抄録
Although the achievement of satisfactory results with our newly devised diagnostic system using a fuzzy inference has been already reported, these results were based on subjective evaluation of the findings by only one physician. In order to determine the usefulness of this diagnostic system for general use, ultrasonic images from 50 cases of histologically confirmed thyroid nodules, consisting of 22 cases of papillary carcinoma and 28 of benign nodules, were evaluated by nine examiners; six physicians and three medical technicians with varying degrees of experience. Then their subjective diagnostic results were compared with those of our diagnostic system with a fuzzy inference. The distribution of the degrees of variation in the evaluation of the seven items measured using a fuzzy scale differed among the examiners for the items of shape, margin, boundary and internal echoes, but was relatively similar for the items of hyperechoic spots, cysts and echo level. The sensitivity of cancer diagnoses made with our diagnostic system with a fuzzy inference with input data from eight of the examiners was greater than that of their subjective diagnoses. The diagnostic results for the remaining examiner were the same. The mean diagnostic sensitivity using our system was 84.9%, while that of the diagnoses without the system was 71.7% (P<0.05). There was no statistically significant difference in the mean diagnostic accuracy due to the results of the specificity. We believe that this diagnostic system should be useful for any physician, but especially for novice physicians.