Value of Multiple Computed Tomography Criteria for Prediction of Malignancy in Patients with Ovarian Mass
Keywords:Computed tomography, Histopathology, Morphological characteristics, Ovarian cancer
Objective: Computed tomography (CT) can be used as a safe, accurate and noninvasive technique for the prediction of ovarian malignancy with several criteria. We aimed to determine the sensitivity and specificity of presence of multiple CT malignancy criteria for the prediction of ovarian malignancy in patients with ovarian mass.
Study Design: From a total of 734 patients diagnosed with an adnexal mass, ovarian mass was determined in 91 contrast-enhanced abdominopelvic CT images were examined for the presence of tumoral diameter (>50 mm), thick septa, wall thickness, solid component, contrast involvement, invasion, ascites, and bilaterality. The ratios of these parameters and the value of their combined use for the prediction of ovarian malignancy was assessed.
Results: Of the 91 patients included in the study, in patients with benign [in 66 (72.5%) patients] and malign [(in 25 (27.5%) patients] ovarian mass, the mean (range) ages were 43.5 (17-82) and 58.0 (34-88) years, respectively. A statistically significant relationship was determined between ovarian malignancy and all the CT criteria (p<0.05) except the tumoral diameter (>50 mm) and wall thickness (p>0.05). The ROC analysis revealed that with the presence of 3 or more criteria among the 8 CT criteria, the ovarian mass can be predicted as malignant at least with a sensitivity of 76% and specificity of 70%.
Conclusion: The presence of 3 or more parameters among the 8 selected CT criteria, the ovarian malignancy can be predicted at least with a sensitivity of 76% and specificity of 70% in patients with ovarian mass.
How to Cite
All the articles published in GORM are licensed with "Creative Commons Attribution 4.0 License (CC BY 4.0)". This license entitles all parties to copy, share and redistribute all the articles, data sets, figures and supplementary files published in this journal in data mining, search engines, web sites, blogs and other digital platforms under the condition of providing references.