Experience With Patient Dosimetry and Quality Control Online for Diagnostic and Interventional Radiology Using DICOM Services

Transcription

1 Medical Physics and Informatics Review Vano et al. Online Quality Control for Diagnostic and Interventional Radiology Medical Physics and Informatics Review FOCUS ON: Eliseo Vano 1,2 Jose I. Ten 2,3 Jose M. Fernandez-Soto 1,2 Roberto M. Sanchez-Casanueva 1 Vano E, Ten JI, Fernandez-Soto JM, Sanchez- Casanueva RM Keywords: diagnostic reference levels, interventional, patient doses, quality control, radiation safety DOI: /AJR Received October 14, 2012; accepted without revision October 23, This study was supported by the Ministry of Science and Innovation (grant SAF ). 1 Medical Physics Service, San Carlos University Hospital, Instituto de Investigacion Sanitaria del Hospital Clínico San Carlos, Profesor Martin Lagos s.n., Madrid 28040, Spain. Address correspondence to E. Vano (eliseov@med.ucm.es). 2 Radiology Department, Medicine School, Complutense University, Madrid, Spain. 3 Diagnostic Radiology Service, San Carlos University Hospital, Madrid, Spain. AJR 2013; 200: X/13/ American Roentgen Ray Society Experience With Patient Dosimetry and Quality Control Online for Diagnostic and Interventional Radiology Using DICOM Services OBJECTIVE. This article describes the different automatic approaches used to collect and process patient dose values and other procedural data during diagnostic and interventional radiology and discusses their benefits for clinical practice and quality control online. Approaches for automatic processing of patient dose and other procedural data for computed radiography and for flat-panel detectors extracting information from DICOM headers or via DICOM services are described. The method to perform image retake analysis is also discussed. CONCLUSION. Automatic systems to manage patient doses and procedural data are feasible and will improve radiation safety and quality in radiology. The current level of technology makes such systems achievable at a reasonable cost and with great benefit to clinical practice. S ome years ago, patient dosimetry in diagnostic radiology was performed to calculate the mean or median values of different dosimetric quantities (e.g., entrance surface air kerma, air kerma-area product, and doselength product [DLP]) in a small sample of procedures as part of the clinical audit and to compare them with the local, national, or regional diagnostic reference levels, as recommended by the International Commission on Radiological Protection [1]. Today, with the introduction of digital radiology, it is possible to have automatic systems to collect and archive patient dose data individually, for all the imaging and interventional procedures, in addition to demographic, geometric, and other procedural parameters, as part of the DICOM header or through other DICOM services (e.g., modality performed procedure step [MPPS] or radiation dose structured reports) [2 4]. These automatic systems mean significant benefits for patient dosimetry and quality control (QC) because they offer the possibility of processing data from all the procedures (not from just one small sample), the automation of the data collection process, and the possibility of processing a wide range of data in addition to patient dose data (e.g., radiographic parameters, C-arm angulations, distances, and acquisition protocols) [3, 5 7]. The distribution of patient doses for the different imaging procedures in a hospital can now be analyzed in full, not with statistical descriptive parameters only. Several refined optimization actions can be launched, not only in the case of median or mean values that are higher or lower than the diagnostic reference levels, but also when other parameters are out of the normal range or when some individual procedures result in doses that are higher than two or three times the diagnostic reference levels. During the transition from conventional to digital radiology, the European Commission has supported several research programs to promote the radiation protection of patients (e.g., DIMOND and SENTINEL) [3, 8, 9]. Some pilot actions to launch automatic systems for patient dosimetry and QC online were developed at the San Carlos University Hospital in Madrid as part of these European programs [3, 10, 11]. Different approaches were used, depending on the availability and level of implementation of the DICOM standard, including extracting the technical information from the DICOM image headers, using the radiation dose structured reports (which contain accumulated dose over several irradiation events), analyzing the MPPS messages sent by the modalities to the radiology information system, and implementing optical character recognition techniques on saved screen images containing information about the delivered dose. The implementation of the DICOM radiation dose structured reports and the software used AJR:200, April

2 Vano et al. to process the information contained in the reports are likely to improve all these options in the coming years, especially for CT and interventional radiology procedures. The International Commission on Radiological Protection has published a document on patient dose management in digital radiology that addresses some of the most relevant issues [2]. Digital radiology has many advantages over film-screen radiography [12]. Nevertheless, both patient dose audit and QC are challenges, especially during transitions to new technologies [13]. The retake analysis in digital imaging is another relevant aspect that should be addressed in digital departments. Several initiatives are currently being run in many countries and within the radiology industry. Standardization committees and regulatory authorities have begun to show an interest in this issue, and the coming years should see significant advances [14, 15]. In its proposal for a Council Directive on Basic Safety Standards [16], the European Commission requires that the report of the dose delivered to patients be included in every single x-ray examination. This task would be unachievable without the help of automatic patient dose management systems. This article describes the different automatic approaches used to collect and process patient dose values and demographic, radiographic, geometric, and other procedural data during diagnostic and interventional radiology examinations and discusses the benefits for the clinical practice and for the QC online. Materials and Methods One of the first attempts to use digital technology in Europe for patient dose management was made in 1999, when the first digital generators and digital image systems were installed in several hospitals. It was then possible to transfer radiographic and geometric data to PCs and to calculate entrance surface air kerma from the x-ray output curves [10]. When the digital detectors were available and the DICOM headers started containing radiographic, geometric, and dose parameters, automatic systems were updated to collect information from the images and offer some graphical interfaces. The experience at the San Carlos University Hospital in Madrid allowed data collection from most of the digital modalities existing at the Hospital [5, 13]. The concept of alarm or trigger levels to launch corrective or optimization actions was already considered in the systems managing patient dose values. These alarms or trigger levels were set also for radiographic or other procedural parameters (e.g., inappropriate radiographic techniques, compression force in mammography, and high temperature in the flat-panel detector imaging device). A dedicated module called DOLIR (dose online for interventional radiology) was later launched for interventional radiology and cardiology, using the information sent by the x-ray systems as part of the patient dose reports [17]. Different Approaches for Automatic Collection and Processing of Patient Doses The different approaches to collect and process data for patient dosimetry and QC depend on the level of technology of the digital x-ray systems existing at the hospital and on the informatics infrastructure, including the hospital information system, radiology information system, and PACS and the connectivity among the different systems. One relevant aspect of the usefulness of systems for patient dose management is the staff dedicated to these tasks. Sometimes the infrastructure and the tools for automatic management exist, but the team to exploit the data, including experts in informatics, medical physics, and radiology, is missing. The Concept of Quality Control and Patient Dosimetry Online (QCONLINE) Initially, the x-ray systems were not fully digital and the informatics infrastructure at the hospital was not fully developed. Radiographic and geometric data were forwarded from the generators of the x-ray systems to a PC to calculate patient entrance surface air kerma and to archive and process individual dose data. Periodic validation and calibration of the output of the x-ray tubes were always necessary to validate dose values. If individual patient monitoring was requested, demographic data of patients needed to be introduced manually in the systems and linked to the entrance surface air kerma calculations. This Fig. 1 QCONLINE (Quality Control and Patient Dosimetry Online) I system. Screenshot shows alarms used for computed radiography modality. Image depicts procedures in three groups (green, yellow, or red light) depending on mean dose values of last 10 procedures. In this case, only pelvis imaging is likely to require possible corrective action because mean dose value is higher than local diagnostic reference level. was the initial experience for real-time auditing of patient doses at the San Carlos University Hospital. It was mainly applied for computed radiography (CR) [10, 18]. Average values of entrance surface air kerma and kerma-area product from the 10 most recently examined patients were calculated and compared with local diagnostic reference levels, generating warning messages if reference levels were exceeded and prompting corrective action if appropriate (Figs. 1 and 2). This first version of the system was called QCONLINE I [3, 19]. The main disadvantage of this system was the lack of physical link between the clinical images and the radiographic and patient dose data. Today, this approach still remains one of the few options to automatically audit patient dose values for CR in many hospitals. Other similar systems, but only limited to dose parameters, were later developed by the industry [20]. Systems Extracting Information From the DICOM Header (QCONLINE II) When systems with flat-panel detectors for imaging became available, most of the information to audit patient doses and other procedural data appeared at the DICOM header of the images (Fig. 3). This allowed us to capture the data, transfer them to a database, and process them in real time. This second step in the software development at the San Carlos University Hospital was called QCONLINE II. A problem still remained: the lack of standardization in many of the DICOM header tags containing useful information. Different vendors used private tags and different quantities and units for dose-related parameters, changing them from one version of the DICOM conformance statement to another after upgrades. Staff effort was required to maintain these systems in operation. In addition, connectivity from 784 AJR:200, April 2013

3 Online Quality Control for Diagnostic and Interventional Radiology Fig. 2 QCONLINE (Quality Control and Patient Dosimetry Online) system. Screenshot shows radiographic parameters and patient dose data (entrance air kerma and kerma-area product) for individual procedures (lumbar spine lateral projections). Minimum, maximum, mean, and SD values are shown to left of figure. the different modalities or from the PACS to a central workstation that processed all the dosimetric and procedural data represented additional difficulties related to the involvement of several hospital departments as well as of technical personnel from the vendors. The necessary confidentiality and protection of personal patient data represented another challenge with such systems. Fig. 3 DICOM header. Example shows some of most relevant tags captured and processed by QCONLINE (Quality Control and Patient Dosimetry Online) II system for chest images. Results The use of automatic patient dose collection and analysis at the San Carlos University Hospital started in 1999 when the first fully digital radiology department was launched in a public hospital in Spain. The feedback received during the European research programs DIMOND and SENTINEL allowed enriching the experience in such fields as QC online, training possibilities for residents and radiographers, and optimization of clinical practice [8, 9]. The main advantages of the QCONLINE system after the introduction of the flat-panel technology and the availability of many data in the DICOM header include a physical link between the clinical images and the radiographic and dose data, easy auditing of patient dose values and radiographic data, easy auditing of image quality related to these dose and radiographic data (Fig. 4), the possibility of auditing the full imaging procedure (e.g., timing, retakes, patient dose, and image quality), the possibility of automatic routing of images from a PACS or from the modality to the QC station, the possibility of comparing patient dose values (mean values of samples) with local and international diagnostic reference levels (Fig. 5), inclusion of demographic data if appropriate and if confidentiality local regulations permit, and the setting of different alarms to notify of x-rays system or procedurals problems (e.g., the use of inappropriate tube voltage values, low compression force in mammography, and mistakes in the use of automatic exposure tools). The experience at the San Carlos University Hospital was especially positive for mammography [11] because it allowed auditing of entrance surface air kerma and glandular doses from over 25,000 images during the initial step. The level of mechanical compression (included as one of the tags in the DICOM header) allowed the system to suggest corrective actions when operators failed to apply the appropriate compression. As for projection radiography, a database with 205,000 patient dose values was used to compute the evolution of patient doses over time and to introduce optimization actions during the transition from conventional file-screen imaging to digital radiology [13]. The local software for the QCONLINE system using different upgrades has been running for patient dosimetry auditing since 2000 [10, 13]. All the digital modalities of the Radiology Department sent their images to a PACS connected to a workstation of the Medical Physics Service and extracted the information contained in the DICOM header. Four interventional cardiology laboratories (working with an independent PACS) were also connected to this workstation but could only rely on the very limited information contained in the DICOM header of interventional systems at that period. At the workstation, relevant parameters (depending on the information contained in the DICOM header of each modality) were analyzed, and their current values for a given imaging procedure were compared with values considered suitable (i.e., diagnostic reference levels, in the case of dosimetric data). A warning message appeared on the workstation screen whenever parameters were out of range, and corrective action could then be undertaken if required. Images received from flat panels, CR systems, and interventional laboratories (only one frame per series) were, by default, presented on the QC workstation screen for basic image quality inspection, which allowed monitoring in real time. Images giving rise to a warning were stored on AJR:200, April

4 Vano et al. the workstation hard disk along with the alarm source, including it as another attribute, in a private field at its DICOM header. These images were later used in training programs for residents and radiographers. The results of the DICOM header analysis of each modality with the most relevant tags selected for the QCONLINE system are as follows: First, CR images contain information about the examination, plate identification and number of uses, exposure level (parameter defined by the manufacturer), and processing parameters. The number of exposures in the plate and the exposure level were the parameters used for the auditing. No information was provided about radiographic parameters that could be used to calculate patient doses. The computer application allows online comparison of the mean exposure level for a recent sample with the local reference value, permits auditing of this dose-related parameter, and suggests corrective actions if appropriate. Second, mammography digital radiography images contain all the technique parameters (tube voltage, tube current, focus size, distance focus-detector, anode and filter selection, manual or automatic exposure mode, compression force, compressor position, patient thickness, and detector temperature) and a calculation of entrance surface air kerma and glandular dose. Most of these parameters were audited, and dose calculations were periodically verified by using the results of the local QC program. Third, chest and trauma digital radiography images also contain all the technique parameters and a calculation of entrance surface air kerma and kerma-area product, which are audited and periodically verified by the Medical Physics service. Fourth, patient dose parameters from interventional radiology and cardiology modalities were measured with a kerma-area product meter. Patient dose values were included in the DICOM header (depending on the manufacturer). The number of frames per series, the number of series per procedure, tube voltage, tube current, pulse time, distances, and C-arm angulations are other useful information to be collected for auditing purposes. Finally, the DICOM headers of the first CT units provided information on a few parameters, such as and tube voltage and tube current. The volume CT dose index and DLP were available only via the MPPS service. Only recently have they been available via the radiation dose structured reports. Fig. 5 Screenshot of alarm setting for QCONLINE (Quality Control and Patient Dosimetry Online) II system. European reference values (ERV) and local reference values (LRV) are shown for different modalities together with actual mean values and range of values in audited sample. Fig. 4 Example of images as presented by QCONLINE (Quality Control and Patient Dosimetry Online) II system. Relevant data from DICOM header are shown with image allowing audit of image quality together with radiographic technique, proper use of automatic exposure control, and entrance surface air kerma value. Dose Index in Computed Radiography With nondigital generators (e.g., mobile systems for pediatrics and neonatal units), we opted to perform an experimental measurement of the curves by relating the dose index (a parameter related to the light emitted by the CR plates) with the entrance surface air kerma, for different patient thicknesses and tube voltages. The dose index is included by most of the new CR systems at the DICOM header of the images and allows estimating entrance surface air kerma values for individual patients [21]. Retake Analysis The QCONLINE system also makes it possible to analyze repeated images in digital imaging [22]. This analysis formed a classic item in any quality assurance program in all the radiology departments using film-screen technology but was, unfortunately, difficult to adapt to digital technology. We chose to use the DICOM header information to implement a method aimed at detecting potential retakes in digital imaging [23]. In most cases, neither the CR workstation nor the PACS itself is designed to support reject analysis. We also used QCONLINE to identify images bearing the same patient identification number, same modality, description, projection, date, cassette orientation, and image comments. In many cases, the most frequent difficulty is having the wrong image identification information. The system automatically detects retakes and deficiencies in the department performance, such as wrong identification, positioning errors, wrong radiographic technique, bad image processing, equipment malfunctions, artifacts, and so forth. In addition, automatically collected retake images can be used for staff training. Transportability of the System to Other Hospitals The evolution of QCONLINE (from version II) includes a module that is able to analyze, collect, and process relevant information transferred via MPPS. The transportability of the system was tested in two remote hospitals 786 AJR:200, April 2013

5 Online Quality Control for Diagnostic and Interventional Radiology over several months. The MPPS module proved to be an asset in the connectivity and portability of the system to other hospitals [24]. The system is currently being upgraded to process radiation dose structured reports already available for some of the new CT and interventional systems. Fig. 6 Screenshot of DOLIR (Dose Online for Interventional Radiology) I module to audit interventional radiology procedures. Several dosimetric and operational quantities can be followed in graphical display (arrow). Two colors in vertical bars indicate contribution of fluoroscopy (green) and cine (blue) to kerma-area product. Horizontal line indicates median value in this catheterization laboratory. Interventional Radiology Before radiation dose structured reports become available for most interventional systems, a new module developed for interventional radiology applies the method of the QCON- LINE as an interim solution. Two approaches have been developed: using the full information contained in the patient dose reports (with details of all the archived series fluoroscopy, cine, and digital subtraction angiography), and using the MPPS DICOM service that typically allows one to manage only the kerma-area product and the cumulative air kerma at the patient entrance reference point. On the one hand, the first approach allows managing more information and provides better capacity to audit the full procedure and to help with the optimization, but it requires frequent updates whenever changes occur in dose-reporting formats. On the other hand, the MPPS approach is easier to implement, but provides less information for individual procedures. The DOLIR system is based on homemade software that nevertheless uses most of the performances of QCONLINE. It allows automatic archiving and analysis of the major study parameters and patient doses for fluoroscopy-guided procedures performed in cardiology and interventional radiology systems. The x-ray systems need to have the capability to export the technical parameters of the study and the patient dose values at the end of the procedure and via . The software develops queries, retrieves all the study reports sent by the imaging modality from a mail server, and stores them on a Microsoft SQL Server database. It is connected to seven interventional systems processing all the produced patient dose reports. In the case of some technical parameters and patient doses, alarms are added to receive malfunction alerts so as to immediately take appropriate corrective actions [17]. The DOLIR module allows generating alarms when high patient dose values are produced in catheterization laboratories and conducting corrective actions or clinical follow-up for high skin doses, if appropriate. The functionalities of this module (customized for Philips Healthcare Allura systems in our hospital) are discussed here. First, several interventional catheterization laboratories can be simultaneously audited from a central workstation. In our case, seven laboratories are connected (six in the San Carlos University Hospital and one located in a hospital 25 km distant from the San Carlos University Hospital). Second, audits can be made for kerma-area product, cumulative air kerma, fluoroscopy time, total number of series per procedure, total number of frames, and so forth. All these parameters allow a complete audit of the different procedures (Fig. 6). Third, kerma-area product for fluoroscopy (green bars in Fig. 6) and kerma-area product for cine (blue bars in Fig. 6) are shown in the graph. Fourth, the third quartile (local diagnostic reference levels) and median values are shown as horizontal lines in different colors. For cumulative air kerma, a trigger (alarm) level has been established to easily detect the procedures that need to be audited individually for potential skin injuries and clinical follow-up. Fifth, a comparison between the different laboratories performing similar procedures can Fig. 7 Screenshot of full patient dose report in DOLIR (Dose Online for Interventional Radiology) I module, including skin dose distribution (for cardiac procedures) in percentages of 2 Gy (blue bar at bottom) and values of cumulative scatter dose and maximum scatter dose rate at C-arm (arrows). easily be made. Sixth, it is possible to visualize in a single graph the results of procedures. Seventh, if some individual high doses are detected in the graphical interface, immediate access to the full patient dose report is available along with the detail of all the archived series of images. Eighth, patient reports include details of the skin dose distribution in 10 skin zones, with percentages (i.e., 100% or 2 Gy) for cardiology procedures (Fig. 7). Ninth, in some laboratories, thanks to the addition of a wireless electronic system for dosimetry (DoseAware, Philips Healthcare) [25], the cumulative scatter dose per procedure at the C-arm and the maximum scatter dose rate per procedure at the C-arm can also be included (in this case, still manually) in the DOLIR system (Fig. 7), which supplies information on the level of occupational risk for the different procedures. Finally, total or partial data export (filtered per procedure, per laboratory, AJR:200, April

6 Vano et al. DLP for 8500 Procedures (mgy cm) Cerebral (unenhanced) Coronary arteries Neck to pelvis (contrast enhanced) Cerebral (contrast enhanced) 819 Abdomen (contrast enhanced) Abdomen, pelvis (contrast enhanced) Aorta Chest, abdomen, pelvis (contrast enhanced) or per date) to a Microsoft Excel spreadsheet is possible. DOLIR version II is based on a MPPS DICOM service and provides easy portability to other hospitals, and no endless updates are necessary whenever changes occur in dose-reporting formats. However, this MPPS version does not allow access to the full patient dose reports. This will be possible when the radiation dose structured report is implemented in all the interventional systems. Advances for CT Using the DICOM Header and the Radiation Dose Structured Reports The module for CT systems is able to extract the information contained in the DICOM header but also that contained in the radiation dose structured report when available. A pilot experiment including CT data exploitation has been conducted with three 64-MDCT units (one Brilliance [Philips Healthcare] and two Optima CT660 [GE Healthcare]). With the Brilliance unit, the dose information was extracted from the DICOM header of the localizer image with the Philips Healthcare proprietary Standard Extended Service-Object Pair Exposure Dose Sequence information. With the Optima systems, once the clinical procedure is completed, all patient dose information is extracted from the radiation dose structured report generated by the modality. Alternatively, it is also possible to analyze DICOM Abdomen (unenhanced) Lumbosacral spine (unenhanced) Chest, abdomen (contrast enhanced) Mean European Values: Chest 8.3 msv 593 mgy cm Abdomen 12.5 msv 836 mgy cm Neck (contrast enhanced) 443 Chest (contrast enhanced) Facial and sinus (unenhanced) Chest (unenhanced) 309 Chest high definition Fig. 8 Example of results from CT module. Sample of 8500 procedures was analyzed using mean values of dose-length product (DLP). Some effective doses are shown applying conversion factors of European Commission guidelines [26]. MPPS messaging when available. When the QCONLINE system receives images, it automatically generates a structured file containing the complete contents of the DICOM header of each study, as well as the information extracted from the radiation dose structured report. When the reception of the study is over, the structured file is processed and the software performs calculations to store the desired information in the database. The system includes a tool to set levels of alert to detect anomalous situations and automatically communicates alarms by via intranet. Finally, this module is capable of estimating the effective dose derived from the procedure according to the generic estimation method proposed by the European Commission [26]. The full process is automatic and was tested over a period of 6 months on 11,500 procedures with three 64-MDCT units. Data were captured with a relatively small number of errors (< 0.1%). Examples of the results are shown in Figures 8 and 9. As in previous versions, the system allows not only extracting, archiving, and processing in a relational database the parameters related to patient dose contained in the DICOM headers and radiation dose structured reports, but also managing all the additional information contained in the headers for a QC online of the operational procedures. Information is gathered at four levels: patient (e.g., demographic data), procedure (e.g., x-ray system and name of the procedure classified according to the national Society of Diagnostic Radiology criteria, www. seram.es), series (e.g., number of images per series), and images (e.g., radiographic and geometry parameters). Once all the information has been entered, several trigger conditions can be implemented to generate alarms and to launch corrective actions when, for instance, individual dose values per examination are three times higher than the diagnostic reference level, or median values of the last 30 procedures are higher than the diagnostic reference level, or acquisition rotation time and tube voltage fall out of the established protocols, or patients undergo several CT examinations in a short period. The system allows the exporting of data for statistical process. Initially limited to the examinations performed at one hospital, the personal patient dose record can now be supplemented by connections with other hospitals and outpatient centers using the same system. Before issuing a formal patient dose report, the medical physicists verify and correct all patient dose data. The reports present and compare mean and median DLP values for the most common CT procedures with the existing references available to decide whether optimization actions are required to refine some clinical protocols. Discussion When dealing with systems for automatic transfer and processing of patient dose parameters in digital imaging, problems arise that should be considered and corrected in the future. The most easily detected problems (e.g., different dose units for the values in a specific DICOM tag, or some abnormal values that are very high or very low because of the errors in the transfer process of data) can easily be corrected if only a few x-ray systems are involved in the sending of information to a central database, but a strict and frequent QC still is necessary. Other problems could, however, be classified as systematic mistakes when coming from operators who were unable to introduce the correct procedure identification (e.g., a full-body CT that is identified as a chest or abdominal CT or a coronary angiography, for procedures also containing some therapeutic procedure identified as an angioplasty). Patient identification is another frequent mistake in some centers. These mistakes, if not detected and corrected in an early stage, may contaminate the full database, and 788 AJR:200, April 2013

7 Online Quality Control for Diagnostic and Interventional Radiology Frequency (No. of Examinations) the results obtained could deviate greatly from the real values. In many centers, operators fail to apply and validate calibration factors to the dose parameters included by the x-ray systems in the DICOM header or in the radiation dose structured report, which becomes another obstacle to overcome. Factors to Consider to Turn Automatic Patient Dose Systems Into an Attractive Tool The software for these automatic tools is, of course, the key part of the system, but many other factors, in part linked to the software but also related to the real expertise in the field of optimization of medical imaging, are necessary to offer a friendly instrument to improve radiation safety and quality management in the clinical work. The following important factors could help make the use of automatic patient dose and QC management systems more attractive: the possibility for easy data processing and exploitation (including statistical analysis and graphs); the possibility of automatically detecting, correcting, or removing data introduced incorrectly in the database (e.g., type of procedure or demographic data); attractive and friendly graphical interfaces (displays); the possibility of setting different alarms (tendencies in median values and very high doses for individual procedures) and automatic alerting using or telephone messages; the possibility of producing standard periodic reports for clinical audit and procedures improvement; analysis of repeated images and retakes; automatic preselection n = 1298 Abdominal Examinations mgy cm msv Median rd Quartile 978 Mean DLP Greater Than Two Times Median 15.6% DLP Greater Than Three Times Median 5.6% European Mean Value for CT of Abdomen, 10.6 msv >50 Effective Dose (msv) Fig. 9 Example of results from CT module. Sample of 1298 abdomen examinations was analyzed with values of dose-length product (DLP) and estimated effective doses using conversion factors of European Commission guidelines [26]. Percentage of procedures with DLP higher than two and three times median values is shown in inset. of procedures and patients requiring clinical follow-up for high skin doses in fluoroscopyguided procedures; periodic evaluation of median dose values to be compared with local or national diagnostic reference levels and collective doses for the different modalities and examinations, which would allow protocols to be systematically reviewed and old imaging systems to be renewed; analysis of any other parameters (different from radiation doses) included in the DICOM header or in the radiation dose structured report for QC online (e.g., use of collimation, proper filtering, focus-skin distance, and patient-image detector distance); and finally, contribution to a national or international repository or registry of patient doses, as suggested by the recent International Atomic Energy Agency statement [27]. Conclusion Automatic systems to transfer, archive, and process radiation dose and procedural data contained in the DICOM header and in the radiation dose structured report on an individual basis are feasible and necessary to improve radiation safety and quality in radiology. The current level of technology allows doing so at a reasonable cost and with a great benefit for the clinical practice. Diagnostic reference levels will be effortlessly reviewed with such systems, and the current international agreement on patient dose tracking will be easily implemented in the future. Nevertheless, better standardization and connectivity, along with more support from the medical physicists in the imaging departments, are still required. References 1. International Commission on Radiological Protection. The 2007 Recommendations of the International Commission on Radiological Protection: ICRP publication 103. Ann ICRP 2007; 37: International Commission on Radiological Protection. Managing patient dose in digital radiology: a report of the International Commission on Radiological Protection. Ann ICRP 2004; 34: Vano E, Fernandez JM, Ten JI, Gonzalez L, Guibelalde E, Prieto C. Patient dosimetry and image quality in digital radiology from online audit of the X-ray system. Radiat Prot Dosimetry 2005; 117: Ng CK, Sun Z. Development of an online automatic diagnostic reference levels management system for digital radiography: a pilot experience. Comput Methods Programs Biomed 2011; 103: Vano E, Fernandez Soto J. Patient dose management in digital radiography. Biomed Imaging Interv J 2007; 3:e26 6. Tsalafoutas IA, Metallidis SI. A method for calculating the dose length product from CT DICOM images. Br J Radiol 2011; 84: Wang S, Pavlicek W, Roberts CC, et al. An automated DICOM database capable of arbitrary data mining (including radiation dose indicators) for quality monitoring. J Digit Imaging 2011; 24: Faulkner K. The DIMOND project and its impact on radiation protection. Radiat Prot Dosimetry 2005; 117: Faulkner K, Malone J, Vano E, et al. The SENTINEL project. Radiat Prot Dosimetry 2008; 129: Vano E, Fernandez JM, Ten JI, Guibelalde E, Gonzalez L, Pedrosa CS. Real-time measurement and audit of radiation dose to patients undergoing computed radiography. Radiology 2002; 225: Chevalier M, Morán P, Ten JI, Fernández Soto JM, Cepeda T, Vano E. Patient dose in digital mammography. Med Phys 2004; 31: Vano E, Faulkner K, Orton CG. Point/counterpoint: a major advantage of digital imaging for general radiography is the potential for reduced patient dose so film/screen systems should be phased out as unnecessarily hazardous. Med Phys 2006; 33: Vañó E, Fernández JM, Ten JI, et al. Transition from screen-film to digital radiography: evolution of patient radiation doses at projection radiography. Radiology 2007; 243: Shih G, Lu ZF, Zabih R, et al. Automated framework for digital radiation dose index reporting from CT dose reports. AJR 2011; 197: Sodickson A, Warden GI, Farkas CE, et al. Exposing exposure: automated anatomy-specific AJR:200, April

8 Vano et al. CT radiation exposure extraction for quality assurance and radiation monitoring. Radiology 2012; 264: European Commission. Proposal for a Council Directive laying down basic safety standards for protection against the dangers arising from exposure to ionising radiation. European Commission website. ec.europa.eu/energy/nuclear/radiation_ protection/doc/2012_com_242.pdf. Published May 30, Accessed October 10, Ten JI, Fernandez JM, Vañó E. Automatic management system for dose parameters in interventional radiology and cardiology. Radiat Prot Dosimetry 2011; 147: Fernandez JM, Ordiales JM, Guibelalde E, Prieto C, Vano E. Physical image quality comparison of four types of digital detector for chest radiology. Radiat Prot Dosimetry 2008; 129: Morán P, Chevalier M, Ten JI, Fernández Soto FOR YOUR INFORMATION ARRS 2013 Breast Imaging Symposium JW Marriott Phoenix Desert Ridge Resort and Spa Phoenix, AZ September 18 21, 2013 Registration opens May 15 at JM, Vañó E. A survey of patient dose and clinical factors in a full-field digital mammography system. Radiat Prot Dosimetry 2005; 114: Li S, O Dea TJ, Geise RA. Establishing a quality control program for an automated dosimetry system. Med Phys 1999; 26: Vano E, Martinez D, Fernandez JM, et al. Paediatric entrance doses from exposure index in computed radiography. Phys Med Biol 2008; 53: Jones AK, Polman R, Willis CE, Shepard SJ. One year s results from a server-based system for performing reject analysis and exposure analysis in computed radiography. J Digit Imaging 2011; 24: Prieto C, Vano E, Ten JI, et al. Image retake analysis in digital radiography using DICOM header information. J Digit Imaging 2009; 22: Vano E, Ten JI, Fernandez JM, Prieto C, Ordiales JM, Martinez D. Quality control and patient dosimetry in digital radiology: on line system new features and transportability. Radiat Prot Dosimetry 2008; 129: Sanchez R, Vano E, Fernandez JM, Gallego JJ. Staff radiation doses in a real-time display inside the angiography room. Cardiovasc Intervent Radiol 2010; 33: European Commission. European guidance on estimating population doses from medical x-ray procedures: Radiation Protection Report 154. European Commission website. ec.europa.eu/energy/nuclear/ radiation_protection/doc/publication/154.zip. Published Accessed October 10, International Atomic Energy Agency. Joint position statement on the IAEA patient radiation exposure tracking. International Atomic Energy Agency website. rpop.iaea.org/rpop/rpop/content/news/ position-statement-iaea-exposure-tracking.htm. Published Accessed October 10, AJR:200, April 2013