A Meta-analysis Of Ultrasonography, Computed Tomography And Magnetic Resonance Imaging For Distinguishing Benign And Malignant Tumor Of Ovarian Tumors

OBJECTIVE: TO perform a meta-analysis to compare the effectiveness of current ultrasonography (US),computed tomography(CT),and magnetic resonance imaging(MRI)in differentiation of malignant from benign ovarian tumors.METHODS: Through a MEDLINE and CNKI literature search(January 1991 to December 2007)articles with imaging-histopathologic correlation and date that allowed calculation of true-positive(TP), true-negative(TN), false-positive(FP),and false-negative(FN)were identified. The diagnostic performance of US, CT, and MR imaging in the differentiation between benign and malignant ovarian tumours were compared.RESULT: 52 articles were included, yielding 107 studies. Data on the accuracy of the different imaging modalities were analyzed and compared by constructing summary receiver-operating characteristic (SROC) curves. The pooled sensitivity and specificity and 95% confidence interval were 0.87(0.83～0.90),0.84(0.83～0.84) for US, and 0.86(0.78～0.92),0.88(0.82～0.92) for CT, and 0.93(0.90～0.95),0.89(0.86～0.90) for MRI. Summary receiver operating characteristic analysis demonstrated better discriminatory power for US and CT. The value of Q* point for MRI (0.91) was higher than US (0.88) and CT (0.87).CONCLUSION: MRI showed a trend toward better performance than US or CT in differentiation of malignant from benign. But US morphologic assessment still is the most important and common modality in detect ovarian cancer.