Automatic techniques for cochlear implant CT image analysis

Abstract

Cochlear Implants (CIs) are neural prosthetic devices that provide a sense of sound to people who experience severe to profound hearing loss. Recent studies have demonstrated a correlation between hearing outcomes and the intra-cochlear locations of CI electrodes. Our group has developed image-guided CI programming (IGCIP) techniques that use image processing methods applied to computed tomography (CT) images to provide the patient-specific intra-cochlear locations of the implanted CI electrodes. With this information, IGCIP permits to select a patient-customized active electrode set. Clinical studies have shown that IGCIP is effective in improving hearing outcomes in both adults and children. Prior to this dissertation, several image analysis techniques used in IGCIP required human interventions and the sensitivity of IGCIP with respect to these techniques was unknown. In this dissertation, we first present several fully automated image analysis techniques for the accurate localization of several types of the CI electrode arrays in clinical CTs. Then, we present a fully automated active electrode set selection method. Last, we use micro-CTs and clinical CTs of temporal bone specimens implanted with different types of CI arrays to create a highly accurate ground truth dataset. It is used to rigorously characterize the accuracy of the image analysis techniques used in IGCIP and the sensitivity of IGCIP with respect to these. The validation studies show that the image analysis methods developed for array localization and electrode set selection are both accurate and robust.