High Tech Imaging

Dr. Stewart Spies of Northwestern University Medical School, Feinberg School of Medicine, takes us through the emerging trends in imaging informatics.

EHM. What is imaging informatics?

SS. An overriding definition would be the leveraging of technology to aid in the acquisition, analysis, reporting, transmission and archiving of patient imaging studies. This encompasses steps in the performance of imaging procedures and also addresses ancillary techniques and capabilities that are used in accomplishing those goals.

EHM. Imaging informatics is rapidly evolving. What is the main technology currently making an impact?

SS. If you asked that question of 100 people working in imaging, most of them would probably say that the use of digital picture archiving and communications systems (PACS) has been the major technology that has impacted the handling of medical images over the last 10 to 15 years.

These are systems that have changed the way images are interpreted, as we’ve moved from the old days of looking at X-ray films on viewboxes to doing virtually all of our image interpretation on computer monitors. That has had some advantages – it’s harder to lose files than it is to lose films – but there are some challenges as well. Speed of access can vary: it really depends on the network capabilities. This is usually not a problem in a hospital, but it can be more of a problem when you start looking at remote access of images.

The more we move toward digital data, the more there is a desire to be able to access it from various places not necessarily within the hospital or the radiology department. For example, we’re looking at accessing data at home to review urgent cases in the middle of the night; we’re looking at accessing data in the operating rooms, for the surgeons to be able to see images as they’re actually performing procedures, and also accessing data at other points of care areas within the hospital. This creates a challenge for the infrastructure people to make sure that the proper networking is in place, and to make sure that the workstations are communicating properly and can provide the appropriate quality of images for users in a timely fashion.

EHM. What other technologies do you see developing?

SS. There are many other technologies out there, and they have strengths and weaknesses. There definitely isn’t a global, universal solution that will satisfy everyone’s needs – from those of a hospital radiology department to those of an outpatient imaging center.

Speech recognition (SR) technology, although not typically thought of as imaging related, is something that’s beginning to see increased penetration in reporting of radiologic procedures. It has tremendous advantages in allowing us to get reports out rapidly, but there are also challenges in terms of shifting some of the responsibility of editing reports from transcriptionists to radiologists, some of whom are very adept with word processing and some of whom are not.

This can slow down the radiologists’ workflow, because it does take longer to report an examination using speech recognition software than it does to report one with the old digital dictation systems or on a cassette tape. But the advantages of using SR are significant, and I think this will be the prevailing method for generating reports in the future.

Another area that impacts imaging sections is what’s usually referred to as decision support – when an examination is ordered for example, ensuring that it’s the appropriate examination to perform in a given clinical setting. That has implications in terms of effective patient care and also reimbursement. Are you doing procedures that are likely to be reimbursed because they are appropriate and likely to positively effect patient outcomes? This is an evolving technology; I would say it’s certainly nowhere near mature at this point, but it will become increasingly important.

EHM. What would you say are the greatest challenges in the field?

SS. To me, the biggest challenge is integrating information from the diversity of information systems that store patient data. We have hospital information systems, radiology information systems and PAC systems, and maybe speech recognition systems, and right now, these play together with variable success. There are standards that are out there that try to make things work seamlessly, and it’s certainly the right direction to move in.

There’s the DICOM standard for images, which specifies what the file structure of digital images should be. It’s a complex standard that has evolved over many years. This standard has gone a long way to allowing interoperability, so that if someone has a PET scan on a scanner from one vendor and we want to look at it on our scanner or on our workstation, there’s a pretty good chance we will be able to look at the patient data and analyze it. But the devil is in the details with standards like this. Manufacturers are permitted to modify the standard when they have specific needs that may be unique to their application, but that flexibility creates a potential for incompatibilities from vendor to vendor. This is something that we deal with every day.

Then there’s Health Language 7 – the HL7 standard – for messaging or communication of orders and results as well as other information within the hospital, and that is separate from the DICOM standard. There is an effort called IHE, (Integrating the Healthcare Enterprise), which tries to reconcile and coordinate these standards and create rules and strategies or paradigms so that medical information can be shared and flow across diverse systems. It’s a huge challenge at the moment, and we certainly are nowhere near to fully solving this problem.

Other challenges that are more easily handled are things like storage requirements. More of our images are digital, and more examinations are starting to generate a lot more data than they used to. For example, a CT scan used to generate a certain amount of information that needed to be stored, but we are now creating additional projections (‘views’) of patient images and performing other processing steps which need to be stored, and the amount of data that needs to be archived has expanded significantly.

The computer industry is also evolving their technologies for mass storage, and they’re doing an excellent job of that, but what if you have an issue where you may have an archive in the hospital, it’s expensive to replace it, and you can’t just adopt a new storage technology every year. One downside of having limited storage capacity is the time it takes to retrieve prior studies from a long-term archive, which is usually much slower than a short-term archive. In the future, hopefully we’ll be moving more toward having all of our data online, regardless of how old it is – so that the access will be fast.

EHM. What are some of the hot topics in imaging informatics at the moment?

SS. Decisions support will continue to evolve over the coming years, and so will computer aided diagnosis and detection (CAD). As our computer systems become more capable and powerful, we can manage our processes in more depth and implement more sophisticated algorithms. We will have a greater ability to, for example, detect breast tumors on mammograms or detect pulmonary nodules on CT exams or chest X-rays. Ultimately this is going to be an important area to help improve the efficiency and accuracy of diagnosis.

Another topic which is not completely mainstream, but is worth mentioning, is the role of so-called open source solutions for solving medical imaging problems. Open source programs are developed by a sophisticated user community. They’re made available to anyone who wants them for free. Anyone is free to modify them and then submit their changes to be included in the existing body of information (if appropriate).

In the computer realm at large, open source is a very powerful concept. For example, the Linux operating system, is an open source operating system that does what Windows does or what the Macintosh operating system does, and yet it’s essentially free. Anyone can download it, install it and use it.

There are certainly examples of this already in medicine. At this point, it’s a vertical area and not widespread, but I suspect we will see more and more of it in the future.