Anzeige
Erschienen in:
02.08.2018 | Original Article
The accuracy of image-based safety analysis for robotic cochlear implantation
Erschienen in: International Journal of Computer Assisted Radiology and Surgery | Ausgabe 1/2019
Einloggen, um Zugang zu erhaltenAbstract
Purpose
To evaluate the accuracy and reliability of image-based safety analysis for robotic cochlear implantation (RCI) in an ex vivo assessment.
Methods
The accuracy was evaluated in a study on 23 human temporal bones. For image analysis, a computer-assisted safety analysis based on intraoperative cone beam computed tomography was implemented. The method automatically segments the drill tunnel and predicts the distance between the tunnel and the facial nerve. In addition, the drilling error at the target is predicted. The predicted distances were compared with the actually drilled distances measured in postoperative high-resolution micro-computed tomography scans. The automatic method was compared to accuracies associated with a manual analysis of the image data.
Results
The presented computerized image-based analysis enabled the proximity of the facial nerve to the drill trajectory to be predicted with an accuracy of 0.22 ± 0.15 mm and drilling error at the target to be predicted with an accuracy of 0.11 mm ± 0.08 during N = 19 RCI procedures. The manual assessment of facial nerve proximity was performed with an accuracy of 0.34 ± 0.20 mm by a trained clinical expert.
Conclusion
The assessment of intraoperative CT-based imaging presents multiple benefits over alternative safety mechanisms including early detection and applicability even in cases of malformation of the mastoid. This work presents a computer-assisted approach to image analysis that enables procedure safety measurements to be reliably performed with superior accuracy to other proposed safety methodologies, at a safe distance from the facial nerve. Its application must, however, be considered in relation to associated costs (time, cost, irradiation) and the dependence of the measure on a reliable preoperative segmentation.