College of American Pathologists

  All aboard—new DICOM standard for digital images


CAP Today




September 2010
Feature Story

Anne Ford

It may take only two to tango, but it takes many more than that to make a new technological standard useful. No matter how needed or valuable the standard is, its utility is limited until a critical mass of vendors and users adopt it. Think of it in these terms, says Wil Baas, senior director of software development at Philips Digital Pathology, the Netherlands: “If there is only one mobile phone, you cannot call anybody. There have to be at least two. The ball has to start rolling.”

When it comes to new Digital Imaging and Communication in Medicine Standard (DICOM) Supplement 145, “Whole Slide Microscopic Images,” the ball may not be rolling quite yet, but with the standard’s approval in August, it has been given a hearty shove. Developed over the past five years by DICOM Working Group 26, with input from the CAP’s Diagnostic Intelligence and Health Information Technology Committee, Supplement 145 aims to allow pathologists to retrieve and manage whole-slide images and associated information from any digital imaging system, regardless of the manufacturer. And while it will take time for manufacturers to implement the standard in their systems, its eventual consequences for pathologists are expected to be considerable.

Working group co-chair and ­DIHIT Committee member Bruce Beck­with, MD, chair of pathology at North Shore Medical Center, Salem, Mass., explains one of the standard’s implications for pathologists: “If their institution has a PACS [Picture Archiving and Communication System]—which is currently where most hospitals store their electronic radiology images—the pathology images would be able to be stored in those same repositories, and they might be able to view them with some of the same viewing software they have now.”

And that’s a big deal for a lot of reasons, says Working Group 26 member John Gilbertson, MD, “given the cost of maintaining and building and supporting large PACS environments.” He adds, “This won’t happen right away, because the vendors have to incorporate this new supplement into their designs. But over the next few years most vendors will do that, and you’ll be able to put your whole-slide images in a normal PACS environment.” Dr. Gilbertson is associate chief of pathology and director of pathology informatics at Massachusetts General Hospital and associate professor of pathology at Harvard Medical School.

To look at the big picture (no pun intended), “The idea is to facilitate interoperability between equipment from different vendors,” Dr. Beckwith says. “In radiology it’s very common that you might buy your CT scanner from one vendor and your MRI machine from another vendor, and your PACS will come from maybe a third vendor, and they can all talk to each other. That’s what we’d like to get to in pathology—where you can buy a scanner from vendor A and viewing software from vendor B, and you can store it all on your current hospital PACS you got from vendor C.”

That will mean, says CAP medical imaging technology consultant James MacDonald, that “the pathologist doesn’t have to worry about: ‘Well, I can’t read this image, because it came from’” an incompatible system. “Or, if I’m a general pathologist in Indiana and you’re a derm path in California, I can send my image to you to read. It improves patient care, and it opens up revenue and consultation opportunities, because pathologists can do on-shoring or off-shoring, whatever the trend may be.”

The DICOM standard is a medical image exchange standard that has achieved widespread acceptance since it was introduced in 1993. By now, “most all PACS in medicine use DICOM as their core,” Dr. Gilbertson explains. From a pathologist’s point of view, however, DICOM’s usefulness has been greatly limited because it was conceived primarily as a tool for radiologists.

In its original form, “DICOM had the ability to handle simple pathology images, but didn’t have much of a way to describe what they were,” Dr. Beckwith says. “The DICOM standard didn’t have the concept of a specimen, really. It just assumed that all the images were going to be of a patient.” Since then, he says, “what we’ve tried to do is make some extensions to it so that it can be more widely adopted within pathology.” A first step was taken in 2008, when Working Group 26 issued Supplement 122, which details the identification and description of specimens.

The next challenge—image size, particularly as it pertains to whole-slide imaging—led to the creation of Supplement 145. “Whole-slide images are too big for the DICOM standard,” Dr. Beckwith explains. “There are some restrictions [in the standard] on the size of images, so you couldn’t just take a whole-slide image and stuff it into DICOM, so to speak.”

Whole-slide images are not only large but also interactive, Dr. Gilbertson points out. “Pathologists are always zooming and panning and zooming some more.” Why is that a problem? Because with a typical radiology image, Dr. Beckwith says, “DICOM would assume you’re going to transfer the whole image and look at the whole image. You could zoom in if you wanted to, but the images are small enough that you can just move the whole image around.” By contrast, whole-slide images are so large that “for performance reasons, you don’t want to move all that data around every time; you just want to look at a tiny piece of it.”

These challenges produced much discussion within the working group about “whether we should ask for there to be a new type of DICOM images that have an unlimited size or a very large upper limit on size,” Dr. Beckwith says. “This would be a major change to the architecture of DICOM. The problem was that it was felt to not be backward-compatible. All the existing stuff wouldn’t work, not even minimally. So we’ve tried to make sort of a backward-compatible method for integrating whole-slide images into the DICOM standard, so that we don’t break existing systems that are out there.”

The solution proved to be breaking whole-slide images into smaller regions, or tiles, of up to 64,000 by 64,000 pixels each. That way, “when the pathologist is looking at the slide and says, ‘Okay, I want to zoom in on this region at 10×, and I need to pull out those 20 tiles,’ only those 20 tiles will be sent from the server to the computer desktop that the pathologist is using,” Dr. Beckwith explains. That will allow the user to access even very large images at different resolutions and in different regions of interest quickly and easily.

Another implication for end users: Supplement 145 allows for the possibility of different focus planes. “You can change the focus and see different planes within the tissue,” Dr. Beckwith says. “So if your slide scanner can scan multiple focus planes, we can incorporate that into the digital image in DICOM. There are a lot of slides where you need that.”

So how do vendors feel about all this? Well, some were initially resistant to the notion of a common standard for whole-slide imaging “because each of them had their own algorithms and proprietary software,” MacDonald says. But after several of them were invited to participate in the supplement’s creation, he says, they realized, “It’s a business opportunity, and we had better get on it in the infancy stage.”

“All the big players understand that supporting an open standard is, in the end, beneficial to all of us,” confirms Philips’ Baas. “It stimulates competition in the market by avoiding lock-in to a certain vendor.”

While vendors go about implementing Supplement 145, one future user, at least, is reflecting more broadly on the importance of standards to the field of pathology. “It’s not just about DICOM,” Dr. Gilbertson says. “It’s about standards in general that I think pathology needs to think about actively. If we don’t have standards, and you have to, say, change your LIS, it’s extremely expensive and very hard to do. But if you have standards, it makes things a lot easier.”

Anne Ford is a writer in Evanston, Ill.