CAD companies focus on productivity to improve radiology workflow

December 20, 2004

To gain acceptance of their offerings by the radiology community, computer-aided detection vendors have focused their sales pitch on diagnostics, stressing the ability of their systems to increase radiologists’ confidence in finding subtle areas of potential malignancy. An equally important selling point is efficiency. In the face of a shortage of radiologists and an explosion in the number of images demanding review, CAD suppliers at the RSNA meeting focused attention on the time- and work-saving features of their products.

To gain acceptance of their offerings by the radiology community, computer-aided detection vendors have focused their sales pitch on diagnostics, stressing the ability of their systems to increase radiologists' confidence in finding subtle areas of potential malignancy. An equally important selling point is efficiency. In the face of a shortage of radiologists and an explosion in the number of images demanding review, CAD suppliers at the RSNA meeting focused attention on the time- and work-saving features of their products.

As one indication of CAD's effect on productivity, Confirma reported that its CADstream software could substantially reduce the time physicians spend interpreting breast MRs and technologists spend preparing the images and doing the calculations. CADstream's core and advanced automated features simplify both the preparation and reading of breast MR studies, said Meghan Mitchell, market manager.

With CADstream software, DICOM data can be processed in 10 to 15 minutes compared with about 30 minutes manually. Interpretation takes just a fraction of the 45 minutes or more that radiologists may need to interpret images without help from software, according to the company. The more complex the information, the more time savings may be possible compared with manual processes, Confirma claims. The actual time will depend on the size of the study and how much information is being processed; whether images are registered and reformatted in multiple planes, for example, or whether angiogenesis maps or interactive real-time curves are calculated.

CADstream's angiogenesis maps streamline a particularly time-consuming exercise for technologists by allowing them to categorize the uptake and washout of contrast material as rapid, plateauing, or persistent, merely by clicking on a pixel.

The lung nodule detection software from R2 Technology, which is available on the company's stand-alone workstation or as an integral part of PACS developed by Sectra and Kodak, offers such labor-saving tools as line measurement and temporal comparison. By automatically segmenting lung nodules, ImageChecker CT CAD software spares radiologists the need to take caliper measurements of every lung nodule found on CT. The software also takes a volume measurement, which is more accurate than a linear calculation.

According to findings from an R2-sponsored paper presented at the RSNA meeting, the variability of volume measurement of phantom nodules was less than 2% when the company's software was used. And 3D views of lung nodules corroborate lung nodule measurements by showing whether an abnormality abuts or intersects a blood vessel. Temporal comparison, which registers previous and current CT examinations, precisely locates the position of specific lung nodules and computes their change in size over time.

"To determine how quickly nodules are growing, the radiologist typically had to zoom up, take a forceps tool, measure one nodule, measure another, write the sizes down on a piece of paper, and bring out a calculator to determine tumor-doubling time," said Dr. Kevin A. Kreeger, vice president of advanced development for R2 Technology. "Our software automatically steps through all the nodules and determines whether they grew in size or not. In one-tenth of a second, radiologists have done what used to take them 30 seconds for each nodule."

A CAD product for assessing lung filling defects, due from R2 next spring, will take it a step further, automatically processing any chest CT case for suspected pulmonary embolism. The software will compute the diameter of an artery and the percentage of blockage. It will also include 3D views to distinguish abnormalities that reside in arteries from those in lymph nodes and trace the path of a filling defect across arterial bifurcations.

"CAD-driven review is not just reviewing scans with a CAD mark," Kreeger said. "It is using CAD information to drive, organize, and control workflow and increase radiologists' efficiency."