片桐 誠¹, 有田 清三郎², 福林 雅裕³, 山根 康彦¹, 原田 種一¹

Makoto KATAGIRI¹, Seizaburou ARITA², Masahiro FUKUBAYASHI³, Yasuhiko YAMANE¹, Tanekazu HARADA¹

¹川崎医科大学内分泌・甲状腺外科, ²川崎医科大学数学教室, ³川崎医科大学中央検査部

¹Division of Endocrine Surgery, Kawasaki Medical School, ²Department of Mathematics, Kawasaki Medical School, ³Department of Clinical Laboratories, Kawasaki Medical School

キーワード : Ultrasonic diagnosis, Fuzzy scale, Fuzzy inference, Thyroid cancer

A newly developed system for ultrasonic differential diagnosis of thyroid carcinoma was devised. The system is based upon a "fuzzy scale" by which the degree of variation in findings of seven items, i. e. shape, margin, boundary, internal echoes, hyperechoic spots, cystic patterns and echo level, is evaluated, and upon a fuzzy inference using membership functions. Using this method, the diagnostic accuracy of the images obtained by a 7.5 MHz probe in 90 cases, consisting of 31 cases of papillary carcinoma and 59 of benign nodules, was compared with that obtained by a 5.0 MHz probe in 102 cases, comprising 43 casesof papillary car cinoma and 59 of benign nodules. The sensitivity rate for cancer diagnosis with the images obtained by the 7.5 MHz probe was 96.8%, and the accuracy rate was 81.1%. On the other hand, the sensitivity rate with the images obtained by the 5.0 MHz probe was 55.8%, and the accuracy rate was 59.8%, both worse than for the 7.5 MHz probe (pClearer pictures were obtained by the 7.5 MHz probe than by the 5.0 MHz probe, enabling us to detect important factors, which led to the differences in results.