Virtual Colonography + CAD in Practice

	Siemens Medical Solutions syngo Colonography PEV detects polyp-shaped objects between 6mm and 25mm.

As U.S. sites ramp up virtual colonography programs, experienced providers weigh in on software and CAD.

With reimbursement for virtual colonoscopy on the horizon and clinical studies proving the technology’s diagnostic capabilities, facilities are beginning to gear up for the procedure. A virtual colonoscopy program requires a few essential investments: a 16- or 64-slice CT scanner, CO2 insufflator, post-processing software, and a review station. Increasingly, CAD is part of the formula for success. Even experienced readers can miss clinically significant lesions, says Seong Ho Park, MD, assistant professor of radiology at University of Ulsan College of Medicine Seoul in South Korea. Colon CAD software delivers sensitivity comparable to double-reading, or in the case of a single-reader model, CAD improves sensitivity 10 to 15 percent without adding significant time to the review, says Denis Marion, MD, a gastroenterologist at Hotel dieu Hospital in Lyon, France.

Park taps into Siemens Medical Solutions Polyp Enhanced Viewing (PEV), in conjunction with Infinitt Xelis Colon software, to minimize human errors and increase diagnostic sensitivity. “Quality of CAD, or diagnostic accuracy, is the foremost feature to be considered when choosing a CAD system,” says Park. The downside of the best-of-breed approach is a less than optimal integration. The University uses Xelis software for interpretation, which translates into separate workstations for initial review and CAD. After initial image review at the virtual colonoscopy workstation, Park turns to the CAD workstation for review of CAD marks. Robust 3D and CAD processing help Park overcome the hurdles associated with the dual-work-station approach. PEV false positives are low, averaging five to six per patient, resulting in a minimal increase in interpretation time.

Hotel dieu Hospital uses a different approach to its virtual colonoscopy program, tapping into Median Technologies LMS-Colon for 2D and 3D interpretation and CAD. After acquisition, images are automatically sent to the Median server for processing. Radiologists review images on a dedicated workstation or PACS review station.

Median software offers concurrent and second-reader review modes. In the concurrent mode, CAD marks appear with the initial 2D and 3D displays, and the user simultaneously interprets the datasets and confirms or rejects CAD marks. The second-reader model presents CAD marks after initial review by the radiologist, only marking lesions not already detected.

The University of Arizona Health Sciences Center in Tucson, Ariz., uses TeraRecon’s Aquarius Workstation and Philips Healthcare ViewForum to complete virtual colonography studies. Each system has advantages that translate into time-savings, says Arnold Friedman, MD, chief of body imaging. Aquarius offers an intuitive user interface, which can streamline the interpretation process, while Philips’ unfolded view saves time by limiting the exam to a single trip from the rectum to secum.


Questions to consider

The virtual colonoscopy market continues to evolve as vendors push to eliminate the stand-alone workstation model by integrating colonography software and CAD on the PACS workstation. As vendors focus on integration, practices need to consider other issues in the evaluation and implementation process.

• Is the reporting feature robust? Does it readily import images into the report?
• Assess each system’s integration capabilities. Can the software be integrated with the HIS? Does it send DICOM reports to PACS? Can referring physicians access reports and key images in PACS?
• Does the single model approach provide the diagnostic accuracy needed in a CAD paradigm? Does it sacrifice software functionality?
• Does CAD processing interrupt human interpretation workflow? Are CAD results ready when the radiologist finishes initial review?
• Does simultaneous CAD processing interfere with speed by limiting hardware resources available to support 2D/3D interpretation?

With reimbursement edging toward reality, radiology practices that analyze and understand software options will have an advantage that could deliver early and long-term referrals. Do the homework, consider the math.