Software segments CT studies for routing to designated radiologists

Whole-body scans are a snap. Efficiently reading them is not.

Modern CTs scan the length of a patient in seconds, covering multiple body parts. They may also require multiple readings and billings. But the PACS typically assembles the data under a single order, which may have to be broken manually into segments--head, neck, chest, abdomen, pelvis, and legs--with individual studies sent to the appropriate radiologists. In those cases, staff must manually process the paperwork to bill out for the different exams.

Problems can arise down the road when the PACS attempts to prefetch a multiple-procedure study but can find no information regarding the orders.

DeJarnette Research Systems has developed a solution. The software-based dyseCT automatically segments CT scans according to body parts, routes the segments to designated radiologists, and sets the codes for billing. The company is continuing to evolve the system, most recently announcing Release 2.0, which was demonstrated at the 2004 RSNA meeting. It features an improved segmentation algorithm, specific procedures for neurological studies, and improved handling methods for different CT makes and models.

About a dozen institutions have so far implemented dyseCT. That is fewer than Wayne DeJarnette, president and founder of DeJarnette Research Systems, would like. But things are looking up.

"We are starting to see some momentum," he said. "People are starting to put it in their budgets."

DeJarnette sees dyseCT as a $23,000 labor and money saver. Work-study analyses performed at five clinical sites indicate that inefficiencies associated with typical multislice operations running in a PACS environment cost an institution about $150 per day per CT, about $50,000 per CT per year. DeJarnette's dyseCT promises to go a long way toward making involved departments more efficient.

The product modifies the DICOM modality work list, using a group procedure to replace the multiple procedures. It analyses the images and associates them with anatomical regions, then matches them to the appropriate orders.

DeJarnette relies on "bone moments" to split the body into segments. These moments, based on measurements of where bone density is greatest, provide profiles that match the different body segments. A body scan can be segmented into any number of parts, depending on the reading paradigm.

"Right now (without dyseCT) the neuro guys are getting all the images, even though they don't care about anything below the neck, just as the guys reading the lower part of the body are getting--but don't care about--the head," said Dejarnette, who founded the privately held company 20 years ago. "For radiologists who want to read by body part or segment, dyseCT offers a huge improvement over what they have now."

DeJarnette has tried to convince CT manufacturers to built dyseCT into their scanners, but to no avail. They all see the problem of dispatching the proper images as being the responsibility of the PACS, not the CT, he said.

Bending to industry opinion, DeJarnette has sought and found help from PACS vendors. Fuji and Amicas so far have helped place units in a couple of installations. The company has signed an agreement with SourceOne to distribute dyseCT, which has recently been listed among the products sold by Novation, the supply chain manager for VHA and the University HealthSystem Consortium.

Banking on the belief that the idea has potential outside CT, DeJarnette is developing an encore product. Company engineers are migrating the capabilities built into dyseCT to a version designed for MR scanners.