Comparison between borderline ovarian tumors and carcinomas using semi-automated histogram analysis of diffusion-weighted imaging: focusing on solid components.

PURPOSE: This study aimed to evaluate whether histogram analysis of the apparent diffusion coefficient (ADC) of a solid tumor component could distinguish borderline ovarian tumors from ovarian carcinoma.

MATERIALS AND METHODS: Sixteen pathologically proven borderline tumors and 21 carcinomas were retrospectively examined. Magnetic resonance (1.5-T) image data sets were coregistered, and the solid components of each tumor were semiautomatically segmented. ADC histograms of the solid components were extracted; modes, minimums, means, and 10th, 25th, 50th, 75th, and 90th percentiles of the histograms were compared between the two tumor types, and receiver-operating characteristic (ROC) analysis was performed.

RESULTS: The mode, minimum, mean, 10th, 25th, 50th, and 75th percentile ADC values of solid components of borderline tumors were significantly larger than those of carcinomas. Among these, the 10th percentile values had the lowest p value (p = 0.0003). At ROC analysis, the area under the curve (AUC) in the 10th percentile was the greatest (0.854), and the best cutoff value in the 10th percentile provided the highest specificity (93.8 %).

CONCLUSIONS: ADC histograms of solid tumor components facilitated the distinction between borderline ovarian tumors and carcinoma. The 10th percentile ADC values had the best diagnostic performance.