Portal de Periódicos da CAPES

Medical Imaging Workstations: What Is Missing and What Is Coming?

2009; American Medical Association; Volume: 133; Issue: 4 Linguagem: Inglês

10.5858/133.4.542

1543-2165

Michael W. Vannier,

Advanced Radiotherapy Techniques

During the past 2 decades, radiology has undergone a transformation from film to digital. Now moving from glass slides to digital imaging, pathology is in an early stage of a similar evolution and may benefit from a report on how radiology made this change in technology. More importantly, where will radiology be in the future? Many of those experiences were not entirely pleasant, and your awareness of the potential pitfalls might prove useful for planning and decision making.To a clinician, the most visible and important components of a digital imaging system are medical imaging workstations and, more recently, image-guided therapy workstations, particularly those used in surgery and radiotherapy. This article explains how workstation applications are evolving in clinical practice. I introduce workflow issues and some key acronyms. My objective is to help you understand the workstations, what motivated their development, and their interfaces to imaging modalities. Knowing this, you can appreciate how radiology, surgery, and radiation oncology can work together productively in a clinical enterprise.There are 2 distinct types of workstations. The first type is called a “thick client” workstation. These high-performance and high-resolution workstations are where radiologists spend most of their time and make most of their decisions. In contrast, there are “thin client” enterprise workstations that might be found on the clinic floor, on a physician's desk or in his or her home office, or most recently as part of a wireless device. These thin client workstations include specially programmed Web browsers.There is another way to think about all of this. The thick client workstations are somewhat akin to the familiar gaming platforms (eg, Xbox [Microsoft Corporation, Redmond, Washington] and PlayStation [Sony Computer Entertainment America, Inc, Foster City, California]). The gaming platforms contain high-performance computing capabilities and computer graphics hardware, both of which are run by local software. The gaming platforms can operate autonomously, similar to the way we use thick client workstations in radiology. We download sets of images to these workstations and locally operate the workstations (ie, without the need for a remote host computer to manipulate the images).The thin client portals are display stations like those you might have on your office or home desk or on a wireless device, such as a Web-enabled cell phone or personal digital assistant. These thin client displays work similar to Google (Mountain View, California) in that they both provide computing resources using a remote set of host computers. Imaging Web services can now be found everywhere, especially in a modern hospital or medical center. It is common today for radiologists to carry a cell phone and pager; therefore, if paged while on call, we can immediately and remotely review images from anywhere. Today, we can combine the cell phone and image review station in a single portable wireless appliance, so images are available anytime and anywhere they are needed.Surgery represents an important type of image-guided therapy where workstation technology is significantly changing clinical practice. Surgeons place demands on workstations for high-quality images and for local interaction, similar to how radiologists are using thick client systems. However, surgical workstations have become highly specialized and are now able to do certain tasks exceptionally well.In radiotherapy, we provide imaging services by moving data between modalities. For example, the results of a computed tomography (CT) scan or magnetic resonance imaging (MRI) can be sent to a picture archiving and communications system (PACS) and then through the hospital network to various workstations. Radiotherapy workstations may look similar to the workstations radiologists use; however, they offer virtual simulation, dosimetry, and treatment planning. Radiotherapy workstations are highly optimized for delivering therapy, for example, by controlling a linear accelerator. In this specialized treatment environment, the general purpose nature of workstations is of less importance than the specialized, suited-to-task performance.Because any given patient often has multiple encounters with an imaging modality, such as a CT scan or MRI, an image-based patient and disease model can be used to integrate information across episodes of care. A patient-specific model, informed by these encounters, provides the basis for making decisions and for individualizing the delivery of care. In clinical practice, we see many patients over and over again. We may initially study the patient with 1 modality (eg, MRI, positron emission tomography, or CT scan) for diagnosis and staging. Next, treatment is given, and follow-up scans are obtained. The imaging information from various sources can be referenced to an abstract model using an individualized anatomic framework. Accomplishing this integration, automating the steps needed to build a 3-dimensional model, and making it available to surgeons and radiation oncologists while delivering value at every step of the process is one of the principal challenges facing radiologists today.Added pressures include time constraints and a limited number of skilled support personnel. We cannot laboriously manually outline anatomic structures in images, identifying normal tissue structures and tumor targets using extremely large data sets. For this reason, practical medical imaging workstations have software to automate these processes. The automated software tools we use can be informed and guided by prior experience, so that we may reuse the organ definitions from archived cases with human operator supervision. The prior image information, with boundary outlines and anatomic labels, is used as a deformable template that is adjusted to fit each patient's anatomy.A more general model is emerging in which image data on a given patient are obtained from multiple sources and multiple encounters. These data sets are inserted into an integrated model. The sums and differences of morphologic and functional changes over time are used to design a treatment plan, deliver the plan to the therapy device, administer the treatments, and provide follow-up. This is done with a feedback loop, so the experience gained from an individual or from a group of individuals is available at the workstation. Software tools allow users to access and apply this accumulated experience.Imaging workstations are used in many clinical domains, and each domain provides its own set of challenges. In today's operating room, several workstations may be present. For example, an operating room may have an ultrasound workstation, multiple imaging devices including a fluoroscopy system, various monitors, and a surgical navigation system among others. All of this information is available to the surgeon; however, each system likely came from a different vendor. Each component can work individually, but interfaces and standards are lacking. Therefore, the interoperability of the equipment in the operating room may be nonexistent (ie, the staff cannot exchange data). Moreover, they may not be amenable to an integrated system.Take for example endoscopic minimally invasive spinal surgery. In this environment, the surgeon seeks to deliver the care using meticulous surgical procedures without a laminectomy. He or she makes a very small incision and enters the spine with a suitable endoscope. To do this, the patient is anesthetized and monitored. There are live fluoroscopic images that are registered to CT and MRI preoperative images, and the results must be delivered in the endoscopic context. It is a major technical challenge to integrate all of these devices and to use most effectively the limited amount of time that is available.DICOM 3.0 (DICOM, Rosslyn, Virginia), which is an acronym for digital image communications, is a set of voluntary medical imaging data communication standards developed during the past 2 decades. The American College of Radiology found that the inability to interoperate equipment from different vendors was not only a vexing problem but also a barrier to progress in the early days of PACS. Therefore, DICOM was developed in conjunction with the medical imaging system industry. Once implemented, DICOM enables medical imaging systems or workstations to interoperate, using data acquired with various modalities, such as CT, MRI, ultrasound, positron emission tomography, and single-photon emission computed tomography. DICOM is a voluntary standard because there is no regulatory or legal requirement to provide this capability. Today, however, the medical imaging market demands this level of compatibility to ensure that imaging systems, at least the ones in radiology, will interoperate. Is DICOM for diagnostic radiology alone? No. DICOM is widely used in radiation oncology, cardiology, dentistry, and other disciplines. DICOM standards are available in the public domain and can be found at the DICOM Web site (http://medical.nema.org, accessed October 1, 2008). DICOM/National Electrical Manufacturers Association is a very large virtual organization with extensive participation in its numerous specialty groups by industry, academia, and government.Understanding some key imaging informatics terms and their definitions is important. First, recognize that workflow is an organizing principle that applies across specialties. The sequence of steps that is necessary to perform an imaging task or to complete an interventional procedure may be termed a surgical workflow or a radiologic workflow.Standards for imaging are developed and disseminated by DICOM, the nonprofit digital imaging and communications group sponsored by the National Electrical Manufacturers Association. DICOM is the world's leading authority on standards for medical imaging formats, thereby enabling interoperability of image (like CT scans and MRIs) generation devices with workstations, archives, networks, and so on. The workgroup is an important entity within the DICOM framework. For example, member-users from industry, academia, and private practice collaborate to develop specific standards in their clinical disciplines. There are DICOM workgroups for radiology, surgery, cardiology, dentistry, and others. Members from the principal specialties are represented in the DICOM standards-generation process.Hospital workflows are ubiquitous. As patients enter the health care system, there are workflows for administration to perform admission, discharge, transfer; for diagnosis; for numerous radiology-specific tasks including image acquisition, postprocessing, and image processing; for delivering and administering drugs; and for sedation and anesthesia, to name a few. Ultimately, these workflows have to work in a coordinated fashion to allow a complex surgical procedure to take place. Within the operating suite, there is a surgical workflow that includes the processes, implements, and information that are present in this setting.Here is an example of how we use workstations for surgical planning. Consider an adult patient with a fracture nonunion in the midhumerus that needs to be corrected. The original radiographic image shows an easily recognizable displaced fracture. In making a surgical plan, the doctor can load the image into a workstation that automatically outlines the distal fracture fragment on the screen and permits the user to interactively move the image so that he or she can determine exactly what type of bone plate to use in the fixation. Similar steps are used in hip replacement surgery. Integrated data at a workstation can help to properly size a hip prosthesis, to insert a hip prosthesis, to determine the dimensions of a custom bone plate, to specify the placement of screws, and so on. The integration of knowledge from a variety of sources and the delivery of this information to the surgeon in a convenient and comprehensible manner can be achieved using workstations.Workstations have advanced far beyond what can be achieved with a software module used at a single location. By leveraging an investment in computer networks, we are now able to access validated best practices for a particular type of procedure, using scripts contained in a shared repository. The technical infrastructure for capturing and disseminating workflows is now under development. In the future, the infrastructure necessary to access a worldwide workflow archive will exist, and members will be able to reuse the best available experiences obtained at other centers.Workflows are often illustrated as flow charts. Specific types of workflows can be captured in a repository for reuse and refinement. Robots can capture and automate workflows to better assist and augment the surgeon's performance of complex procedures. Thus, robots could reduce errors and shorten interventions. To this end, we must capture gestures and motions in the context of images, a process that demands closer and closer integration of workstations. To some extent, workstations have already been integrated in the surgical operating room. For example, robotic prostatectomy is common in many centers. The surgeon sits during the procedure, does not wear gloves, and does not scrub in. The surgeon must wear scrub clothes only because he or she is seated next to the patient. An attendant in the operating room places the robot's arms through ports in the patient's body. The actual surgical procedure is done remotely by the surgeon, who views the endoscopic camera images through 2 visual channels, with stereo viewing at the workstation. The actual effectors that move the robot are controlled by the surgeon's hands. To be safe and effective, robotic surgery requires highly reliable interfaces among the integrated workstations or components in the system.Another image-guided therapy workstation that provides high-intensity focused ultrasound surgery is now being introduced in many centers. This is an example where a workstation is used while the patient lies inside a magnetic resonance scanner. A focused ultrasound array transducer is placed outside the patient. This array can deliver intense ultrasound to a focused spot within the body. The intense ultrasound waves converge at a point and ablate the tissue at this point without an incision. As the patient lies on the magnetic resonance table, images are obtained that reveal the temperature distribution in the body. For ablation of uterine fibroids, for example, we can ablate the lesions by delivering the ultrasound signal under magnetic resonance guidance. There is no incision, and little, if any, anesthesia is required. Therefore, patients can be treated in an outpatient setting. This procedure requires the integration of many specialized hardware and software tools that are implemented using an imaging workstation.In the future, the autopsy suite may be augmented with premortem and postmortem images, head-mounted displays, and integration of instruments. In addition to tracking these components, such a computer system will advise the user where to sample tissue and how best to conduct the investigation.We learned from prior difficulties and experience that we need more than just DICOM standards to use workstations to their full potential. Therefore, a new set of procedures called the integrated health care enterprise (IHE) has been developed to achieve the promise of digital imaging that was not realized with DICOM alone. Consider the motivation behind this new set of procedures to understand why its development was necessary.The CT scan volume per year during the last several years has grown and continues to grow, with a 10% to 15% yearly increase in the number of examinations at a time when the number of images per examination continues to grow overall. We have seen an increasing number of examinations, the examinations are more complex, the number of images per examination is greater, and many examinations add a requirement for postprocessing. While this has occurred, the number of radiologists has remained almost constant. Therefore, in the same number of hours per day, radiologists must complete a much greater amount of work. This has created a “perfect storm.”1 In 2000, there was a severe shortage of diagnostic radiologists and many unfilled positions. The total imaging workload grew much faster than the usual number of radiologists in practice, creating a persistent shortage.By 2003, the crisis was largely over; the shortage was no longer as intense. What happened? Increased productivity was achieved through the introduction of digital imaging and PACS. We improved the information technology (IT) infrastructure, and now radiology departments are filmless and paperless. It was the adoption of DICOM and the procurement of DICOM-compatible systems, especially workstations, which made it possible to do the postprocessing and gain access to interpret and manage images through the enterprise network infrastructure.A paper presented at the 2004 Radiological Society of North America's annual meeting in Chicago discussed the IHE-based integration success of PACS, the radiologic information system, imaging modalities, and the hospital information system at Mayo Clinic in Jacksonville, Florida.2 The authors met with their vendors and decided to use IHE integration profiles to automate different transactions to streamline their workflow. The results were dramatic. They reduced errors, lowered staffing requirements, improved turnaround time, and accomplished all of this with the same equipment and personnel.Today, we are focused on workflows that involve more than just radiology. We want to gain productivity in other aspects of imaging and in other departments. There are now sources of images in pathology, cardiology, and other fields. However, it is impractical to buy every department or area its own imaging management system and then try to interface all of the systems. It is just not practical. Each clinical enterprise will eventually have a single image archive or PACS that serves all of the image-generating modalities and medical specialties.We need to think about standards. The importance of standards is clear when we deal with different electrical plugs and voltages or with plumbing or tires or CDs and DVDs. The interoperability of items from different manufacturers is crucially important, and when these items are compatible, the users benefit greatly. The antithesis of this would be to buy proprietary products so you are locked into buying everything you need from a single vendor. This is good for the vendor but bad for the user. In the computer industry, proprietary standards have largely evaporated along with a number of the manufacturers that provided them. We do not need to make this mistake again.The need for standards is obvious when we travel from country to country with an electrical appliance. The electrical connector that works in the United Kingdom will not work in France. The same holds true for medical images. The DICOM standard, which is currently in version 3.0, is the medical image format that originated from an agreement between the American College of Radiology and the National Electrical Manufacturers Association. The National Electrical Manufacturers Association now sponsors DICOM, which manages a dissemination process as well as a certification process. The certification process provides proof or documentation that a specific device complies with the standard and is interoperable with other designated devices. Because the DICOM standard is constantly evolving, it is only incompletely developed and partially implemented in any real system.DICOM really represents a plug-and-play interface that provides assurance—if you plug something in, it will actually interoperate. The IHE is like the wiring diagram. Integrated health care enterprise specifies how things, especially information items, are moved from place to place, how these systems actually integrate to implement a useful workflow. The DICOM standard is restricted in that it does not specify how to implement an interconnection. DICOM defines the data elements; however, it does not define the database structure, the hardware, or the operating system.The practical significance of a DICOM conformance statement is very high. We learned from hard lessons that because DICOM is a voluntary standard, many early “DICOM-compatible” systems did not interoperate. Manufacturer A would say, “We have a DICOM device,” and manufacturer B would say, “We have a DICOM device.” However, when we plugged the 2 devices together, they did not work. Then, we went back to the first manufacturer who said, “I ran our tests, and everything we promised is working.” Manufacturer B said the same thing. Therefore, we learned that we needed a DICOM conformance statement in which manufacturers certify that their devices work with other DICOM-certified devices. This document states that a specific product conforms to DICOM functionality and that the product will work in an integrated system with other DICOM-designated devices. This is not a perfect system, but it has proven to be a major step, albeit learned the hard way, that moves us toward a fully integrated system.Integrated health care enterprise is a user-driven, standards-based, global effort with a number of implementations. However, it is not a standard in and of itself. Interfaces are expensive, and as users, we often do not understand exactly how to specify each interface. Therefore, we risk repeating the 1970s or even the 1990s when we paid a premium price to vendors for every interface. Now, with IHE implementation profiles, integrating devices can be less expensive than adding a new telephone line because IHE eliminates some options in published standards to constrain manufacturers to work together effectively. Integrated health care enterprise is not a standard, but it uses standards. It is a blueprint for solving workflow problems.The IHE sponsors meetings where domain experts apply their knowledge and expertise to achieve integration and implement workflows. There is already an enormous base of intellectual property that has come about as a result of many professional organizations working on the IHE. Integrated health care enterprise is a joint initiative between the American College of Radiology, several radiology-related professional organizations, and the College of American Pathologists. Integrated health care enterprise has been around since 1997 and has grown significantly since then.Integrated health care enterprise works in various domains. There is a radiology domain, a laboratory domain, a pathology domain, and so on. Integration profiles are available in these various areas and are developed through a standard-based process. The IHE process involves assembling a group of experts, providing a formal way of making decisions, and promulgating solutions.Integrated health care enterprise profiles can be fairly complex. This technology is more suited to IT experts for understanding the details, but it provides a workflow or way of organizing information about images, evidence documents, reporting, postprocessing, and scheduling. Integrated health care enterprise is a layered, compartmentalized framework in which to develop workflow for integration. This information is published in documents that are free to the public. On the IHE Web site, there is a framework for anatomic pathology (a 2-volume edition); one volume contains integration profiles (41 pages), and the second volume contains a technical framework (82 pages).There are also specialized IHE meetings called “connectathons.” A connectathon includes a group of vendors who have all connected their systems together. If a user is considering integrating some of these different systems, he or she can “road test” the systems at one of these meetings. The testing is done in different domains, such as radiology and cardiology, and often involves many systems. Therefore, a connectathon is a good way to accomplish some essential testing of components and to guide users' in their decision-making process.Drawing on my experience on how not to do things, I can think of several potential mistakes that could create problems. The first mistake is buying proprietary systems for which you can use only one vendor's products. We have tried that with poor results. Second, talk to your friends and colleagues to select the best-in-class system based on their prior experience. Later you will learn that last year's system is no longer that good, and it had some shortcomings that you never learned about until the successor appeared. So, you do not want to buy the first of anything. This is not to say that vendors are dishonest, because I think they believe many of their promises. However, if technologies are not appropriately and thoroughly tested, a user may not have the capabilities to resolve an impasse.In some cases, if you are not careful in how you write your procurement orders and in how you deal with vendors, you cannot really expect them to continue to be interested. Some vendors are under duress because of market conditions. Therefore, the limited ability of these vendors to resolve complex problems for individual users is understandable.If you become enamored with a particular workstation, consider what is going to happen to the computer systems that you are using now or that you are about to buy. All of these systems are destined to become obsolete, so buy functionality that you can readily use. You also need to be able to amortize the cost of systems during the time in which you can economically operate them.Google, Wikipedia (San Francisco, California), and medical enterprise computing are not done on ordinary personal computers but on server farms. The server farms are somewhere at the clinical enterprise or are dispersed at multiple remote locations. Server farms are ubiquitous. The serving of computing resources is basically a commodity that can be purchased and used. However, if you want to invest in this sort of thing, you will be competing with organizations that have a lot of horsepower and specialized knowledge. The strategy for Web hosting for an enterprise should focus on keeping down costs, trying to consolidate services, using the fewest number of machines, and keeping in mind the effect on workstations.Because workstations are rapidly evolving, know where to find additional information online. For example, KLAS (Orem, Utah) is an organization that provides consumer reports for health care IT and workstations. KLAS is an independent organization that uses customer surveys to evaluate software and systems. Similarly, in the United Kingdom, there is an entity called PACSnet, which is currently part of the Centre for Evidence-based Purchasing funded by the National Health Service Purchasing and Supply Agency. PACSnet supports PACS users, provides consumer feedback, disseminates useful nonproprietary information on products, and helps users make decisions.Open Source and Open Systems are user groups that collaboratively develop technology and provide public access to software. The open systems have been growing in importance and play a significant role in medical imaging. People who infrequently use medical images may be satisfied with a workstation that uses Open Source software, which is free and part of the public domain. For example, OsiriX (Apple, Inc, Cupertino, California) is a very sophisticated system that operates on Apple Mac computers and is available on the Internet at no cost. Some small clinics use a free Open Source PACS software system. Although the user does not pay for the software, he or she may require consulting services and must procure suitable computer equipment on which to effectively operate the software. Wayne Rasband, MS, National Institutes of Health, has contributed ImageJ, a widely used public-domain image processing and visualization software package written in JAVA.At the United States National Library of Medicine, which is part of the National Institutes of Health, an effort was made to build open-source software tools for image processing and analysis. This project is called Insight, and its principal software product is called the Imaging Tool Kit, a very large, sophisticated, and no-cost set of software modules for imaging. The open-source Insight project also resulted in the establishment of an innovative, online software journal called the Insight Journal, which has an open-source review policy. You will also find online pathology blogs and other information related to microscopic imaging.Carl Jaffe, MD, will present the National Institutes of Health initiatives, particularly the Cancer Biomedical Informatics Grid, which focuses on cancer informatics but represents a tremendous resource that ties together many IT technologies with a complex infrastructure that is highly visionary. There are pathology software tools and pathology work spaces inside of the Cancer Biomedical Informatics Grid that should be considered.Those who have a pathology background are probably familiar with the annual conferences called “Advancing Practice, Instruction and Innovation Through Informatics.” Most of what we know about pathology informatics comes as a result of these meetings. Finally, for imaging informatics in general, a useful resource is the Society for Imaging Informatics in Medicine.The integration of medical imaging workstations and IT is rapidly evolving. However, integration is not yet at an end-stage. While it is challenging to follow and understand the trends, we have to make decisions to realize the benefits of these systems and to realize the transformation from analog to digital imaging. Those who use the available tools to integrate successfully will prosper, and those who do not face marginalization and peril.