Online Journal
IF値: 0.677(2017年)→0.966(2018年)


Journal of Medical Ultrasonics

にて英文誌のFull textを閲覧することができます.


2002 - Vol.29

Vol.29 No.06

Original Article(原著)

(J545 - J555)


ROC Analysis for Objective Evaluation of the Diagnostic Accuracy of Ovarian Malignancy by Ultrasonography and Color Doppler

崔 華1, 竹内 久彌2, 福久 健二郎3

Hua CUI1, Hisaya TAKEUCHI2, Kenjiro FUKUHISA3

1順天堂大学浦安病院産婦人科, 2山王病院産婦人科, 3独立行政法人放射線医学総合研究所重粒子医科学センター

1Department of Obstetrics and Gynecology, Juntendo University, Urayasu Hospital, 2Department of Obstetrics and Gynecology, Sanno Hospital, 3Medical Information Processing Office, Reseach Center for Charged Particle Therapg, National Institute of Radiological Sciences

キーワード : blood-flow velocity, color Doppler, ovarian tumor, receiver operating characteristics analysis, ultrasonography

The diagnostic value of color Doppler in supplementing sonographic data is not clear. Although we reported on the usefulness of color Doppler in diagnosing ovarian malignant tumors in 2001, we suspect that comprehensive use of blood-flow data may be of greater diagnostic value than blood-flow impedance alone. Here we used receiver operating characteristics (ROC) analysis to evaluate the usefulness of a comprehensive color Doppler method combining B-mode and color Doppler to diagnose ovarian cancer. We studied 308 tumors of 282 patients who had histopathologically proven ovarian tumors and had undergone sonographic and color Doppler examination. Seventy-six of these patients had 82 malignant tumors (borderline cases included). ROC analysis of the 308 tumors showed that the areas under the receiver operating characteristics curves (AUC) for echo pattern classification, B-mode image interpretation, color Doppler, and combined B-mode and color Doppler were 0.808, 0.908, 0.9191, and 0.949, respectively. Best results were obtained with the combined method, which was significantly superior to echo pattern (p=0.001) and B-mode image (p=0.0081), although the combined method did not differ significantly from results obtained with color Doppler (p=0.0972). Color Doppler was significantly better than echo pattern (p=0.0005), however. ROC data were then used to analyze 106 tumors identified by color Doppler using resistance index (RI), peak systolic velocity (PSV), and acceleration time index (ATI). Only the resistance indexes of the malignant and benign tumors differed significantly (0.628, nonmalignant or benign. These results indicate that color Doppler has better diagnostic value than echo pattern or B-mode imaging and is effective for diagnosing malignant tumors. Although results obtained using the combined method did not differ significantly from those obtained using color Doppler, it showed the greatest area under the curve, indicating that it may be superior to color Doppler alone. Combined color Doppler and B-mode imaging rather than B-mode imaging alone, and full utilization of blood-flow data may thus enhance our ability to detect malignant ovarian tumors.