Visual Search Patterns of Radiologists in Training

At the present time, the information contained in a roentgenogram can be extracted only by a radiologist observing it. That this link can sometimes fail has recently been emphasized by Tuddenham (1) and Garland (2), who have reported frequent perceptual errors by radiologists viewing roentgenograms demonstrating fairly obvious lesions. Such errors have also been shown by Zeidner and Sadacca (3) and Enoch (4) in their studies of photointerpreters reading aerial photographs. There are two stages in the visual perception of detail. First, the object must be focused upon the fovea; second, this focused image must be translated into meaning by the observer's brain. The fovea subtends an angle of only 2°, so that at 30 inches the area of distinct vision is a circle about 1 inch in diameter. Thus, some 300 separate eye fixations are necessary to cover the area presented by a 14 × 17-inch roentgenogram at this distance. The eye searches the film by moving in small rapid jumps of high velocity (saccadic movements), stopping at intervals and remaining steady before making another jump. It is only during these steady intervals or fixations that we are able to see clearly, for while the eye is moving there is no perception of detail. During any visual task an observer makes many unconscious eye movements and fixations. The eyes receive and transmit to the brain a vast amount of information, several million “bits” a second, to use the language of communication theory, but the brain can make decisions at a rate equivalent to only about 25 bits a second. Therefore, some selective mechanism must exist controlling the information which rises to conscious levels. Experience and training must be among the factors affecting this selective mechanism, and knowledge of the methods of visual search adopted by the radiologist is necessary if the causes of his perceptual errors are to be discerned. Method The present study is based upon the initial visual search of five radiologists entering training, who viewed six roentgenograms and three projected Rorschach ink blots for thirty seconds each, while their eye movements were recorded by an optical eye-marker camera. The film records were then analyzed for search patterns, fixation times, and distribution of fixations. Rorschach slides were introduced both as a “control” presentation and as part of a larger study. Viewing was done at a distance of 42 inches because of equipment limitations, although Tuddenham and Enoch have pointed out that the ideal display for initial search subtends an angle of 9° at the eye, which would then place a conventional chest film 8 ft. from the viewer. “Normal” chest films were used in this first study because it was desired to examine the initial search patterns, and these would be disturbed if any abnormalities became apparent.