Advanced visualization stumbles in today’s PACS environment, demo suggests

May 16, 2008

Although it offers tremendous opportunities for advances in imaging interpretation and communication with referring clinicians, advanced visualization still stumbles when it comes to workflow, according to a presentation Friday.

Although it offers tremendous opportunities for advances in imaging interpretation and communication with referring clinicians, advanced visualization still stumbles when it comes to workflow, according to a presentation Friday.

Using skits and a simulated situation, a team presented the way it works now - with paper shuffling, phone calls, miscues, and lots of time - to get an advanced visualization interpretation to a surgeon. They then repeated the process a few minutes later with PACS and RIS advances that are not yet on the market to show how much better it could be.

The key point: Advanced processing and visualization must be integrated into clinical practice. To date, implementation and use of many postprocessing applications is suboptimal in many institutions.

The task examined was a CT scan of a renal donor. Problems began cropping up from the start. A written request from the referring surgeon was too long to fit into the computerized field read by the technologist, so essential data, like the need for a volume rendering, was left off when the request was put in the RIS. That necessitated repeated calls and extra steps.

Then the scan data were stored in the PACS in a thick-slice format. When the radiologist requested the thin-slice data for his own reconstruction in order to look for a possible accessory renal artery, the information had to be retrieved again. But the tech didn't know which workstation to send it to, so she sent it to all of them, bogging down the network.

When the radiologist finally got the thin-slice data and did the reconstruction, his suspicion of an accessory renal artery was confirmed. He called the referring clinician to explain, but the clinician wanted to see the image; couldn't the radiologist send it? No, the system was not set up for that, the radiologist explained, just come down and look at it. The clinician agreed to send a resident.

Throughout there were multiple trips to the printer to retrieve worklists and protocols, searches for yellow sticky notes with log-on information, telephone calls to repeat (and once misstate) the patient's name and number, long passages of time, and general consternation that the workflow process was so chaotic.

Among the problems were a lack of automation and integration with other relevant systems and no orchestration of tasks. In addition, many systems are stand-alone, so require separate logins, separate databases, and separate archives; plus, they have no worklists. This also makes billing more complex and work more difficult to perform, due to multiple applications and GUIs. Network traffic is increased, and communication among the various players is poor.

The skit did not represent a specific hospital, but the problems observed were by no means uncommon, said Katherine Andriole, Ph.D., of Brigham & Women's Hospital and Harvard University, who played the technologist.

She and the other presenters, Dr. Paul Chang of the University of Chicago Pritzker School of Medicine, who played the radiologist, and Dr. Luciano M. Prevedello of Brigham & Women's Hospital, who played the referring surgeon and the tech in the 3D lab, argued that today's systems negate many of the benefits of multislice CT scanning and advanced postprocessing capabilities.

One approach is a system that stores thick slices from the CT in PACS, Andriole said. This allows a 10-fold reduction in storage requirements, but wipes out much of the value of near-isotropic CT scans and prevents postprocessing at a later date.

A midlevel solution, and the one most commonly used today, is to store key images from thin-slice data in PACS. But with this approach, interactive processing and visualization is not possible and the exam can't be postprocessed.

Better is a true 3D thin-slice PACS model, which requires tighter integration of advanced processing capabilities with PACS, the presenters argued. Archiving of thin-slice data can be phased in incrementally, but ultimately all the data will need to be archived on the PACS for future postprocessing, they said.

Several vendors currently provide advanced processing directly at the interpretation display via a central server approach. But not all functions offered by stand-alone workstations are available in the embedded solutions.

With a thin-slice system and better, tighter integration and orchestration, the visibility of new technologies to referring clinicians and to patients can improve relationships and customer satisfaction, Andriole said. Faster demonstration and better communication of results via advanced presentations can be a benefit as well.

Other potential benefits include more efficient radiologists who may be able to increase their reading volumes. New exams or services are possible.