Most rads miss the gorilla imprinted on chest CT

The unusual image of a jungle animal randomly appearing in an otherwise familiar scene should cause a person to sit up and take notice. If that person is hyper-focused on another task, however, research has shown that a phenomenon called inattentional blindness is likely to make the person completely miss even such absurd scenes as a gorilla-suited man dancing across a basketball court or a chest CT.

A new study, conducted by researchers from Harvard Medical School in Boston, has shown that even radiologists can miss seeing a gorilla stamped on a diagnostic image. More than four-out-of-five radiologists did not see the image of a gorilla printed on a stack of chest CTs, according to a study slated for publication in Psychological Science.

The study is an adaptation of a famous test from the field of attention research. In that test, subjects view a video (see below) featuring a group of basketball players wearing either white or black shirts. Subjects are instructed to count the number of times a player wearing white passes the ball, and since there are multiple balls and the players are in constant motion, the subjects must concentrate on the task. Halfway through the video, a man in a gorilla suit nonchalantly walks through the action, but subjects are so lost in concentration that only about half notice the gorilla when asked after the video.

The current study, conducted by Trafton Drew, PhD, of Harvard Medical School, and colleagues, asked 24 radiologists to scroll through five lung CTs while looking for nodules. Each case contained an average of 10 nodules, but in the final trial, the authors inserted a 29x50mm image of a man in a gorilla suit that faded into and out of view over five slices of the image. The size of the gorilla was over 48 times the size of the average nodule in the images, explained Drew et al.

After viewing the images, the radiologists were asked if they saw anything unusual and, more bluntly, if they saw a gorilla on the final trial. Twenty of the 24 radiologists failed to see the gorilla, according to Drew and colleagues. Eye-tracking revealed that 12 of the radiologists who did not report seeing the gorilla looked right at it while it was visible.

To compare the results from expert radiologists with untrained viewers, a group of 24 naïve observers were also tested. They were given basic training in how to search for a nodule and completed the same task as the radiologists. None of the untrained observers detected the gorilla in the final trial.

Both the radiologists and the naïve observers were able to easily locate the gorilla when they were again shown the CT image where it was imprinted, according to the researchers, adding it was reassuring that the experts performed better than the untrained observers. This was likely due to the fact that the radiologists’ attentional capacity was less completely occupied because the task was familiar.

While less than 20 percent of the radiologists spotted the gorilla, the radiologists were much better at spotting the nodules, achieving a 55 percent detection rate on what the authors described as a challenging task, even for expert radiologists.