The influence of prior knowledge on visual search strategies during the viewing of chest radiographs.

Abstract

THE PROCEDURE of “reading” a chest radiograph includes the elements of visual search, detection, and interpretation. These elements are interrelated and occur simultaneously although it is obvious that the procedure begins with visual search and ends with interpretation. In the most general terms, visual search is the procedure used for sampling the complicated visual environment which surrounds the observer (2). Sampling is necessary because of the overwhelming amount of information reaching the retina at any instant (6). In addition, the retina does not have a uniform distribution of spatial resolution (5), and therefore part of the sampling procedure consists of bringing areas of greatest interest into the region of highest spatial resolution, the fovea. The sampling is accomplished by movements of the head-eye system. The eyes move in rapid jumps or saccades with brief stops or fixations. Sensations are produced in the retina only during the initial part of fixation (13). The observer is not generally aware of movement of the eyes or of the intermittent sensations arriving centrally from the retina. The world as perceived is continuous and complete but still has a definite center and periphery depending upon the direction of gaze. It is important to make a distinction between the visual search process and the eye movement pattern. Visual search can be conscious although it is usually subservient to other higher processes. It consists of directing the attention to a particular part of a complex visual scene. For example, attention can be directed sequentially to the heart, lungs, mediastinum, ribs, etc. The eye movements are then superimposed on this attention pattern but do not follow it exactly (9). The eye movements are not under direct conscious control but are entrained by the search process and only reflect in a general way the order and progress of search. The locus of fixations indicates the parts of the visual environment considered important enough to be sampled by the fovea, and the sequence of fixations indicates the order in which parts of the scene were sampled. Very little is known about the search process in radiology. Textbooks (8) admonish students to develop a systematic method of search, and many radiology residents are similarly instructed by their teachers. It is not clear if systematic search is generally used by practicing radiologists, or if its use can improve the detection process. In fact, there is some evidence to the contrary (11). It is also not clear if there are any consistent search patterns at all within an individual's behavior. Previous studies of eye movement during the viewing of chest radiographs have shown that there are no simple search patterns (3). This is not surprising since the final pattern is the result of a complex interaction between search, detection, and interpretation.