Elicitation of Diagnosis Strategy During Scanning Chest X-Rays from Eye Tracking Stimulated Retrospections

Abstract

In the present paper, we apply a debriefing technique aided by eye movement recordings to elicitation of clinical reasoning processes during chest x-rays diagnosis. It can be expected that this technique allows us to obtain elaborated information regarding hidden cognitive processes compared to conventional verbal protocol approach. Two experiments where medical tasks of 20 medical doctors (10 majoring in radiology/ respiratology and 10 majoring in surgery/acute medicine) on diagnosis of chest x-rays images were recorded with a video camera and an eye tracking system were performed. In the first experiment, each one of 5 chest x-rays having four patterns of cancer, pneumonia, normal and others were shown. A participant was asked to make his/her diagnosis decisions about whether each of chest x-rays. Immediately afterwards, a debriefing where each eye tracking recording was used as a cue to verbalize the participant’s implicit diagnosis processes was conducted. In the second experiment, a comparative diagnosis on the current patient’s status was carried out. Five pairs of x-rays images were shown to each participant. In each pair, one was a current image and the other one was an image taken a year ago. The participant examined the current patient’s status by identifying small changes with time. The debriefing stimulated by eye movement recordings was performed just after the task. Based on the verbal protocols form the debriefing session, each participant’s reasoning processes were traced. The results indicated that there were mainly four effective reasoning strategies as followings: Postponement of lung field interpretation, avoidance of preconception by applying redundant scanning rule, critical area revisited, and complying with one’s mental check-list at any moment. At the same time, one problematic strategy was also identified that can be named as “single lesion focusing strategy.” Based on all results as well as implications obtained, we discuss insights relating to effective medical reasoning processes as well as validity of verbal protocols/comments on eye mark recordings.