Kyongtae T. Bae, MD, PhD Pornpim Fuangtharnthip, MD Srinivasa R. Prasad, MD Bonnie N. Joe, MD, PhD Jay P. Heiken, MD Index terms: Adrenal gland, CT, 86.12114, 86.12115 Adrenal gland, neoplasms, 86.317 Computed tomography (CT), quantitative, 86.12114, 86.12115 Published online 10.1148/radiol.2283020878 Radiology 2003; 228:735–742 Abbreviation: ROI = region of interest 1 From the Mallinckrodt Institute of Ra- diology, Washington University School of Medicine, 510 S Kingshighway Blvd, St Louis, MO 63110. Received July 19, 2002; revision requested Sep- tember 18; final revision received Jan- uary 27, 2003; accepted January 15. Address correspondence to K.T.B. (e-mail: baet@mir.wustl.edu). Author contributions: Guarantor of integrity of entire study, K.T.B.; study concepts and design, K.T.B., P.F., S.R.P.; literature research, K.T.B., P.F., B.N.J.; clinical studies, P.F., S.R.P.; data acquisition, P.F., S.R.P.; data analysis/interpretation, K.T.B., P.F., B.N.J.; statistical analysis, K.T.B., P.F.; manuscript preparation, K.T.B., P.F.; manuscript defi- nition of intellectual content, K.T.B., P.F., B.N.J., J.P.H.; manuscript editing, K.T.B., J.P.H., B.N.J.; manuscript revision/review, K.T.B., B.N.J., S.R.P., J.P.H.; manuscript fi- nal version approval, K.T.B. © RSNA, 2003 Adrenal Masses: CT Characterization with Histogram Analysis Method 1 PURPOSE: To evaluate a histogram analysis method for differentiating adrenal adenoma from metastasis at computed tomography (CT). MATERIALS AND METHODS: In a retrospective review of 2 years of clinical CT records, 223 adrenal adenomas in 193 patients (115 with contrast material– en- hanced CT, 43 with unenhanced and enhanced CT, and 35 with unenhanced CT) and 31 metastases (25 patients with enhanced CT) were found. In 158 patients with adenomas at enhanced CT, diagnosis was based on stable mass size for more than 1 year (n = 135) and characteristic signal intensity decrease at chemical shift magnetic resonance imaging (n = 23). In 35 patients with adenomas at unen- hanced CT, mean attenuation was 10 HU or less. Diagnosis of all metastases was based on rapid growth of a mass or new mass in less than 6 months in patients with cancer. Adrenal metastases with extensive necrosis were excluded. Histogram anal- ysis was performed in a circular region of interest (ROI) for mean attenuation, number of pixels, and range of pixel attenuation for all pixels and for the subset of pixels with less than 0 HU (“negative” pixels). Correlation between mean attenua- tion and percentage negative pixels was calculated. RESULTS: Negative pixels were present in all 74 unenhanced adenomas with mean attenuation of 10 HU or less and in 14 of 16 unenhanced adenomas with mean attenuation above 10 HU. Of 184 enhanced adenomas, only 20 had mean atten- uation of 10 HU or less, but 97 contained negative pixels (77 of these 97 masses had mean attenuation above 10 HU). Increase in percentage negative pixels was highly correlated with decrease in mean attenuation of both unenhanced and enhanced adenomas. None of the adrenal metastases had mean attenuation of 10 HU or less or contained negative pixels. CONCLUSION: The histogram method is far more sensitive than the 10-HU thresh- old method for diagnosis of adrenal adenomas at enhanced CT, with specificity maintained at 100%. © RSNA, 2003 Incidental discovery of adrenal masses is common at abdominal computed tomography (CT), occurring in up to 5% of patients (1). Most adrenal masses are adenomas. However, in a patient with a known extraadrenal neoplasm, particularly lung cancer, the finding of an adrenal mass is problematic. Detection of an adrenal mass requires the diagnosis or exclusion of an adrenal metastasis to determine appropriate therapy for the primary tumor. Findings in many studies confirm the usefulness of CT attenuation measurements at both unenhanced (2–5) and delayed contrast material– enhanced (6 –12) CT to differenti- ate benign from malignant lesions. Unlike metastases, adrenal adenomas often contain intracytoplasmic lipid in the adrenal cortex and thus demonstrate low attenuation at unenhanced CT. Korobkin et al (13) reported a high correlation between adrenal lipid content and low attenuation at CT. Boland et al (5) performed a meta-analysis of 10 CT studies and concluded that a threshold of 10 HU or less corresponded to a sensitivity of 71% and specificity of 98% in the diagnosis of adrenal adenomas. In addition, the delayed contrast-enhanced CT attenuation method has been used successfully to differentiate Genitourinary Imaging 735 R adiology