Characterization of Adrenal Nodules With Dual-Energy CT: Can Virtual Unenhanced Attenuation Values Replace True Unenhanced Attenuation Values?

OBJECTIVE. The purpose of our study was to investigate whether virtual unenhanced adrenal nodule attenuation values can replace true noncontrast attenuation values. MATERIALS AND METHODS. Twenty-three incidentally discovered adrenal nodules (19 adenomas and four metastases) were identified in 19 patients (11 men and eight women; mean age, 65 years; age range, 38-84 years) who underwent unenhanced single-energy CT followed by contrast-enhanced dual-energy CT on the same scanner. A virtual unenhanced imaging dataset was generated from each dual-energy CT dataset. CT attenuation of each adrenal nodule was measured at the same location on virtual unenhanced images and true unenhanced images by three radiologists and mean values compared using the Student t test. Correlation between virtual unenhanced and true unenhanced values was determined using linear regression analysis. The mean difference and percentage of diagnostic agreement were also determined. Interreader variability was assessed using the intraclass correlation coefficient (ICC). RESULTS. The mean +/- SD attenuation values for virtual unenhanced images and true unenhanced images were 14.7 +/- 15.1 HU and 12.9 +/- 13.4 HU, respectively (p = 0.2). Strong positive correlation was observed between virtual unenhanced images and true unenhanced images (R = 0.83-0.87). The mean difference between virtual unenhanced images and true unenhanced images was 1.8 +/- 1.7 HU. Diagnostic agreement between virtual unenhanced images and true unenhanced images was 83-91% for three radiologists. No malignant nodules were misclassified as benign on virtual unenhanced images. The ICC was 0.88 and 0.96 for virtual unenhanced images and true unenhanced images, respectively, indicating high interreader agreement. CONCLUSION. Virtual unenhanced and true unenhanced attenuation measurements of adrenal nodules were not significantly different and showed strongly positive linear correlation. This finding resulted in substantial diagnostic agreement between virtual unenhanced images and true unenhanced images for distinguishing benign from malignant nodules.