OBJECTIVES: We undertook to verify the usefulness of computed tomography Hounsfield units (HU) in differentiating cholesteatoma from inflammatory tissue. METHODS: In 91 enrolled cases, the lesions were classified according to the gross pathology, and the specific locations of each lesion were documented by 1 surgeon within 1 day after the operation. The densities in HU of cholesteatoma and non-cholesteatoma lesions were retrospectively measured 3 times by the same examiner, and the difference between the two groups was analyzed. The interobserver reliability among the 3 examiners was assessed to verify the confidence level of the HU measurements in preoperative detection of cholesteatoma. RESULTS: The mean HU values of cholesteatoma were 35.7 to 66.6 HU, and those of non-cholesteatoma lesions were 32.9 to 51.3 HU. A general linear model-repeated-measures analysis of variance did not show any significant difference between the cholesteatoma and non-cholesteatoma lesions (p = 0.305). The general linear model-repeated-measures analysis of variance showed a significant difference of the measured HU levels among the 3 examiners (p = 0.021). CONCLUSIONS: This study showed that the HU values on preoperative computed tomography did not suffice for the detection of cholesteatoma lesions. A clinician's physical examination together with an interpretation of computed tomography is still the "gold standard" method.
OBJECTIVES: We undertook to verify the usefulness of computed tomography Hounsfield units (HU) in differentiating cholesteatoma from inflammatory tissue. METHODS: In 91 enrolled cases, the lesions were classified according to the gross pathology, and the specific locations of each lesion were documented by 1 surgeon within 1 day after the operation. The densities in HU of cholesteatoma and non-cholesteatoma lesions were retrospectively measured 3 times by the same examiner, and the difference between the two groups was analyzed. The interobserver reliability among the 3 examiners was assessed to verify the confidence level of the HU measurements in preoperative detection of cholesteatoma. RESULTS: The mean HU values of cholesteatoma were 35.7 to 66.6 HU, and those of non-cholesteatoma lesions were 32.9 to 51.3 HU. A general linear model-repeated-measures analysis of variance did not show any significant difference between the cholesteatoma and non-cholesteatoma lesions (p = 0.305). The general linear model-repeated-measures analysis of variance showed a significant difference of the measured HU levels among the 3 examiners (p = 0.021). CONCLUSIONS: This study showed that the HU values on preoperative computed tomography did not suffice for the detection of cholesteatoma lesions. A clinician's physical examination together with an interpretation of computed tomography is still the "gold standard" method.