Automated Analysis of Intra-Cochlear Electrode Contact Locations from Post-Operative CT-Scans

Background

A postoperative computed tomography (CT) scan can provide feedback about the intra-cochlear position of a cochlear implant (CI). It is, however, difficult to determine the cochlear anatomy from postoperative images alone due to the metallic artefacts. This can be improved by obtaining the anatomy from pre-operative images and the CI position from aligned post-operative images. In previous studies we have shown that this method is reliable, but time-consuming, in the hands of expert evaluators.

Objective

Our goal is to develop and test the reliability of an automatic algorithm, which can do this faster and with less human effort.

Methods

After aligning pre- and post-operative CT scans, we approximate the electrode locations using watershed analysis. These locations are fitted with a spiral and the intensity profile along this spiral is extracted. Using a priori knowledge about the electrode, we then construct a model for each CI type by using a sinusoid with peaks that follow the known contact spacing. Finally, we maximize the dot product between the model and the intensity profile by optimizing the scale of the model and the shift along the intensity profile to locate the electrode contact centers.

Results

We compared automatically and manually determined intra-cochlear positions for the HiFocus Mid-Scala electrode array and found an average 2D-distance of 0.16mm (n=103).

Conclusions

Our method will allow for valuable feedback to the CI surgeon and an intricate analysis of the influence of surgical and anatomical parameters on the intra-cochlear position of the CI.