Debate over lossy compression renewed at SCAR 2000

By Erik L. Ridley, AuntMinnie staff writer

June 5, 2000 --

PHILADELPHIA - The use of lossy compression in medical imaging and PACS applications can reduce image transmission times and storage requirements. But clinical concerns over the diagnostic quality of compressed images, and potential medical-legal issues, continue to limit its popularity.

The pros and cons of lossy compression were discussed in a point-counterpoint debate June 4 at the Symposium for Computer Applications in Radiology. The debate was moderated by Dr. Bradley Erickson of the Mayo Clinic in Rochester, MN.

The degradation of images with lossy compression has been overstated, according to Dr. Heber MacMahon of the University of Chicago Hospital. Recent research found no difference in diagnostic accuracy at reasonable compression ratios, and radiologists have even been found to prefer images compressed at 8:1, likely due to the noise reduction achieved from the compression process, MacMahon said.

"Even if you take a conservative [approach], such as around 10:1 compression, there are very few observers who can distinguish [a compressed image] from the original, even when comparing them carefully in a side-by-side situation," MacMahon said. "We feel that if the observer cannot distinguish between the images from the original, there is no reason to assume that there will be any loss in diagnostic accuracy."

But Dr. Don Schomer of the M.D. Anderson Cancer Center in Houston believes that scientific validation of lossy compression of medical images has not been achieved.

"Compression validation studies are very difficult and very expensive to perform, and are generally conducted on one image format or one type of imaging finding," Schomer said. "It's very difficult to generalize these findings across the entire radiology imaging domain. And validation of one algorithm is not necessarily generalizable across the entire spectrum of algorithms."

And while no lawsuits involving lossy image compression have occurred to date, they are almost certain to occur in the future, Schomer said.

"Plaintiff's lawyers are very creative," he said. "Would you like to be on the other side of this argument: 'Doctor, my client is now brain damaged because you couldn't wait a few extra moments to download the full-resolution image?' It doesn't matter if both images are equivalent. A skilled litigator is going to take you to the bank on that one."

Proponents of lossy compression believe that such risk is negligible. De facto forms of "compression" already occur routinely in clinical practice, such as exam rejection, selective image retention, and streamlining of imaging protocols, according to Dr. Thurman Gillespy of the University of Washington in Seattle. In addition, the potential loss of image quality is small in relation to the degradation that occurs as part of the normal image acquisition process, he said.

Further complicating the matter is the exponential increase in local area network bandwidth (via ATM and gigabit Ethernet), the proliferation of digital subscriber lines (DSL) and cable modems for wide area networks, and the dramatic reductions in storage costs achieved in the last few years. For example, storage costs are approximately 50 times cheaper than 10 years ago, and network bandwidth has also dropped significantly in price, Schomer said.

"Bandwidth is essentially going away as a problem," he said.

These improvements are largely being counteracted, however, by dramatic increases in image data and volume, Gillespy said.

Dr. John Carrino of the Brigham and Women's Hospital in Boston, an opponent of lossy compression, believes that progressive image transmission approaches, such as "just-in-time" protocols like South San Franciso, CA-based Stentor's dynamic transfer syntax (DTS), offer a better alternative to irreversible compression.

By Erik L. Ridley staff writer
June 5, 2000

Let know what you think about this story.

Copyright © 2000


To read this and get access to all of the exclusive content on create a free account or sign-in now.

Member Sign In:
MemberID or Email Address:  
Do you have a password?
No, I want a free membership.
Yes, I have a password:  
Forgot your password?
Sign in using your social networking account:
Sign in using your social networking