Pixel-based quantitative evaluation of CT values in differentiating cystic-solid ovarian tumors
10.3969/j.issn.1002-1671.2017.09.015
- VernacularTitle:按像素量化CT值方法鉴别卵巢囊实性肿瘤的应用价值
- Author:
Bo ZHAO
;
Kun CAO
;
Haitao ZHU
;
Yuhong QU
;
Yingshi SUN
- Keywords:
ovarian tumors;
computed tomography
- From:
Journal of Practical Radiology
2017;33(9):1382-1385
- CountryChina
- Language:Chinese
-
Abstract:
Objective To assess the ability of pixel-based quantitative evaluation of CT values in differentiating benign and malignant cystic-solid ovarian tumors.Methods CT images of 41 cystic-solid ovarian lesions from 39 patients were reviewed,with 27 benign and 14 malignant confirmed by post-operation pathology or follow-up.Regions of interest (ROIs) were drawn along edges of tumors on all slices of contrast-enhanced images with ImageJ software.CT values of each pixel were extracted.CT values of 20,25,30,35 and 40 HU were used respectively as the threshold to divide cystic and solid components.Solid proportion,the mean and median CT values of solid component were calculated and compared between benign and malignant groups.Results Mean CT values of solid components were all higher in malignant than in benign ovarian masses under all the threshold values (P<0.05).For median CT values, the same trend existed under the threshold of 20,25,35 and 40 HU (P<0.05).For the solid proportion,difference was found only under the 40 HU threshold, with lower value in malignant group (0.67±0.25) than in benign group (0.47±0.31).ROC curves were drawn to differentiate benign and malignant lesions.The highest AUC was obtained by using the mean CT value of solid components defined by 40 HU threshold (AUC=0.735).Conclusion Pixel-based quantitative evaluation on CT images could help to define cystic and solid components of ovarian masses, with 40 HU to be an optimal threshold.Cystic-to-solid proportion and CT value of solid components derived from whole lesion can help to differentiate benign or malignant lesions.
