Like all disruptive technologies, picture archiving and communication systems (PACS) solve some problems with new ways of working, but also bring fresh challenges in their wake. Information overload is one of those issues occupying some experienced PACS users.
As the technology enables radiology departments to see more patients per day, produce more studies per patient and more images per study, so the potential grows for huge amounts of information to be generated in an organisation’s PACS.
Questions then arise about how this is managed. Which studies should be processed? Where should they be processed and by whom? Which images should be sent for primary interpretation and which stored in archives - and for how long?
Not a new problem
Visitors to the major radiological congress, UKRC, this year were treated to an authoritative account of the issues from one the US’s experts on the subject, Dr Katherine Andriole of the Brigham and Women’s Hospital, Boston. She is also chair of the sub-committee managing the TRIP (Transforming the Radiological Interpretation Process) Initiative which is dedicated to three basic objectives:
Dr Andriole had some reassuring news: the problem is not entirely novel. Since medical imaging was invented in the late nineteenth century, managing the information produced has always presented difficulties.
A position paper (subscription needed) by Dr Andriole and other TRIP sub-committee gives a full account of the history. The authors record that the early adopters of Wilhelm Rontgen’s “amazing new ray” had serious problems storing the large numbers of thick glass plates generated by the new diagnostic process. Not only were the plates heavy but volatile chemicals were used in processing.
Overload followed almost immediately as doctors sent large numbers of requests to early radiology departments, hoping to resolve longstanding questions and open up new lines of enquiry in all disciplines. The solution for hospitals in the twentieth century was to bring in more people to manage the growing workload – not an answer likely to win much approval today.
Dr Andriole told the UKRC that investigations on the current overload issues had even stretched as far a Hollywood where the film industry is learning to deal with huge amounts of computer generated graphics produced during modern filming and animation. The solution there seems to be to spend more money – again, not a popular option in cost-conscious Western healthcare systems.
Addressing the problem
TRIP has set itself the task of spearheading research, education and the discovery of innovative solutions to address the problem of information and image data overload.
Attempting to put a figure on the issue, the TRIP position paper refers to an informal study at the department of radiology at the Mayo Clinic, Jacksonville, Florida, which shows that roughly 1,500 images were generated and stored per day in 1994. At the same practice in 2002 and average of 16,000 images were acquired each day. It was estimated that, if the rate of growth continued at the same level, 80,000 images a day would be generated today.
Borrowing a term from the aeronautics industry, the TRIP sub-committee concluded that radiologists were getting “beyond the limits” meaning the computerised element of their work was getting too complex for simple human control.
“Radiologists today sit at the controls of their increasingly ‘difficult to fly’ workstation facing a challenge in imaging and information overload that transcends the traditional quick fixes of more personnel or more time,” they wrote.
Dr Andriole explained the approach TRIP was taking, working with corporate partners in the imaging and IT industries. The quest sounds simple: “to optimise what the radiologist does well and let the computer do what it does well.”
But the reality is much more subtle. For example, as Dr Andriole explained, it is very difficult for expert radiologists to tell you what they are doing when they look at an image.
Nevertheless TRIP has defined six areas where it believes research must be done if progress is to be made in managing the ever-increasing volume of data and make best use of both human and technical resources. These are:
Research to develop standards for image quality and display.
Studies to develop aids to the perception of features on an image and to develop a supportive workstation for the future. This would incorporate computer aids such as cueing, overlays and annotation. Decision support tools, simple reminder techniques, data mining and access to reference libraries are also included in this research area.
This area of research looks at new ways of displaying medical images on screen.
Research which looks at the increasing use of 3D and motion in imaging, virtual reality fly-throughs and hand-eye cueing instruments. In addition, hand-held devices will make images more readily available at the point of care.
The TRIP sub-committee says advances will be essential in the integration of hospital and radiology information systems with PACS and speech recognition systems. It advocates greater adoption of the framework developed by Integrating the Healthcare Enterprise (IHE) http://www.ihe.net/ .
A research area covering the development of objective methodologies for radiological imaging and as well as standard datasets for testing, comparison and collaborative research.
TRIP is a US initiative but some of the issues the initiative raises will resonate with UK users. In particular issues such as integration with a wide variety of other hospital systems will surface, especially as NHS England’s drive for ‘ruthless standardisation’ appears to be undergoing modification and moderation.
Research interests coincide too. For example, at the recent Clinical Care Computing conference in London, Dr Paul Taylor of the Centre for Health Informatics and Multiprofessional Education (CHIME) at University College London spoke about his research on clinical decision support tools in mammography screening.
Currently he has found the tools tend to have high sensitivity but low specificity and work is needed to tune any prompting system to reduce inappropriate false positives so that they are not routinely ignored. A slide presentation from the Clinical Care Computing conference programme and presentations section is available at www.abies.co.uk.
With the NHS’s drive to install PACS widely across the country in full swing and the decision to order and install the equipment taken, the coming years seem likely to bring another interesting wave of work focused on making the best use of PACS, the computer aids that can be used with the new systems - and the people who work with those tools.