News Release

In this digital age, compressing medical images can put the squeeze on diagnostic accuracy, UF researchers find

Peer-Reviewed Publication

University of Florida

GAINESVILLE, Fla.---A picture's worth a thousand words, but when it comes to digital medical images, one of those words should be "caution."

That's the message from researchers at the University of Florida, who have found that even mildly compressing coronary angiograms for delivery over telephone or computer lines can lead to diagnostic errors.

At issue is the increasingly common practice of storing medical images such as angiograms, X-ray films and CT or magnetic resonance scans on compact discs and other portable media. More and more, patients present CDS to specialists and surgeons when seeking second opinions. Doctors use that same technology to transmit pictures directly to their colleagues at hospitals or outlying clinics, speeding the process of examining key patient information.

But because the images contain so much data, storing them or sending them electronically requires mathematically manipulating them to shrink their size. Doing so quickens transmission -- in some cases cutting hours off delivery -- and saves money.

Compressing the data helps, but raises its own set of problems: Many times the file can only be reconstructed to reproduce a close approximation of the original, not an exact replica. Depending on the degree of compression, studies are finding, the images can be so distorted that experienced clinicians begin to misinterpret what they show.

"Data compression helps with costs and with speed of transmission -- that's the upside," said Dr. Richard Kerensky, an associate professor of medicine in the division of cardiovascular medicine at UF's College of Medicine and medical director of the cardiac catheterization laboratory at Shands at UF medical center. "The downside is you might start to degrade the image the more you compress it. As you change it, you lose information and add artifacts that weren't present on the original image."

Kerensky and his colleagues recently focused their attention on the issue, publishing findings from one of three related studies involving an international team of scientists in the April issue of the Journal of the American College of Cardiology. Also involved in the work, which dealt solely with the series of moving X-ray pictures known as coronary angiograms, were the Cleveland Clinic Foundation, the Mayo Clinic, Duke University, the University of Pennsylvania and the University of Kiel in Kiel, Germany. Study authors included Dr. Carl J. Pepine, chief of cardiovascular medicine at UF's College of Medicine, who was the first to call for standardizing the exchange of data in the early 1990s, while head of the American College of Cardiology's cardiac catheterization committee.

The goal: to set guidelines for cardiologists to follow during this time of technological transition. After all, most experts agree it's just a matter of time before digital imaging completely replaces film or videotape. The recommendations will be added to the Digital Imaging and Communications Standard for Medicine, which the international cardiology community, led by the ACC, established to standardize the digital file format and image exchange medium for angiography.

The idea is to ensure that doctors can reconstruct a digital file to view a true representation of the original exam, said study co-author Jack Cusma, a consultant in the division of cardiovascular diseases at the Mayo Clinic in Rochester, Minn., and an associate professor in the Mayo Medical School. "That seems logical, but it's tougher to do than it sounds," he said. The trio of studies arrived at the same conclusion: Image quality was progressively lost with increasing compression. In the UF study, 100 different angiograms were compressed at varying degrees. Researchers evaluated observers' ability to detect subtle diagnostic features in the angiograms and discovered slight changes were introduced when the images were reduced 10:1.

At 16:1, clinicians interpreting the images began to make mistakes; they were less likely to detect common diagnostic features, including arterial narrowings or calcifications and the presence of stents, wire coils designed to prevent arteries from closing. The potential impact, UF researchers say, is that clinicians would be less confident in their clinical decision-making and therefore less likely to take decisive action in terms of treatment.

Although highly technical, the discussion should be somewhat familiar to anyone who has transmitted or received photographs through the Internet. The recent studies focused on a compression method originally developed for still photography -- JPEG, for Joint Photographic Experts Group. JPEG reduces photographic file sizes so they can be easily transmitted over telephone and computer lines. The approach commonly is used outside medicine in products such as digital cameras and to send photographs for reproduction in newspapers or magazines, or on the Internet.

"Chances are every picture you look at in the newspaper today was transmitted as a JPEG image over a computer network. Everyone is getting very accustomed to the widespread availability of digital information," Cusma said. "We're able to get all this information over the Internet, and now a lot of people think once we have diagnostic images in digital format we should be able to do the same thing. Then they find out that it takes 60 minutes over a 'fast' connection to more than eight hours to reproduce the entire original exam without compromising quality. That's a lot more difficult task then just sending over a Web page from USA Today."

But time is only one of the worries. The primary concern is image quality: Is the end product of high enough resolution for a doctor to make a diagnosis or advise a patient about treatment? What's more, there are costs and risks associated with having patients undergo repeat angiograms if they arrive for a consultation without a usable image.

For one, exposing patients to unnecessary additional radiation wouldn't be ideal, Cusma said. A single chest X-ray is well below the annual suggested limit for radiation exposure, but repeated angiograms -- each equivalent to about 1,000 chest X-rays' worth of exposure - would boost the total dose the patient receives to much higher levels, he said.

Based on the latest findings, the standards committee has decided not to permit JPEG compression for original image recording and archiving, and while it has not yet precluded data compression for network transfer, it most likely will not be allowed. Other compression approaches are under development that may someday yield better performance. But for now, "we couldn't take that risk," Kerensky said.

"You can never be sure if a surgeon or another doctor might not see some subtle characteristic that they really need to see," he said.


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.