OBJECTIVE: To evaluate imprint cytology (IC) as the intraoperative pathological consultations for ovarian epithelial tumors (OET). METHOD: We reviewed ICs obtained from 354 consecutive surgical specimens of OET. Cytological specimens were classified into five categories. Final pathological diagnoses were made according to the WHO classification. We performed logistic regression analysis, calculated the limits among benign, borderline, and malignant lesions, and analyzed the diagnostic accuracy. We also made receiver operating characteristic (ROC) curves regarding IC. RESULTS: The accuracies to differentiate benign and malignant lesions were 87.1 and 83.6%, respectively. In contrast, that of borderline lesions was 30.0%. The areas under ROC curves to diagnose benign, and malignant lesions were 0.888 (P<0.05) and 0.951 (P<0.05), respectively, that meant IC was significantly useful for diagnosis of malignancy. CONCLUSIONS: IC applied to OET was proved to be practically useful in establishing an intraoperative diagnosis by ROC curves.
OBJECTIVE: To evaluate imprint cytology (IC) as the intraoperative pathological consultations for ovarian epithelial tumors (OET). METHOD: We reviewed ICs obtained from 354 consecutive surgical specimens of OET. Cytological specimens were classified into five categories. Final pathological diagnoses were made according to the WHO classification. We performed logistic regression analysis, calculated the limits among benign, borderline, and malignant lesions, and analyzed the diagnostic accuracy. We also made receiver operating characteristic (ROC) curves regarding IC. RESULTS: The accuracies to differentiate benign and malignant lesions were 87.1 and 83.6%, respectively. In contrast, that of borderline lesions was 30.0%. The areas under ROC curves to diagnose benign, and malignant lesions were 0.888 (P<0.05) and 0.951 (P<0.05), respectively, that meant IC was significantly useful for diagnosis of malignancy. CONCLUSIONS: IC applied to OET was proved to be practically useful in establishing an intraoperative diagnosis by ROC curves.