Role Of Dynamic Contrast Enhanced Mr Imaging And Diffusion Weighted MR Imaging In Qualitative Diagnosis Of Ovarian Complex Tumors

Objective To evaluate the effectiveness of dynamic contrast enhanced MR imaging（DCE-MRI） and diffusion-weighted MR imaging （DW-MRI） in qualitative diagnosisof complex ovarian tumors.Materials and methods The MR data of34patients with complex ovarian tumors（age range14⁵3years, mean42years） confirmed by histopathology were analyzedretrospectively.12cases were benign complex tumors and22cases were malignantcomplex tumors （5borderline tumors and17invasive tumors）. All patients hadundergone conventional MR imaging, dynamic contrast-enhanced MR imaging anddiffusion weighted MR imaging on a GE Signa HDe1.5T scanner. Diagnosis ofbenign or malignant tumors were made by the conventional morphological criteriaand compared with the histopathological findings to obtain the sensitivity, specificity,positive predictive values （PPV） and negative predictive values （NPV） as well as thesemi-quantitative parameters of EA, T1/2max, MS of time-signal intensity curve （TIC）and the ADC and eADC measurements of diffusion-weighted imaging acquired fromthe tumor solid regions. The statistical significance was determined by Mann-WhitneyU tests. Receiver operating characteristic analyses were performed to determineoptimum thresholds for tumor qualitative diagnosis in that malignant tumors would bedifferentiated from benign tumors or invasive from borderline tumors, and thesensitivity, specificity, positive predictive values （PPV） and negative predictive values（NPV） were also calculated. Diagnostic tree was then made in an attempt to facilitatedifferential diagnosis in ovarian complex tumors.Results （1） Sensitivity, specificity, PPV, NPV and accuracy of conventional MR were90.9%,33.3%,71.4%,66.7%and70.6%, respectively, in determining malignantovarian tumors.（2） Statistically significant differences were noted in EA, T1/2max,MS, ADC and eADC between malignant and benign tumors （p<0.05）. The differencesin ADC, eADC and T1/2max were significant between borderline and invasive tumors （p<0.05） in spite of no significant differences in MS and EA （p>0.05）.（3） Themaximum Youden index was chosen to determine the optimum thresholds of all theparameters in ROC analyses to differentiate malignant from benign tumors. Statisticalanalysis showed a threshold value of260.38for EA to provide sensitivity, specificity,PPV, and NPV of72.7%,91.7%,94.1%, and64.7%, respectively. A threshold value of3.12%per second for MS provided sensitivity, specificity, PPV, and NPV of95.5%,83.3%,91.3%and90.9%, respectively. Using the same approach, when a cut-offvalue was determined as24.24seconds, T1/2max had a sensitivity of77.3%,specificity of91.7%, PPV of94.4%and NPV of68.8%. With a cut-off value of1.29×10^-3mm²/s for ADC, a sensitivity, specificity, PPV, and NPV of50%,91.7%,91.7%, and50%, respectively, for indication of malignant tumors. Moreover, athreshold value of eADC>0.28divided the two groups of tumors, yielding50%,91.7%,91.7%, and50%of sensitivity, specificity, PPV, and NPV.（4） The maximumYouden index was also chosen to determine the optimum thresholds of all theparameters in ROC analyses to differentiate invasive from borderline tumors. Athreshold value of ADC<1.64×10-3mm2/s provided sensitivity and specificity of100%and100%for prediction of invasive tumors and a threshold value of T1/2max<17.14s provided sensitivity and specificity of64.7%and100%.Conclusions The MS of TIC of the solid component is the most superior diagnosticperformance in distinguishing malignant from benign ovarian complex tumors. TheADC value of the solid component is optimal diagnostic performance indifferentiating invasive from borderline ovarian complex tumors. CombiningDCE-MRI and DW-MRI with conventional MR imaging increases the accuracy ofpreoperative imaging in qualitative diagnosis of ovarian complex tumors.