Objectives/hypothesis: The objectives of this prospective study were to assess the reproducibility of the measurements of the cochlea and lateral semicircular canal (LSCC) and to determine if abnormal measurements predict sensorineural hearing loss (SNHL).
Methods: Two readers independently measured the cochlear height on coronal section and the LSCC bony island width on axial section on 109 temporal bone computed tomography scans; audiologic data on these patients were collected independently from medical records. Inter- and intrareader variability was evaluated using intraclass correlation coefficients (ICCs) based on a random-effects model. The positive and negative predictive values of abnormal measurement for hearing loss were determined.
Results: There was excellent inter- and intraobserver agreement for both measurements (ICC >80%). The average cochlear height was 5.1 mm (normal range, 4.4-5.9 mm) and average LSCC bony island width was 3.7 mm (normal range, 2.6-4.8 mm). Review of the original radiology reports demonstrated that both cochlear hypoplasia and LSSC dysplasia were overlooked in >50% of patients with both abnormal measurements and SNHL. Cochlear hypoplasia (< 4.4 mm) had a positive predictive value of 100% for SNHL, whereas cochlear hyperplasia and bony island dysplasia were less predictive.
Conclusion: The measurements of coronal cochlear height and axial LSCC bony width have excellent reproducibility and identify bony labyrinth abnormalities missed by visual inspection alone. In addition, cochlear hypoplasia is highly predictive of SNHL. To reliably identify inner ear malformations, measurement of the cochlear height and LSCC bony island width, in addition to the vestibular aqueduct, should be routinely performed on all temporal bone studies.