Imaging Software May Aid Medical Diagnoses

Originally Published MDDI October 2003

R&D DIGEST

Gregg Nighswonger

Mammograms, x-rays, and other imaging modalities provide important tools for identifying disease. But their utility is severely limited if warning signs are too subtle for detection by the unaided eye. Indications of abnormality may also elude detection if they are too complex for interpretation with computer-based systems. 

Now, scientists at the U.S. Department of Energy's (DOE's) Idaho National Engineering and Environmental Laboratory (INEEL; Idaho Falls, ID) have developed a system that may increase the diagnostic utility of digital imaging methods. Called the Change Detection System (CDS), the technology is capable of highlighting slight differences between digital images. In a rather novel demonstration of the system's usefulness, lead researcher Greg Lancaster used it to compare scans of his own brain after he'd had a tumor removed, convincing his doctors of the program's effectiveness. 
The new software-based system is a direct result of applying national security technology, which was initiated through funding from the DOE's Applied Technology Program. One medical technology firm is already looking to license the program. 

Medical diagnoses often involve making a side-by-side image comparison to discern subtle changes. But identifying minute differences between two pictures can be nearly impossible. 

One method of comparison has been the flip-flop technique, in which rapidly alternating two similar digital images on a screen creates an animation effect. While identical elements seem stationary, differences appear as movement. But this approach has been found to be severely limited by the need for a stationary camera. 

The CDS technology developed by INEEL's Lancaster, James Litton Jones, and Gordon Lassahn combines the advantages of computer analysis with the powerful human reflex elicited by the flip-flop technique. 

The CDS program aligns images, to within a fraction of a pixel, from hand-held or otherwise imprecise cameras. The alignment compensates for any differences in camera angle, height, zoom, or other distractions. 

Flipping between two seemingly identical images aligned by CDS reveals otherwise imperceptible differences. The alignment process takes only seconds and the software is simple enough to be operated by a 10-year-old child, the researchers say. The compact nature of the program enables it to operate on a standard PC or a handheld computer. 

The researchers believe the program has a range of potential applications— from research and security use to medical practice. Despite the technology's origin in national security, Lancaster first considered medical applications of the technology as a result of his personal experience during research for the CDS project. 

After his physicians removed the brain tumor, they monitored Lancaster's brain with twice-yearly MRI scans to make sure the tumor didn't return. He watched his physicians review each image, searching for the tiniest change. As Lancaster watched the process, he worried that they might miss something. 

“They just stared at them to try to find differences,” Lancaster explains. “I said, ‘Man, that's so archaic.'” Lancaster determined that he would demonstrate the new system to his doctors and test their powers of perception as they reviewed images with and without CDS. 

“I took an image, altered it ever so slightly, brought in both pictures and said, ‘Can you see a difference?' They looked at the two images and admitted, ‘Well, no,'” Lancaster says. “But with the [CDS] method, it really pops out. They said, ‘Wow! What a tool!'” 

Copyright ©2003 Medical Device & Diagnostic Industry