Differentiation of follicular carcinomas from adenomas using histogram obtained from diffusion-weighted MRI

To evaluate the role of histogram analysis of apparent diffusion coefficient (ADC) maps from diffusion-weighted imaging (DWI) in the differentiation of follicular thyroid carcinoma (FTC) from follicular adenoma (FA) in nodules indeterminate on ultrasound-guided core needle biopsy (USCNB). This study was performed with institutional review board approval. Seventeen patients who were planned to undergo diagnostic lobectomy for an indeterminate thyroid nodule (atypical of unknown significance/follicular lesion of undetermined significance [AUS/FLUS] or suspicious for follicular neoplasm/follicular neoplasm [SFN]) on USCNB were enrolled prospectively. All patients underwent DWI on the day before surgery. Histogram parameters were derived from ADC values obtained from the whole extent of the tumours. The parameters were compared with the final diagnosis based on histopathological examination after surgery. The accuracy of the parameters in differentiating FTC from FA was evaluated using receiver operating characteristic (ROC) curve analysis. Twelve patients were confirmed as having FA and five patients as having FTC. Histogram parameters including the 10th (ADC10), 25th (ADC25), and 50th (ADC50) percentiles of the ADC values were significantly lower in FA than in FTC (p < 0.05, all). ROC curve analysis revealed that ADC25 resulted in the highest AUC (0.867; confidence interval, 0.616–0.980), with a cut-off value of 0.352×10−3 mm2/s. Histogram parameters from ADC maps could differentiate FTC from FA effectively in indeterminate nodules on USCNB, with ADC25 being the most promising parameter. •Histogram analysis of ADC maps from DWI can help differentiate FA from FTC.•The ADC histograms of FA had a higher peakedness than those of FTC.•ADC25 is the best predictive parameter, with cutoff of 0.352 × 10−3 mm2/s.