Abstract
HYPOTHESIS: Computed tomography (CT) density measurement can be used to objectively distinguish otosclerosis from normal bone and to determine histologic grades of otosclerosis. BACKGROUND: Otosclerosis can be seen on CT as subtle radiolucent areas. An objective radiologic measurement that corresponds to known otosclerosis pathology may improve diagnostic accuracy, and could be used as a radiologic biomarker for otosclerosis grade. METHODS: A blinded, randomized evaluation of both histologic grade on histopathology slides and CT density measurement was performed on 78 human temporal bone specimens (31 with otosclerosis and 47 controls) that had undergone high-resolution multi-detector CT before histologic processing. Assessments were performed at 11 regions of interest (ROIs) in the otic capsule for each specimen. RESULTS: The CT density measurement mean (Hounsfield Units) ± standard deviation for all ROIs (Nos. 1-9) was 2245 ± 854 for grade 0 (no otosclerosis, n = 711), 1896 ± 317 for grade 1 (inactive otosclerosis, n = 109), and 1632 ± 255 for grades 2 and 3 combined (mixed/active otosclerosis, n 35). There was a strong inverse correlation of CT density to histologic grade at ROIs Nos. 1-5 (ANOVA, p < 0.0001). The inter-rater reliability for CT density was very good (correlation coefficient 0.87, p < 0.05). ROC curves suggested a cut-off of 2,150HU to distinguish otosclerosis from normal bone, and 1,811HU to distinguish low grade from mixed/high grade otosclerosis. CONCLUSIONS: In human temporal bone specimens, CT density may be used to distinguish normal bone from bone involved by otosclerosis. A higher histologic grade (i.e., indicating a more active otosclerotic focus) correlated with lower density.