Dose reduction software that places each pixel in context can cut the radiation load of an angiogram by up to 50 percent, according to a study at Johns Hopkins Hospital.

By running ContextVision's GOPView iRV Plus adaptive filtration program, radiologists were able to produce virtually equivalent images of a pig's vascular anatomy using 40 to 50 percent of the radiation dose, according to Dr. Eleni Liapi, a professor of radiology at Johns Hopkins in Baltimore and lead investigator of the study.

She presented the results of the experiment, sponsored by the Stockholm, Sweden-based ContextVision, in December at RSNA 2009.

In the small qualitative experiment, two radiologists looked at sets of randomly drawn images from angiograms on a pig; one was produced with the standard dose (160-200mA), another with the software-enhanced low dose (63-125mA). (The study was blind, and the rads didn't know which image was taken at which dose.)

The results: the rads found the two sets of pictures to be nearly identical, and actually made out slightly fewer visual artifacts in the low dose set.

"The addition of the real-time adaptive filter in all low dose angiograms led to significant improvement in diagnostic acceptability without any decrease in contrast and visibility of the evaluated structures," Dr. Liapi noted in her report.

ADAPTIVE FILTRATION

Initially developed over 20 years ago by scientists in Sweden, the software works by figuring out the context of each pixel, such as if it's embedded in a curve or line. Then, using the process called adaptive filtration, it removes the noise for greater picture clarity.

"The critical thing is we do it on each frame separate from the next, and each pixel separate from the next," Donald Barry, product general manager of the X-ray division of ContextVision, explains to DOTmed News.

This frame-by-frame, pixel-by-pixel processing helps protect the image against the effects of temporal blurring without having the program resort to motion correction. In temporal blurring, movement can distort the picture, smearing, say, the edges of blood vessels during an interventional procedure.

"Most companies have some sort of motion correction for smearing, but going at higher and higher speeds, they're forced to turn this off," Barry says.

FUTURE STUDIES

Barry believes a larger, human study could be done by one of ContextVision's OEM customers. ContextVision, not in the business of selling directly to customers, works with OEMs to package its products unlabeled inside their machines.

"Many OEMs are interested in dose reduction, and my expectation is they would use this data for a human-based study," Barry says.

In the meantime, Barry is working on a project to compare image quality with the ContextVision enhancement software without the need to recruit people for a massive comparison study. In this experiment, a computer will randomly inject noise into existing radiographs, taken from humans, to mimic the effects of a low-dose image, which could then be corrected with the company's software. "If you know [that] the pattern you're putting in reflects a dose reduction of X percent, then you have a good surrogate of a low dose image," Barry says.

Read more and watch a video on DOTmed:

Johns Hopkins University Study Proves Low-dose Capability of ContextVision Interventional Image Enhancement Tool
https://www.dotmed.com/news/story/11437

RSNA Video Profile: ContextVision
https://www.dotmed.com/news/story/10957

ContextVision's GOPView CT Achieves Unprecedented Image Enhancement Speed
https://www.dotmed.com/news/story/10514


ContextVision Software Is the "Brains" Behind Many Modalities
https://www.dotmed.com/news/story/9457