An automated system that uses red, blue and green coding channels from color fundus images may be able to accurately diagnosis glaucoma and reduce the need for traditional testing, a new study reports.

Using a technique called bit-plane slicing, researchers separated the red, green and blue channels from fundus photos and split the channels into bit planes. Doing so enables the automated system to better analyze the relative importance played by each bit (one pixel contains eight bits) of an image.

Investigators then extracted local binary pattern statistical features from each bit planes—in effect turning the fundus image into a computer code—and fed them into three learning algorithms, which classified the fundus images as normal or indicative of glaucoma.

“Our experimental results suggest that the proposed approach is effective in discriminating normal and glaucoma cases with an accuracy of 99.3% using 10-fold cross validation,” the researchers wrote in their paper.

The system is ready to be tested on large and diverse databases and has the potential to assist eye care practitioners in screenings to confirm and increase accuracy of their diagnosis, the investigators noted.