Study validates irreversible compression for clinical US

Many vendors support compressed image transfer syntax as specified in the DICOM standard, making it possible to both store and distribute irreversibly compressed clinical images. The DICOM standard doesn't address the amount of compression acceptable to users, however, so researchers at the Mayo Clinic in Rochester, MN, are conducting a study to determine what levels of compression their facility would tolerate.

The Mayo Clinic considered it important to conduct the study (J Digit Imaging 2002;15(1):15-21) before including irreversible compression in the imaging chain for its ultrasound practice.

"We found that color and gray-scale images compressed with JPEG at approximately 9:1 were indistinguishable from the originals," said Kenneth Persons, lead analyst programmer for Mayo's Imaging Information Systems Section.

Mayo has been using irreversible JPEG compression prior to interpretation on all of their ultrasound images for two years. (Mayo prefers the term "irreversible compression" instead of "lossy compression" because "lossy" implies degradation, and no evidence of degradation was found in their study. )

"There have been no problems or concerns of any sort raised by our staff regarding image quality of the compressed data sets," Persons said.

The use of irreversible compression enables Mayo to maintain all of its ultrasound data online, contributing to the efficiency of the practice and improving patient care.

"Irreversible compression is not something to be afraid of, but its characteristics should be well understood before implementing it in your practice," Persons said. "It can help make practical the online storage and distribution of the numerous large data sets produced by a medical imaging department."

Image quality is the primary consideration, and a compression technique and conservative compression level that does not compromise image quality should be chosen. Mayo recommends that other practices perform their own study to determine the proper compression level, since each facility is different, and one size may not fit all.

"Conducting such a study will also demonstrate for the radiologists that the resultant image quality will be diagnostically acceptable," Persons said. "Our study provides a guideline for what levels of compression might be tried."

The effects of compression on other imaging modality types is being examined at Mayo. A current study is evaluating the use of irreversible JPEG and JPEG2000 compression to determine the best algorithm and an acceptable conservative compression level for MR and CT images.