AJR.17.18087 Received February 13, 2017; accepted after revision May 4, 2017. S. T. Schindera is a consultant for Bayer Healthcare. A. Parakh has previously received personal fees from Bayer Healthcare that were unrelated to the present work. Based on a presentation at the Radiological Society of North America 2016 annual meeting, Chicago IL. Supported by funding from Bayer Healthcare. 1
Clinic of Radiology and Nuclear Medicine, D epartment of Radiology, University Hospital Basel, University of Basel, Petersgraben 4, Basel, CH 4031, Switzerland. Address correspondence to S. T. Schindera ([email protected]).
2 Department of Radiology, Massachusetts General Hospital, Boston, MA. 3 Department of Radiology, Duke University Medical Centre, Durham, NC. 4 Department of Radiology, Hospital Centre of Biel, Biel, Switzerland. 5 Department of Radiology, Kantonsspital Aarau, Aarau, Switzerland.
AJR 2017; 209:1–6 0361–803X/17/2096–1 © American Roentgen Ray Society
OBJECTIVE. The purpose of this study is to compare diagnostic reference levels from a local European CT dose registry, using radiation-tracking software from a large patient sample, with preexisting European and North American diagnostic reference levels. MATERIALS AND METHODS. Data (n = 43,761 CT scans obtained over the course of 2 years) for the European local CT dose registry were obtained from eight CT scanners at six institutions. Means, medians, and interquartile ranges of volumetric CT dose index (CTDIvol), dose-length product (DLP), size-specific dose estimate, and effective dose values for CT examinations of the head, paranasal sinuses, thorax, pulmonary angiogram, abdomen-pelvis, renal-colic, thorax-abdomen-pelvis, and thoracoabdominal angiogram were obtained using radiation-tracking software. Metrics from this registry were compared with diagnostic reference levels from Canada and California (published in 2015), the American College of Radiology (ACR) dose index registry (2015), and national diagnostic reference levels from local CT dose registries in Switzerland (2010), the United Kingdom (2011), and Portugal (2015). RESULTS. Our local registry had a lower 75th percentile CTDIvol for all protocols than did the individual internationally sourced data. Compared with our study, the ACR dose index registry had higher 75th percentile CTDIvol values by 55% for head, 240% for thorax, 28% for abdomen-pelvis, 42% for thorax-abdomen-pelvis, 128% for pulmonary angiogram, 138% for renal-colic, and 58% for paranasal sinus studies. CONCLUSION. Our local registry had lower diagnostic reference level values than did existing European and North American diagnostic reference levels. Automated radiationtracking software could be used to establish and update existing diagnostic reference levels because they are capable of analyzing large datasets meaningfully. adiation doses from medical procedures, particularly CT, have been under scrutiny in the medical and lay press recently, mainly because of the potential, but small, risk of radiation-induced cancer [1–3]. As a result, the radiology community (radiologists, technologists, medical physicists, and vendors) has been spending considerable time and monetary resources to optimize the radiation dose from CT to ensure patient safety without compromising diagnostic information. Legislative bodies in Europe and the United States have also published various directives and position papers addressing the burden of radiation dose [4–6]. In addition to multiple radiation dose awareness campaigns and referral guidelines for appropriate justification, investments are being made in innovations, such as CT scanners equipped with dose-efficient technology, as well as in new informat-
R
ics solutions, such as clinical decision support tools and radiation dose–tracking software. Radiation-tracking software was recently introduced by several different vendors. Such software tracks, analyzes, and archives dose metrics for large samples to aggregate the big data; therefore, they could facilitate the creation of dose registries that could be compared with various published diagnostic reference levels [7–9]. The application of diagnostic reference levels as an auditing tool for review of protocols and equipment for the optimization of radiation doses has been widely recommended [10, 11]. Historically, the collection of dose data from at least 10 samples for each examination to provide an estimate of exposure has been the standard practice because of the time-consuming nature of manual data collection and analysis. Even today in the United States, as part of the American College
AJR:209, December 2017 1
Downloaded from www.ajronline.org by University of Auckland on 09/15/17 from IP address 130.216.158.78. Copyright ARRS. For personal use only; all rights reserved
Parakh et al. of Radiology (ACR) accreditation program, institutions tend to submit data from small samples, which are considered as the facility’s best and do not tend to represent average or typical work performed in routine clinical practice [12]. An important caveat of using small data samples of dose metrics is that there is greater variability [13]. As practices are becoming larger and more sophisticated, the era of managing by small samples without having a clear overview of the entire practice is decidedly over. Furthermore, the institutions that have invested in doseefficient CT technologies must have the opportunity to measure the outcomes of their enduring investments and efforts. Radiationtracking software provides a pragmatic solution for meaningful radiation dose benchmarking and the monitoring of such dose optimization processes. Recently, there have been publications from academic medical centers in California and Toronto with CT dose data gathered using radiation-tracking software [7, 8]. Valuable CT dose data are also available through the ACR dose index registry [14, 15]. However, these recent CT radiation dose data are mostly from North America and, to our knowledge, there have been no recent publications with comparable data from Europe. The currently available published national diagnostic reference levels from various European countries are a few years old and therefore do not include the latest technical developments for dose reduction, such as iterative reconstruction or automatic tube voltage modulation. The assumptions of some authorities that CT doses in Europe are generally lower than those in the United States prompted us to compare the current dose data from a local European CT dose registry with recently published dose data from United States and Canada [8, 16]. In addition, we also assessed
the relevance of the various published national diagnostic reference levels from European countries. Materials and Methods Site, Scanner, and Software Specifications
The institutional review board of the University Hospital of Basel approved this retrospective observational study and waived the need to obtain informed consent because the patient data were anonymized for the analysis of the radiation dose. Between January 1, 2014, and December 31, 2015, the data from eight CT scanners at six medical institutions located in the same state of Switzerland were automatically tracked with radiation-tracking software (Radimetrics, Bayer Healthcare). These included three regional, two private (one focusing on orthopedic surgery and the other on geriatric care), and one university hospital. All the hospitals had one CT scanner, except the university hospital, which had three CT scanners installed. The geometry of the CT scanners represented a broad range with regard to detector rows (16–256 detectors) and dose reduction capabilities (e.g., five of eight had iterative reconstruction) (Table 1). All the scanners were equipped with automatic tube current modulation. With the aim of establishing a local CT dose registry, radiation-tracking software from the same manufacturer (Radimetrics, Bayer HealthCare) was installed for data from all six participating hospitals. The radiation-tracking software was connected to PACSs from four different vendors (iSite, Philips Healthcare; Impax, Agfa; JiveX, Visus; and Centricity, GE Healthcare). From the PACS, radiationtracking software collects the radiation exposure details, which are stored in the DICOM header or radiation dose structured report, and stores them in three local servers. The dose data at the site of the local servers are then anonymized before they are uploaded to a single master server, called the patient dose repository. The university hospital was identified as the nodal institution, and access
to information from the main server for further analyses was available only to a dedicated dose team from the nodal institution. The dose team consisted of a radiologist, technologist, medical physicist, and information technology specialist. A member of the dose team (the technologist) visited the different facilities to standardize the nomenclature of all of the protocols and created master protocols. The Radlex playbook, published by the Radiological Society of North America, was applied for protocol standardization [17]. This was undertaken to match the various device protocols to a homogeneous unified protocol and to facilitate protocol-based comparison. A radiologist (with 12 years of experience with CT dose optimization in clinical practice and research) confirmed the matching of all of the device protocols with the master protocols to avoid mismatching and to ensure reliable analyses of doses. Radiation-tracking software provided information about the following four dose metrics for further analyses: volumetric dose index (CTDIvol), dose-length product (DLP), size-specific dose estimate, and effective dose. CTDIvol is a measure of radiation delivered to a volume of a standardized phantom (in adults, 16 cm for head or 32 cm for body) from a concatenation of contiguous irradiations and therefore represents the radiation output of a scanner for a particular examination. It is not a direct measurement of patient dose because it does not account for patient size [18]. DLP, which is based on CTDIvol and scan length, reflects the total dose to a phantom and is therefore preferred over CTDIvol in multiphasic studies. Effective dose, which was described by the International Commission on Radiation Protection in 2007 as a protection quantity, is used as a comparative measurement to assess compliance with dose limits [19]. These three metrics were analyzed to compare radiology practice with regard to radiation exposure and not to estimate individual patient risk. Size-specific dose estimate, however, is an estimate of patient dose because it is adjusted to
TABLE 1: CT Scanner Specifications as Part of the Local CT Dose Registry CT Scanner No.
Vendor
Scanner Name
No. of Detector Rows
Iterative Reconstruction
Automatic Tube Current Modulation
Automatic Tube Voltage Modulation
1
Philips Healthcare
Brilliance iCT
256
iDose
Yes
No
2
Siemens Healthcare
Emotion
16
No
Yes
No
3
Siemens Healthcare
Somatom Definition Flash
128
SAFIRE
Yes
Yes
4
Siemens Healthcare
Sensation
16
No
Yes
No
5
Siemens Healthcare
Emotion
16
No
Yes
No
6
Siemens Healthcare
Somatom Definition AS+
128
SAFIRE
Yes
Yes
7
Siemens Healthcare
Somatom Definition Edge
128
ADMIRE
Yes
Yes
8
Siemens Healthcare
Somatom Definition Flash
128
SAFIRE
Yes
Yes
Note—SAFIRE = sinogram-affirmed iterative reconstruction, ADMIRE = advanced model iterative reconstruction.
2
AJR:209, December 2017
the size of the patient and indicates radiation dose to the soft tissues within a scan volume. It is measured for scans of the thorax and abdomen, and for its calculation, the radiation-tracking software determines the effective patient diameter at the mid scan level. CT scans as part of PET and interventions were excluded from the study. In addition, no scans from children were included because the local pediatric university hospital was not a member of the local dose registry during the study period.
Fig. 1—Graph of volumetric CT dose index (CTDIvol) for thoracic CT performed with eight different scanners. Lower and upper ends of vertical bars show 25th and 75th percentile values. Horizontal markers represent 50th percentile.
10 8 6 4 2
Protocols, Sourced Data, and Statistical Analysis
For the dose analysis, we focused on the eight most commonly performed examinations at all hospitals, which were all indication-based and single-phase protocols. These examinations were identified to be unenhanced CT of the head (e.g., to exclude cerebral hemorrhage), paranasal sinuses (e.g., to exclude sinusitis), thorax (e.g., to exclude pulmonary nodules or pneumonia), pulmonary angiogram (e.g., to exclude pulmonary embolism), unenhanced renal colic protocol (e.g., to exclude urolithiasis), enhanced abdomen-pelvis in the portal venous phase (e.g., assessment of an acute abdomen), thorax-abdomen-pelvis (e.g., for tumor staging), and angiography of thorax-abdomen-pelvis (e.g., assessment of an aneurysm or dissection). Radiation-tracking software provided the values of the dose metrics (mean, median, and interquartile range). The graphical representation of data was performed using Microsoft Excel for Mac (version 15.33). The mean, median, and interquartile range from the current study were compared with the recently published data from the University of Toronto medical centers (published in 2015) [7], the University of California medical centers (published in 2015) [8], and the ACR dose index registry (data from semiannual report for adults between July and December 2015) [14]. We also compared our data with the national diagnostic reference levels of the country in which the local CT dose registry was located (Switzerland, published in 2010 [20],
12
CTDIvol (mGy)
Downloaded from www.ajronline.org by University of Auckland on 09/15/17 from IP address 130.216.158.78. Copyright ARRS. For personal use only; all rights reserved
CT Radiation Dose Tracking in Europe Versus North America
0
1
2
the United Kingdom, published in 2011 [21], and Portugal, published in 2014 [22]) (Table 2). The sourced data were compared with our study by matching them with indication and where available for single phase.
Results The total number of CT examinations during the 24-month period was 85,187. The aforementioned eight protocols analyzed in this study were responsible for approximately 51.3% of all of the tracked CT scans (n = 43,761). The eight most common CT protocols were CT of the head (n = 15,539; 35.5% of the analyzed studies), thorax (n = 7684; 17.6% of the analyzed studies), abdomen-pelvis (n = 6132; 14.0% of the analyzed studies), thoraxabdomen-pelvis (n = 4890; 11.1% of the analyzed studies), pulmonary angiogram (n = 3531; 8.1% of the analyzed studies), urolithiasis (n = 2517; 5.8% of the analyzed studies), paranasal sinus (n = 1875; 4.3% of the analyzed studies), and angiography of the thorax-abdomen-pelvis (n = 1593; 3.6% of the analyzed studies). The interquartile ranges of different dose metrics from our local dose registry are enu-
3
4 5 6 CT Scanner Number
7
8
merated in Table 3. For all of the protocols, there was broad variation among the eight CT scanners for the different radiation dose metrics (Table 3). For example, the smallest and largest values of the 75th percentile of the CTDIvol for CT thorax were 2.3 and 11 mGy, respectively. There was a fourfold difference in the median value of CTDIvol for CT of the thorax between scanner 3 (8.4 mGy) and scanner 8 (1.8 mGy), which are the same scanners at two different institutions (Fig. 1). In addition, scanner 8 showed substantially less variability than did scanner 3. A much smaller variation in the dose metrics was observed for head CT among the eight CT scanners. For CTDIvol, the smallest value of the 75th percentile was 34 mGy and the largest value was 64 mGy. The diagnostic reference levels from the ACR dose index registry [14] and California [8] were substantially higher than our data (Fig. 2). Compared with our study, the ACR dose index registry had higher dose values for 75th percentile CTDIvol values by 55% for head, 240% for thorax, 28% for abdomen-pelvis, 42% for thorax-abdomen-pelvis, 128%
TABLE 2: CT Scanner Specifications in the Sourced Data Sourced Data University of Toronto Medical Centers [7] University of California Medical Centers [8]
American College of Radiology dose index registry [14]
Data Collection Period
No. of Scanners
Scanner Specifications
2010–2013
2
64-MDCT
2013
34
One 8-MDCT, five 16-MDCT, rest 64-MDCT or higher; all with tube current modulation; 21 with iterative reconstruction
Report from July–December 2015 Not applicable Not applicable
Switzerland [20]
2007–2009
179
16- to 64-MDCT most common (ranged from single detector to 256-MDCT)
United Kingdom [21]
2010–2011
182
16- to 64-MDCT most common (ranged from 4to 320-MDCT)
Portugal [22]
2011–2012
41
16-MDCT most common
AJR:209, December 2017 3
4 1593
Angiography thorax, abdomen, pelvis
1.5 (1.2–5.7)
9.0 (3.3–16.1) 10.2 (4–14.8)
12.3 (7.6–17.7)
3.7 (2.7–11.3) 5.9 (3.5–13)
7.5 (5.5–13.9)
2.8 (2–10) 4.9 (3.2–11.1)
4.9 (3.8–7.9)
290 (111–532) 384 (220–613)
215 (167–473) 164 (132–297) 486 822 (354–778) (454–1140)
168 (141–388) 135 (43–214) 327 (234–708)
8.3 (4.0–12.6) 17.0 (6–26.5) 9.3 (5.9–13.8)
5.2 (4.0–10.6) 11.8 (6–26.3) 5.7 (4.3–11.3)
4.0 (3.5–8.7) 7.0 (2.6–17.0) 3.9 (2.8–10.5)
CTDIvol (mGy)
243 (142–338)
124 (84–362)
84 (63–273)
7.0 (3–13.1)
3.3 (2.2–9)
15
2.3 (1.6–7.3)
CTDIvol (mGy)
10.3 (7.7–13)
14.4 16.8 (9.7–16.5) (11.4–20.4)
11 (7.8–15)
826 1051 (638–950) (863–1303)
622 (445–761)
14.1 (9.6–17.6)
11.5 (7.7–13.6)
8.6 (6.2–11.2)
5.12 (3.4–13.8)
0.3 (0.09–0.5)
2.4 (1.9–6.2)
6.6 (5.1–8.4)
15.5 (11.1–18.1)
12.5 (8–15.3)
10.2 (6.6–14.5)
544 758 (445–666) (593–977)
13.5 405 (11.7–16.2) (326–531)
1.2 (0.9–5)
7 (2.8–12.9)
3.1 (2.4–10.4)
2.4 (1.9–9.5)
10.5 (8.8–12.5)
86 (67–341)
8.2 (6.5–11.2)
65 (51–242)
5.1 (2.3–11)
2.4 (1.8–8.4)
1.7 (1.7–2.8)
200 (85–468)
25th
75th
50th
SSDE (mGy) 25th
75th
1.8 (1.3–6.5)
50th
DLP (mGy·cm)
678 715 808 (666–1118) (694–1160) (728–1212)
25th
40.3 (34–64)
75th
38.4 (34–64)
50th
CTDIvol (mGy) 37.9 (34–64)
25th
20
Note—Data are mean (range) values from the eight CT scanners.
1875
2517
3531
4890
6132
7684
15,539
No. of Scans
Paranasal sinuses
Renal colic
Pulmonary angiography
Thorax, abdomen, pelvis
Abdomen-pelvis
Thorax
Head
Anatomic Region 2.9 (1.9–4)
4 (1.4–7.9) 11 (8.7–13.8)
1.9 (1.8–2.9) 1.5 (1.2–6.5) 8.2 (6.3–9.4)
5.5 (2–10) 5.3 (2.9–9.4) 0.6 (0.4–0.9) 13.4 (7.2–19.4)
2.2 (1.6–7.3) 3.1 (2.2–7.2) 0.5 (0.3–0.8) 7.7 (5.8–16.1)
13.8 16.6 (10.8–16.2) (13.5–20.6)
75th
50th
ED (mSv)
TABLE 3: Summary of the 25th, 50th, and 75th Percentiles From Eight CT Scanners of the Local Registry for Volume CT Dose Index (CTDIvol), Dose-Length Product (DLP), Size-Specific Dose Estimate (SSDE), and Effective Dose (ED)
Downloaded from www.ajronline.org by University of Auckland on 09/15/17 from IP address 130.216.158.78. Copyright ARRS. For personal use only; all rights reserved
Parakh et al.
80
0
0
Thorax
Present Study National United Kingdom Portugal
Head
25 Present Study National United Kingdom Portugal
AbdomenPelvis
Anatomic Region
ThoraxAbdomenPelvis
Pulmonary Angiogram
Toronto ACR-DIR California
60
40
20
Paranasal Sinus Anatomic Region
A
Toronto ACR-DIR California
10
5
Renal Colic
Fig. 2—Comparison of 75th percentile of volumetric CT dose index (CTDIvol) between dose registries from North America and published European diagnostic reference levels for different anatomic regions. A and B, Graphs show data for CT of head and paranasal sinuses (A) and for CT of thorax, abdomen-pelvis, thorax-abdomen-pelvis, pulmonary angiogram, and renal colic regions (B). ACR-DIR = American College of Radiology dose index registry.
B
AJR:209, December 2017
Downloaded from www.ajronline.org by University of Auckland on 09/15/17 from IP address 130.216.158.78. Copyright ARRS. For personal use only; all rights reserved
CT Radiation Dose Tracking in Europe Versus North America for pulmonary angiogram, 138% for renal colic, and 58% for paranasal sinuses studies. The comparison with the data from California for the 75th percentile CTDIvol revealed similar differences (55% for head, 240% for thorax, 21% for abdomen-pelvis, 28% for thorax-abdomen-pelvis, and 71% for paranasal sinus studies). The largest differences were observed between our data and all of the sourced published data for CT of the thorax (up to 240% difference from California and ACR dose index registry) and for renal colic CT (up to 138% difference from ACR dose index registry), followed by pulmonary angiography (up to 128% difference from ACR dose index registry). Notably, the recently published American and Canadian institutional diagnostic reference levels were also higher than the rest of the European diagnostic reference levels for all protocols except for CT of the head. Discussion Our succinct snapshot of CT radiation dose data for eight indication-based single-phase CT protocols collected within a local European dose registry over a 24-month period using radiation-tracking software offers a basis for benchmarking dose levels across institutions for a heuristic assessment of dose practice patterns with CT. The current data could also be used to judge the relevance of previously published national diagnostic reference levels and to assess intercontinental differences in diagnostic reference levels. The 25th, 50th, and 75th percentile values in the current study could be applied as an index of benchmarking for other institutions. Facilities with exposures consistently less than the diagnostic reference levels, which are defined as the 75th percentiles, could aim for the 25th percentiles, which are defined as the target values [16, 20]. Aiming for such target values triggers an incentive for continuous optimization in CT dose reduction while maintaining an acceptable image quality. Our results show that prior national European diagnostic reference levels might be outdated, likely because of the rapid evolution of CT technology, as well as the wider acceptance of dose-optimized protocols [21]. The most divergent value was noted for the thorax, suggesting that the use of the currently available dose-reduction techniques, especially iterative reconstruction, could be applied to achieve a greater degree of dose reduction. This finding was in agreement with a recent study by Thomas et al. [23], which reported the effect of the use of iterative re-
construction in clinical routine at a national level for frequently performed CT protocols with 20–30% dose efficiency. In contrast to this finding in the thorax, the data for abdomen-pelvis and thorax-abdomen-pelvis examinations in the current study showed results very similar to those of the sourced European diagnostic reference levels. A possible reason for this outcome is that the existing scanning parameters are adequate and necessary, even with the application of iterative reconstruction techniques for acceptable image quality without compromising diagnostic confidence, especially in an anatomic area with low inherent soft-tissue contrast. Our assumption was in accordance with a phantom study, which proved that reducing radiation doses in tandem with an iterative reconstruction algorithm affected the detection of low-contrast lesions [24]. A broad range of differences was observed between our data and the published data from the California medical centers, as well as the ACR data, for various protocols. One protocol that showed this discrepancy well was the renal colic examination, which is a well-accepted example of a lowdose protocol. In addition to renal colic examinations being performed for pathologic entities with high inherent contrast, in which a higher noise level can be accepted, young patients frequently undergo such examinations. According to a scientific report using the ACR dose index registry data from 2014 with 49,903 examinations, the mean institutional DLP for this type of study was 746 mGy·cm [15]. In contrast, the mean DLP for renal colic examinations was 55% lower in our study, which reaffirms that use of doseoptimized techniques is infrequent in the United States. Reasons for the differences in practice are complex and could be explained by a gross discrepancy in knowledge or willingness to use dose-optimized protocols. Consultation of radiologic institutions by an experienced dose-optimization team consisting of a radiologist, technologist, and medical physicist could promote the process of CT dose reduction while maintaining diagnostic confidence [25]. The applied strategies include teaching practical tips, as well as effective use of state-of-the-art dose reduction techniques. Another possible explanation is the advancement in CT technology in the past few years, because the North American data were collected between 2010 and 2013. Dose management systems are analytic solutions capable of gathering and using
big data for effective clinical decision making. Their primary advantage is the ability to process large amounts of robust information about numerous dose descriptors into dashboards, which can be used for quality assurance and to assess intrainstitutional interscanner variations. Dose benchmarking will play an increasing role in CT dose optimization in the future because the initiation of an optimization process should be done in real time and should be based on facts rather than intuition. When benchmarking with others, attention to the technical environment is imperative. It would be futile to compare a single-detector CT scanner with new-generation scanners equipped with various dosereduction techniques. In the future, we intend to use these data within the institutions of the registry to optimize practice. To improve the value of radiation-tracking software in the future, it is of great importance to add objective image quality as a parameter. Addressing only radiation dose parameters while neglecting image quality does not provide a holistic approach to optimizing CT protocols in accordance with the ALARA (as low as reasonably achievable) principle. Unfortunately, the quantitative image quality parameters that are currently available, such as image noise or contrastto-noise ratio, cannot be used for such a task because they do not account for diagnostic accuracy. A promising technique for such a task is the use of model observers. Comparison of useful objective image quality parameters for radiation-tracking software is a topic of ongoing research in our institution. There are a few limitations and hurdles to this study that must be mentioned. First, we did not check the scans and their indications because the data in the master server only include dose metrics and not CT images. However, during the establishment of the dose registry, we exerted great effort in protocol matching using the Radlex playbook to avoid nonstandardization of protocol nomenclature. Furthermore, the large number of CT scans in our study diminished the effect of such cases on the average dose metrics. Second, we did not compare the image quality of the images in our study with other national and international diagnostic reference levels, which also did not provide such information. Because the data were collected from routine clinical practice, it must be assumed that image quality reflects the standard practice of the institutions, interpreted routinely by the radiolo-
AJR:209, December 2017 5
Downloaded from www.ajronline.org by University of Auckland on 09/15/17 from IP address 130.216.158.78. Copyright ARRS. For personal use only; all rights reserved
Parakh et al. gist. Third, the patient-built, protocol, scanner-type, and reconstruction techniques of sourced data were not available. Because of the unavailability of these parameters and the gaps in time, a direct comparison of all the factors was not possible; therefore, summary values have been compared. In conclusion, our local CT dose registry showed lower dose values than did the existing older published European diagnostic reference levels and recent North American diagnostic reference levels. We also found substantial trans-Atlantic differences in dose practice, with European diagnostic reference levels having consistently lower values than North American diagnostic reference levels. Programmatic radiation dose monitoring has emerged as a valuable auditing tool to understand and manage CT practice. Visualization of differences in CT practice by dose benchmarking is a promising self-assessment contrivance to call attention to an institution’s failure to participate in dose reduction and could therefore initiate an appropriate optimization process. References
1. Journy N, Rehel JL, Ducou Le Pointe H, et al. Are the studies on cancer risk from CT scans biased by indication? Elements of answer from a large-scale cohort study in France. Br J Cancer 2015; 112:185–193 2. Redberg R, Smith-Bindman R. We are giving ourselves cancer. New York Times website. www.nytimes. com/2014/01/31/opinion/we-are-giving-ourselvescancer.html. Published January 30, 2014. Accessed December 20, 2015 3. Mathews JD, Forsythe AV, Brady Z, et al. Cancer risk in 680,000 people exposed to computed tomography scans in childhood or adolescence: data linkage study of 11 million Australians. BMJ 2013; 346:f2360 4. Joint Commission. Joint Commission announces new and revised diagnostic imaging standards. Joint Commission website. www.dosemonitor.com/files/ 4413/9035/3871/JCAHO_Radiation_Dose_release_ 2.pdf.. Published December 20, 2013. Accessed August 16, 2017 5. U.S. Food and Drug Administration. White paper:
6
initiative to reduce unnecessary radiation exposure from medical imaging. U.S. Food and Drug Administration website. www.fda.gov/downloads/ Radiation-EmittingProducts/RadiationSafety/ RadiationDoseReduction/UCM200087.pdf. February 2010. Accessed April 29, 2016 6. European Commission. Council directive 2013/59/ Euratom of 5 December 2013 laying down basic safety standards for protection against the dangers arising from exposure to ionising radiation, and repealing Directives 89/618/Euratom, 90/641/Euratom, 96/29/Euratom, 97/43/Euratom and 2003/122/ Euratom. European Commission website. ec.europa.eu/energy/sites/ener/files/documents/CELEX32013L0059-EN-TXT.pdf. Published 2014. Accessed August 4, 2017 7. MacGregor K, Li I, Dowdell T, Gray BG. Identifying institutional diagnostic reference levels for CT with radiation dose index monitoring software. Radiology 2015; 276:507–517 8. Smith-Bindman R, Moghadassi M, Wilson N, et al. Radiation doses in consecutive CT examinations from five University of California medical centers. Radiology 2015; 277:134–141 9. Boos J, Meineke A, Rubbert C, et al. Cloud-based CT dose monitoring using the DICOM-structured report: fully automated analysis in regard to national diagnostic reference levels. Rofo 2016; 188:288–294 10. [No authors listed.] Radiological protection and safety in medicine: a report of the International Commission on Radiological Protection. Ann ICRP 1996; 26:1–47 [Erratum in Ann ICRP 1997; 27:61] 11. [No authors listed.] ICRP Publication 105: radiation protection in medicine. Ann ICRP 2007; 37:1–63 12. American College of Radiology. CT accreditation program requirements. Reston, VA: American College of Radiology, 2013 13. Taylor S, Van Muylem A, Gevenois PA, Tack D. Intra-institution CT dose surveys in adults: what sample size for what precision? Electronic Presentation Online System website. posterng.netkey.at/ esr/viewing/index.php?module=viewing_poster& doi=10.1594/ecr2015/B-0785. Published 2015. Accessed August 4, 2017 14. American College of Radiology. National radiology dose index registry. American College of Radiology website. www.acr.org/~/media/ACR/Documents/
PDF/QualitySafety/NRDR/DIR/DIRSampleReport. pdf. Published 2017. Accessed August 4, 2017 15. Lukasiewicz A, Bhargavan-Chatfield M, Coombs L, et al. Radiation dose index of renal colic protocol CT studies in the United States: a report from the American College of Radiology National Radiology Data Registry. Radiology 2014; 271:445–451 16. Brink JA, Miller DL. U.S. national diagnostic reference levels: closing the gap. Radiology 2015; 277:3–6 17. Langlotz CP. RadLex: a new method for indexing online educational materials. RadioGraphics 2006; 26:1595–1597 18. Strauss KJ. Dose indices: everybody wants a number. Pediatr Radiol 2014; 44(suppl 3):450–459 19. [No authors listed.] The 2007 recommendations of the International Commission on Radiological Protection: ICRP publication 103. Ann ICRP 2007; 37:1–332 20. Treier R, Aroua A, Verdun FR, Samara E, Stuessi A, Trueb PR. Patient doses in CT examinations in Switzerland: implementation of national diagnostic reference levels. Radiat Prot Dosimetry 2010; 142:244–254 21. Shrimpton P, Hillier M, Meeson S, Golding S. Doses from computed tomography (CT) examinations in the UK: 2011 review. Public Health England website. www.gov.uk/government/publications/dosesfrom-computed-tomography-ct-examinations-inthe-uk. September 1, 2014. Accessed December 23, 2015 22. Santos J, Foley S, Paulo G, McEntee MF, Rainford L. The establishment of computed tomography diagnostic reference levels in Portugal. Radiat Prot Dosimetry 2014; 158:307–317 23. Thomas P, Hayton A, Beveridge T, Marks P, Wallace A. Evidence of dose saving in routine CT practice using iterative reconstruction derived from a national diagnostic reference level survey. Br J Radiol 2015; 88:20150380 24. Schindera ST, Odedra D, Raza SA, et al. Iterative reconstruction algorithm for CT: can radiation dose be decreased while low-contrast detectability is preserved? Radiology 2013; 269:511–518 25. Schindera ST, Treier R, von Allmen G, et al. An education and training programme for radiological institutes: impact on the reduction of the CT radiation dose. Eur Radiol 2011; 21:2039–2045
AJR:209, December 2017
