What should radiology residents know about PACS? Poster No.:
P. M. A. van Ooijen, M. Oudkerk; Groningen/NL
PACS, Radiology residents, DICOM
Any information contained in this pdf file is automatically generated from digital material submitted to EPOS by third parties in the form of scientific presentations. References to any names, marks, products, or services of third parties or hypertext links to thirdparty sites or information are provided solely as a convenience to you and do not in any way constitute or imply ECR's endorsement, sponsorship or recommendation of the third party, information, product or service. ECR is not responsible for the content of these pages and does not make any representations regarding the content or accuracy of material in this file. As per copyright regulations, any unauthorised use of the material or parts thereof as well as commercial reproduction or multiple distribution by any traditional or electronically based reproduction/publication method ist strictly prohibited. You agree to defend, indemnify, and hold ECR harmless from and against any and all claims, damages, costs, and expenses, including attorneys' fees, arising from or related to your use of these pages. Please note: Links to movies, ppt slideshows and any other multimedia files are not available in the pdf version of presentations. www.myESR.org
Page 1 of 9
Learning objectives To provide an overview of the general components of a PACS to give residents in radiology an insight into the complexity of the system and to help them better judge problems or questions relating to PACS.
Background The digital environment of the radiologist contains two main entities: the Radiological Information System (RIS) and the Picture Archiving and Communications System (PACS). The RIS contains mainly textual information on the patient such as, examination schedules, worklists, radiological reports, patient demographics, etc. while the PACS contains the image data acquired by the different imaging modalities. The implementation of PACS and RIS is different in every hospital, not only because of the vendor specific features and front-end software, but also because of more fundamental choices such as a RIS guided or a PACS guided workflow. In the early days, connecting several entities (workstations, PACS, imaging modalities, etc) together in a network posed serious problems. Most of these problems have been tackled through the introduction of a number of standards into the medical field. The standardization of the image data is realized by the world standard DICOM (digital image communication in medicine), which is widely adapted by the medical hardware and software industry. For textual information the Health Level 7 (HL-7) standard is adapted world-wide. Most radiology residents are confronted with RIS and PACS nowadays for reading their radiological cases. This directly implies that they not only have the benefits of working at a highly advanced workstation in a digital radiology environment, but they will also encounter the problems and malfunctions that can occur. In both cases it is useful to have an understanding of the technical and organizational background of the digital PACS environment. This reader provides an insight in the structure behind the workstation and aims to equip the radiology resident with a general understanding of the PACS environment. The PACS environment generally consists of a number of different components that will be covered in more detail later in this reader. The two main components are the RIS and the PACS itself. The PACS handles the storage and distribution of the image data. A very important component that is often, but not necessarily, heavily connected to the PACS are the viewing workstations for the radiologist. The RIS handles the workflow
Page 2 of 9
within the radiology department from planning a study to the final report and also handles the storage and distribution of the radiology reports. The RIS and PACS together provide the radiologist with the information necessary to interpret an examination and to deliver the report of that examination to the requesting clinician. Other components in the PACS environment that can either be integrated fully into a certain PACS or RIS or be separately installed are for instance a web server, a broker or a speech recognition system. Important to notice is that, when these components are obtained separately (and thus, in most cases, are provided and installed by a different vendor than the RIS or PACS) the integration of the different systems into one is a very important and often equally difficult issue.
Imaging findings OR Procedure details The different general components of a PACS are shown in figure 1. The communication between the different components is possible through a network. In most current implementations this network is setup as a star configured, switched network, with network speeds of 1 Gbit/s. The main PACS consists of one or more servers that run the distribution of image related data and the database service of the PACS. The PACS stores the images that it receives from imaging modalities and distributes them towards the on-line main storage and the near-line backup storage. Upon request of a viewing workstation, the image data is transmitted to the viewing workstation. In some cases, the data will also be transmitted to a webserver providing images to departments outside radiology. Different output devices are still available to distribute the image data to other entities such as patients or other health institutions for example by publishing CDs or by printing films on a dry laser imager. PACS Core The core of the PACS consists of a system that receives incoming DICOM images, distributes DICOM images to different locations, and sends DICOM images to different systems upon request. In most cases this PACS core consists of multiple servers covering the different tasks. Some of the tasks can be performed on multiple servers to ensure availability by having redundancy. Furthermore, the use of multiple servers for the same task can also increase the capacity of the system by having them all fully operational and balancing the load between them. The PACS core also contains
Page 3 of 9
the database that holds all the information on what is stored in the entire PACS. This database is a crucial part of the system because the inability to access the database will completely block the access to the image data. On-line Main Storage The on-line main storage is directly accessible and based on harddisk technology (figure) providing a very fast access to the image data. The configuration of the on-line main storage can vary based on local preferences. Examples are RAID, DAS, SAN and NAS. RAID technology stands for a Redundant Array of Inexpensive Disks (or sometimes also known as a Redundant Array of Independent Disks). This basically consists of a large number of harddisks connected to and controlled by a dedicated hardware controller with a certain level of redundancy in the setup. This level of redundancy is indicated by the number placed after the RAID acronym. Available levels of redundancy are RAID-0, RAID-1, RAID-3, RAID-4, RAID-5, and RAID-6. The most common used redundancy level in PACS is the RAID-5. A RAID system can either be internal into the host computer or external. The size of an internal RAID is limited by the size of the harddisks used and the number of available slots in the host computer and the RAID level. In general, the higher the RAID level, the higher the redundancy and thus the lower the remaining storage space. An external RAID is typically connected to the host computer using SCSI or fibre channel and can be expanded easily to a multi Terabyte system depending on the ability of the external controllers used. The internal RAID is an example of a Direct Attached Storage (DAS) which is internal into the host computer. However, a DAS can also be configured as a simple JBOD (Just a Bunch Of Disks) without any redundancy or striping. As already mentioned, the main problem of a DAS is scalability since it is limited to the available slots in the host computer multiplied by the maximum size harddisks available. In a Storage Area Network (or SAN) remote storage devices can be attached to a local computer system in such a way that the computer system will treat the SAN as being a locally attached storage device. With a Network Attached Storage (or NAS) the operating system knows that the storage is remote and file-based protocols such as NFS are used. Access to the data is performed on the file level. The choice of storage technology is mainly based on cost and preference within the local IT infrastructure. A typical main storage system for a PACS will hold many TeraBytes (TB) and may in the future gradually shift to PetaBytes (PB) (1 PetaByte = 1000 TB) due to an ever growing increase in the data production of the different radiological imaging modalities.
Page 4 of 9
Near-line Backup Storage The term near-line storage is used for storage on portable media in an automated device (jukebox). The media is not available directly, but can be loaded into a drive automatically by a robotic arm, without manual user intervention. Near-line backup storage can be on several different available portable media such as DVD, UDO, magnetic tape, etc. The near-line backup storage is mostly used as a long term archive. However, problems can occur with the readability of the media caused by aging of the media or simply by the fact that media is discontinued and drives that can read it are no longer available. Therefore, any near-line backup storage holding portable media should be migrated to the latest technology every 5-10 years to keep up to date and to make sure that no data is lost due to the deterioration of the media. When the portable media is removed from the jukebox and stored on a shelve, the term off-line storage is used. Broker/DMWL Server The broker handles the communication between the Radiological Information System (RIS) and the PACS. Health Level 7 (HL7) messages are distributed by the RIS based on certain rules. For example, messages can be generated to inform the PACS of new studies planned or of changes in patient demographics. The imaging modalities receive their worklists from the PACS broker through the DICOM Modality Worklist Server (DMWL Server). The worklist information is transferred from the HIS/RIS to the PACS broker through the CAI-TDM. Based on this worklist data, the patient information is added to the DICOM images acquired at the scanner. After the images have been acquired and sent to the PACS system, several mechanisms are provided to make sure the data is correct and consistent. Studies produced by modalities that are not supporting the DICOM Modality Worklist Client, can automatically be checked on consistency using an automated Quality Control (QC) tool. Such a tool verifies incoming image data against the data in the DICOM Modality Worklist Server and maintains a Verification status for each study. This Verification status can be used to produce difference lists between the PACS and the RIS, the unperformed and unscheduled lists. Manual Quality Control can be applied on studies as well to, for example, correct any data not supplied automatically by the DICOM Modality Worklist. Furthermore, the PACS broker also receives update and change messages from the RIS which are subsequently processed in the PACS database. This ensures consistency even if patient data is changed in the RIS after acquisition. Speech Server
Page 5 of 9
The speech system in most cases consists of two servers. One server will be used to perform the actual speech recognition and the other server will perform the other tasks. In principle, most speech systems can still be used as dictation systems where the radiologist dictates his or her report and a medical typist types in the report based on the voice recording. However, current technology has enabled speech recognition with very high levels of recognition. The speech recognition can be done in batch or in direct mode. In batch mode the reports will be dictated by the radiologists and transformed to text files by the recognition software in the background. After finishing multiple dictations, the radiologist will proceed to the list of dictated reports and will correct and sign the batch. In the direct mode the recognition is done during dictation and the text is instantly shown on the screen allowing interactive correction and adaptation by the radiologist. As soon as the radiologist finishes the dictation of the report he/she can sign it off directly as being the final report. Residents can, instead of signing off, send their report to the supervisor who can then in turn revise the report, comment on it, and eventually sign off the final version. Common Problems and Solutions When a patient has an appointment in the hospital, his or her information is entered into the hospital information system (HIS). This is a very crucial step since here the manual entry of data is done. After this, the information is transferred digitally to other systems when the hospital works digitally and systems are properly connected. Through this connection, the information from the HIS will automatically be used when the appointment for an examination is made in the radiological information system (RIS). As described earlier the examination information is then forwarded to a so-called broker that constructs the DICOM modality worklist so the technician can simply select a patient from the worklist and start the scan. Finally, the image data is transferred to the PACS after the information from the DICOM modality worklist and other information about the examination is included into the header. During this whole process, mistakes can occur at any level. These mistakes could be caused by human error, for example, a misspelled last name in the HIS. Often, these mistakes go unnoticed at first and they will then be corrected later. This means that these corrections have to be communicated to all systems that could have used the data already. Therefore, changes in the HIS have to be communicated to the RIS and changes in the RIS to the PACS to avoid the loss of patient data. Furthermore, problems can also be cause by technical failure, for example in the communication between the different systems because of which certain updates get lost or DICOM modality worklists are not available at the acquisition system. If a DICOM
Page 6 of 9
modality worklist is not available, data has to be entered manually, which is also a source of error. Basically, the manual entry of data has to be kept to a minimum to avoid errors. To correct and track errors, different mechanisms are available. First, the communication of changes by change messages between the different information systems will avoid discrepancies in the database. If these changes would not be communicated, the databases will differ and it might become extremely hard to find certain clinical information. Next, the PACS can check the incoming image data to detect problems or to signal that a problem might have occurred. First, the PACS can check whether the patient information in the header of the DICOM files it receives correspond to a study in the broker. This can be done by checking whether a combination of patient ID and accession number occur both in the broker and in the PACS. If there is a discrepancy, there likely was a mistake somewhere in the line between the broker and the acquisition device, for example because of a manual entry of the patient ID at the acquisition device because of a problem with the DMWL. If a radiologist would dictate such studies, the report could end up with the wrong patient, so studies that have such a discrepancy should be marked as such and dictation by the radiologist should be prohibited until the problem is solved. Another possible way to track problems is the use of unperformed and unscheduled lists. An unperformed list contains all studies that were scheduled in the RIS, but for which data was not received in the PACS. An unscheduled list contains all studies for which data was received in the PACS, but no entry was found in the RIS. A PACS administrator can detect such problems on the unperformed list and try to determine why the study was unperformed. An unperformed study could, for example, indicate that the data of that particular study was archived under a wrong patient by mistake. This can happen when a technician selects the wrong patient from the DICOM modality worklist, for example because multiple patients with the same sex and last name are in the list. However, an unperformed study can also simply mean that the patient did not turn up for the appointment (then the status in the RIS should not indicate that the patient arrived for his/her appointment) or that a study was aborted (for example when a patient get claustrophobic in the MRI). The unscheduled list could show, for example, scientific studies that are not scheduled in the RIS, but it could also again indicate studies with incorrect, manually entered, information.
Images for this section:
Page 7 of 9
Fig. 1: General components of a Picture Archiving and Communications System.
Page 8 of 9
Conclusion Since most radiology departments are digital and running PACS systems, every practicing radiologist should have a basic knowledge on PACS and the surrounding systems. Therefore, training in the background of digital systems such as PACS should be included in the training of radiology residents at a basic level.
Personal Information P.M.A. van Ooijen is a principal investigator at the department of radiology of the University Medical Center Groningen. His main interests are in radiological IT, ranging from PACS to advanced visualization.
References Recommended reading: •
PACS - A guide to the digital revolution. Dreyer KJ, Hirschorn DS, Thrall JH, Mehta A, editors. 2006 Springer Science&Business Media, Inc. ISBN 10:0-387-26010-2/ISBN 13:978-0387-26010-5 PACS and IMAGING INFORMATICS - Basic Principles and Applications. Huang HK. John Wiley & Sons, Inc., Hoboken, New Jersey. ISBN 0-471-25123-2.
Page 9 of 9