Without much fanfare, the Baltimore VA Medical Center passed its 15-year milestone as the first filmless (with the exception of mammography) hospital on June 23, 2008. Instead of marking the occasion with a big celebration, we took this opportunity to reflect on the many things that have changed and, perhaps more interesting, the things that have stayed the same.
The most visible change is the radical redesign of our radiology reading room (Figures 1 and 2), an effort based on a topto- bottom assessment of all our room design mistakes from the first 10 years of filmless operation. One immediately successful technology in the new installation has been an effective and relatively inexpensive sound masking system along with accompanying sound-absorbing materials.
In combination, these environmental additions not only reduce noises from outside the room (such as those from our MR scanner next door) but also minimize the cacophony associated with conversations, dictation, and other routine noises within the room itself.
Other innovative features include icons on each workstation to control overhead lights, individual LED task lighting, adjustable monitor and table heights to accommodate sitting or standing, individual ventilation controls, ergonomic chairs (Figure 3), modular partitions, and biometric and key fobcontrolled room entry. These—as well as unconventional and evolving touches such as scented candles and a treadmill workstation— have combined to create a low-stress, productive, and pleasant place to spend the working day.
In addition to the surroundings, many aspects of our daily work have changed. We now dictate using a speech recognition system. The system requires about 15% more time for a radiologist to dictate an individual case but has reduced our report turnaround times from hours to minutes and report signing times from days to minutes. Our new radiology residents seem to be completely comfortable with the system and do not understand how we ever managed to organize our thoughts in the “old days, when you really couldn’t edit the report once it was spoken, without rewinding and dictating again.”
We now routinely use an advanced workstation and dictate directly from the 3D workstation or use it in almost every case reviewed on our PACS. Our average body CT study has increased from approximately 60 to around 900 images, which are required for review of all CT studies in multiple planes and using maximum intensity projection and volume rendering. Unfortunately, as was the case in 1993, many features that are routinely required, such as fusion for PET/CT and arithmetic summing of thin into thicker sections, are not available on our PACS but must be accessed using a specialized workstation.
Our medical records became entirely electronic within a few years after we acquired our PACS. This provides us with complete access at all locations to the entire patient record (although we are required to open two simultaneous windows and log into the systems separately, which is cumbersome).
Routine access to the Internet from the PACS workstation now allows us to use search engines such as Google or the radiology domain–specific Yottalook. This has substantially increased the frequency with which we look up interesting and obscure information as part of the image interpretation process.
We communicate with clinicians and other physicians by e-mail and via our imaging reports, which are now available almost immediately. As a result of ubiquitous access to images throughout the enterprise and rapid report turnaround, clinical colleagues now visit for in-person consultations only rarely, especially for general radiography where the frequency of these consultations has dropped to less than one-seventh of what it was in 1993.
We are finally making the transition from film to digital mammography, based on the encouraging results of the Digital Mammographic Imaging Screening Trial and recent improvements in workstation-based interpretation of mammograms.
WHAT HAS NOT CHANGED
The basic human-machine interface with the PACS has not changed significantly. The original model, based on Adobe Photoshop with various icons and/or pull-down menus to change window/level or point, zoom, magnify, and draw a region of interest using a mouse or trackball, has not changed perceptibly. This is despite the fact that the way in which a radiologist interacts with a workstation when interpreting a large data set of CT, MR, or other modality studies with or without comparison studies has changed dramatically.
This process used to be more passive, similar to that involved in film interpretation, but it has now evolved into a much more interactive and dynamic experience in which the radiologist/clinician navigates through the images using multiple planes/perspectives. The tremendous amount of work that has gone into improving the human-machine interface in the computer gaming and aviation industries has, regrettably, not yet been adopted by PACS vendors.
The way in which we sign on to and off from the PACS has not changed since 1993. Despite innovations in more efficient ways of signing on to information systems (such as the use of the Lightweight Directory Access Protocol, which can support a single user name and password shared among multiple information systems), our hospital/radiology information system, PACS, speech recognition system, and MR, PET/CT, nuclear medicine, and 3D workstations all still require separate sign-ons. Equally important, newly available technologies that can provide automatic log-on and log-off, such as radiofrequency identification, have not been implemented at our facility.
Capturing cases to a digital teaching file or tickler file system remains a surprisingly cumbersome process. We can send individually selected cases or whole series to a specific destination folder for future reference, but this does not allow us to capture the images to a structured teaching file, to de-identify the studies for Health Insurance Portability and Accountability Act purposes, or to facilitate offline access. An Integrating the Healthcare Enterprise profile, Teaching file and Clinical trial Export (TCE), describes a way to select, deidentify, and then export images and descriptive information from a PACS to a teaching file system. Unfortunately, no commercial PACS vendor or advanced visualization vendor currently supports this profile.
We still have no way to efficiently share images with clinicians, research collaborators, or other healthcare workers outside our own facility. In 1993, we printed film for outside clinicians, amounting to approximately 5% of our volume of studies. We found this to be less than satisfactory because of the limited quality of the dynamic range of digital films and the extra resources and time required to print and send the films. We now transfer studies in DICOM format onto CDs—a process that brings with it new problems, inefficiencies, and quality issues. The creation of CDs requires additional time for technologists or administrators to burn, handle, and mail or give the CDs to patients. Programs embedded on the CDs frequently require administrative access that clinicians do not have on their own PCs, use software that conflicts with other programs, or use nonintuitive software. We have as many complaints about the CDs now as we used to have with printed films.
The amount of time required to retrieve images to our workstations has not improved. In fact, image retrieval time has actually increased slightly because of the complexity of imaging studies and the number of images per study. Solutions are available in the larger world of electronic information management. Large image sets, such as Google Earth and MapQuest maps, are not stored locally on a PC but are “streamed” on an as-needed basis to the user. This approach allows almost immediate on-demand review of images.
It is likely that medical imaging will transition to server-side rendering using this type of streaming technology over the next few years. However, only a minority of PACS vendors are currently investigating or offering this technology.
PACS has not had a fundamental impact on the ways in which we communicate urgent/important findings to our colleagues. We have faster turnaround on reports but do not know whether these reports have actually been reviewed, much less acted upon. Communication is still primarily one way.
In our department, we manually track urgent/important findings and notify the requesting clinician or, when that provider cannot be contacted, another clinician designated with this responsibility. We later check to determine whether a recommendation, such as a follow-up CT, was actually initiated. If not, then we communicate with the clinician again to ask about the reason for the lack of follow-up. This critical communication issue continues to have a major impact on patient care. In 1993, we anticipated that the process would be automated well before 2008, but this has not been the case.
Fifteen years ago, we anticipated that the extraordinary ability to capture all aspects of the image acquisition, interpretation, and communication process would dovetail nicely with similar access to patient electronic medical records to provide the outcomes data that radiology very much needs to defend rapid technological change and innovative techniques. The hope was that it would be a relatively simple matter to automatically follow the effects of different imaging techniques and technologies “downstream” to provide real quantitative metrics about costs and benefits. Although many investigators are certainly at work gathering these data, for the most part they continue to do it with hard and often manual slogs through the records. We have not yet capitalized on the research potential of all the information we now capture.
One final aspect of our work has not changed: the sense of exploration, adventure, and—every once in a while—magic that we have somehow managed to preserve over these past 15 years. Our imaging informatics team and our clinical radiologists from the University of Maryland School of Medicine and the Baltimore VA Medical Center continue to question the status quo in diagnostic imaging, to explore new technologies from other domains, and to maintain a healthy degree of skepticism and curiosity about our changing practice of diagnostic imaging.
