Accuracy of Preoperative Cochlear Duct Length Estimation and Angular Insertion Depth Prediction

Abstract

ObjectiveIn cochlear implantation with flexible lateral wall electrodes, a cochlear coverage of 70% to 80% is assumed to yield an optimal speech perception. Therefore, fitting the cochlear implant (CI) to the patient’s individual anatomy has gained importance in recent years. For these reasons, the optimal angular insertion depth (AID) has to be calculated before cochlear implantation. One CI manufacturer offers a software that allows to visualize the AID of different electrode arrays. Here, it is hypothesized that these preoperative AID models overestimate the postoperatively measured insertion angle. This study aims to investigate the agreement between preoperatively estimated and postoperatively measured AID.Study DesignRetrospective cross-sectional study.SettingSingle-center tertiary referral center.PatientsPatients undergoing cochlear implantation.InterventionPreoperative and postoperative high-resolution computed tomography (HRCT).Main Outcome MeasuresThe cochlear duct length was estimated by determining cochlear parameters (Avalue andBvalue), and the AID for the chosen electrode was (i) estimated by elliptic circular approximation by the software and (ii) measured manually postoperatively by detecting the electrode contacts after insertion.ResultsA total of 80 HRCT imaging data sets from 69 patients were analyzed. The mean preoperative AID estimation was 662.0° (standard deviation [SD], 61.5°), and the mean postoperatively measured AID was 583.9° (SD, 73.6°). In all cases (100%), preoperative AID estimation significantly overestimated the postoperative determined insertion angle (mean difference, 38.1°). A correcting factor of 5% on preoperative AID estimation dissolves these differences.ConclusionsThe use of an electrode visualization tool may lead to shorter electrode array choices because of an overestimation of the insertion angle. Applying a correction factor of 0.95 on preoperative AID estimation is recommended.