From the time we were young children, many of us have enjoyed playing a game comparing two images to find the differences. We look back and forth between two images, seeking out the small, subtle changes. Does the zebra have an extra stripe? Is a chair leg missing?
This game is somewhat similar to the way many radiologists read a mammogram study. They position the current and prior images on monitors side by side and glance back and forth, looking for changes.
Comparing images this way may be a fun game, but it is a less-than-ideal way to read mammograms. Radiologists have to move their eyes and shift their focus repeatedly, searching for irregularities in images that can be full of textures and shadows. This method of reading mammograms can be subject to “change blindness,” which is the inability to detect changes in successive images.
However, there’s another way of viewing mammograms that may help radiologists see the differences they might otherwise not easily see: a method known as image shuffling.
Image shuffling overlays the previous and current mammogram study on the same screen, which allows radiologists to flip back and forth between them. The images “shuffle” without the radiologists having to shift their focus or gaze. When images are flipped in this manner, changes between the two images stand out prominently in the form of motion — something humans are very good at detecting. One small study indicated that image shuffling has the potential to improve both radiologists’ speed and accuracy in reading mammograms, compared with traditional side-by-side viewing.1
Find out more about image shuffling, a feature that’s part of IBM Watson Health’s Merge Unity™, in a new white paper
about this promising technology.