Patient-specific estimation of detailed cochlear shape from clinical CT images

Abstract

Purpose

A personalized estimation of the cochlear shape can be used to create computational anatomical models to aid cochlear implant (CI) surgery and CI audio processor programming ultimately resulting in improved hearing restoration. The purpose of this work is to develop and test a method for estimation of the detailed patient-specific cochlear shape from CT images.

Methods

From a collection of temporal bone \(\mu \) CT images, we build a cochlear statistical deformation model (SDM), which is a description of how a human cochlea deforms to represent the observed anatomical variability. The model is used for regularization of a non-rigid image registration procedure between a patient CT scan and a \(\mu \) CT image, allowing us to estimate the detailed patient-specific cochlear shape.

Results

We test the accuracy and precision of the predicted cochlear shape using both \(\mu \) CT and CT images. The evaluation is based on classic generic metrics, where we achieve competitive accuracy with the state-of-the-art methods for the task. Additionally, we expand the evaluation with a few anatomically specific scores.

Conclusions

The paper presents the process of building and using the SDM of the cochlea. Compared to current best practice, we demonstrate competitive performance and some useful properties of our method.