Nuclear medicine data pose problems for PACS

March 13, 2003

The interactive nature of nuclear medicine images makes it difficult to integrate even the simplest of these studies into a hospital PACS, according to a leading expert on medical image processing.Unlike static radiological image data sets, nuclear

The interactive nature of nuclear medicine images makes it difficult to integrate even the simplest of these studies into a hospital PACS, according to a leading expert on medical image processing.

Unlike static radiological image data sets, nuclear medicine studies often allow users to recompute image displays and parameters "on the fly." Trying to link such dynamic user-control capabilities into a PACS is a challenge, Prof. Anthony Todd-Pokropek told delegates at the EuroPACS meeting, held in September in Oulu, Finland.

Nuclear medicine protocols start with raw data acquisition from different sources, followed by an intermediate results processing stage before a final results screen is generated. This display may include 3D reconstructions, movies, text, and graphical overlays, all linked together and controlled by the user.

Nuclear medicine workstations are usually joined in a stand-alone network, said Todd-Pokropek, a professor of medical physics at University College London. The main issue then becomes setting up interoperability between these systems and the hospital PACS so that each can accept data from the other and perform required tasks correctly. One straightforward solution would be putting the nuclear medicine modules on the PACS workstations, but this is unlikely to work, he said.

"This would be a major task, given the fact that the software is often proprietary, uses nontandard data structures, and often relies on the details of the hardware configuration in which the system is running. It would be unrealistic to do this in a reasonable timescale, and the process of validation and getting approval of it would be horrendous," Todd-Pokropek said.

Similar difficulties may hamper efforts to set up more tele-nuclear medicine networks, which link remote applications workstations. Although such schemes appear attractive because relatively small amount of data are transferred, transmitting and receiving workstations still have to handle complex image-processing protocols.

"The first scenario is that the report is generated at the local center and then transmitted to the remote center for teleconsultation. The alternative is that you send raw data to the remote center and the report is then generated and sent back," Todd-Pokropek said. "In my experience with telemedicine systems, the second scenario is the most common and seems to be the most intractable."

DICOM hanging protocols could pave the way toward viewing modality-specific nuclear medicine results on generic PACS workstations. Current proposals assume a DICOM standard gray-scale display function and soft-copy presentation state, although adding the color needed by nuclear medicine workstations should not be difficult, he said. With the hanging protocols, users should be able to place any object from a desired location on a display and control screen order, transparency, movie speed, and links needed to synchronize different objects.

"The complex software needs to create output in such a form as to create copies of the intermediate data, rather than generating it on the fly," he said. "In other words, we actually need modifications on both sides of the bridge between the nuclear medicine network and the PACS network."

Tomi Kauppinen, Ph.D., a medical physicist in the departments of nuclear medicine and radiology at Helsinki University Central Hospital, agrees that dedicated workstations are best for viewing nuclear medicine data. While nuclear medicine images are linked directly to a PACS spanning the hospital district of Helsinki and Uusimaa (HUSpacs), raw data are transferred back to workstations with specific image-processing tools for reanalysis and comparison with new data.

Images are produced in five nuclear medicine divisions on two separate campuses, where 11 gamma cameras will be linked to the system by year-end. All data from the cameras are first transferred to an online database in DICOM NM format. Images from other modalities can also be transferred for the purposes of image fusion. The data are then archived together in the regional HUSpacs database, for later retrieval by the standard DICOM query/retrieve function.

"With the trend toward hybrid gamma cameras and devices, it is important to fuse anatomic and metabolic images together," he said. "PACS allows us to do online image fusion for treatment planning in therapeutic units, where, for example, CT and FDG-PET studies can be used simultaneously for targeting the treatment fields."