Incorporation of image processing presents next challenge for PACS

Basic technological and organizational problems in PACS have been solved, and it is time for the imaging community to incorporate image processing technology, an advocate said Friday.

Prof. Erwin Bellon of the University Hospitals Leuven in Belgium made the case for bringing image processing techniques, which truly separate soft copy from film, into the digital medical organization. In a few years, this approach will be standard, according to Bellon.

"PACS can move image processing from the labs to clinical usefulness," he said.

For that to happen, it must be readily available for use on real cases so that it can be understood in context.

How to get there, however, remains a conundrum, Bellon said. Most PACS vendors will include a technology if they believe it is useful and there is a demand for it. But proof of the value of image processing won't occur unless users find a way to make it part of their clinical work.

Only real experience using the new processing strategies, albeit in controlled circumstances such as a university setting, can let physicians learn what may be relevant when they interpret an image, Bellon said. This must happen at the physician's reading station and not in an imaging laboratory.

Right now, for example, processing protocols such as multimodality registration of images for oncology and radiotherapy planning and statistical processing for functional MRI are being used clinically at Leuven, but they are too time-intensive to be part of the routine, Bellon said. PACS vendors could make them much easier to invoke, but they will do so only after these protocols have been validated.

While image processing is not yet widely available through PACS, users can adopt strategies that allow them to test these approaches at the workstation, Bellon said.

"Plug-ins," often provided by third-party vendors, can be adapted to feed image processing information into the PACS workstation, using the viewer to present processed imaging data and avoiding the need to load imaging data separately, Bellon said. One such system at Leuven presents PET information from a PET/CT scanner as a color overlay on a corresponding CT slice.

It is also possible to set up "back-end" systems that process and store images in the PACS or electronic medical record for presentation later.

A third approach is to synchronize applications and automatically develop presentations for simultaneous viewing of the images and other data. Leuven researchers developed such a system to assess the clinical relevance of a method for automated extraction of the cardiothoracic ratio. Information assessed was drawn from both the PACS and the RIS.

All of this will work better with the cooperation of the vendors, according to Bellon.

"As a community we should encourage vendors to provide openness," he said.