Diffusion kurtosis imaging of the auditory nerve in unilateral and bilateral deaf patients: An automated processing and analysis pipeline

Abstract

Hearing loss is a condition that affects the microstructure of the auditory nerve. Diffusion Kurtosis Imaging (DKI) is a non-invasive MRI technique that can be used to assess the condition of the nerve in patients with hearing impairment. In this study, an automated pipeline for processing and analyzing diffusion-weighted imaging (DWI) data in subjects with normal hearing, unilateral, and bilateral hearing loss is proposed to research any differences in the auditory nerves of these three groups. The pipeline includes preprocessing steps for DWI artefact correction, the application of the DKI model and DTI-based deterministic tractography established on an automated algorithm for region-of-interest (ROI) generation based on manual segmentation of the subjects’ cochleae. Qualitative analysis of fiber tract bundle segments revealed alternations in shape and size between the left and right sides. Between normal hearing and unilateral deaf subjects, no significant alternations in the diffusion metrics of either side, for both Phase Encoding directions, were observed. Similarly, no differences were detected when comparing normal hearing and two-sided deaf subjects. Variations in mean kurtosis were investigated to assess the impact of hearing loss on microstructure complexity but with no significant outcomes. The automated pipeline was tested in multiple acquisition sessions in 6 out of 10 subjects and statistically proved to be reproducible. Enhancements to the pipeline can improve the quality of the results and should focus on the correction of deformations caused by EPI distortions and the integration of an automated masking method of the cochleae. Therefore, additional studies with a larger dataset and updated acquisition protocol are needed to validate and expand upon these findings to ensure the acceptance of DKI and fiber tractography as reliable methods for evaluating the condition of the disorder.