A histological validation of an automatic cochlear implant electrode placement assessment from clinical computed...

The purpose of this study was to validate the accuracy of the post-operative sub-anatomical cochlear implant placement outcomes of an automatic image analysis tool with respect to histological analysis.

Fourteen fresh-frozen temporal bones (TBs) were implanted. Pre- and post-operative computed tomography (CT) scans were processed through Nautilus, which provided information about electrode placement. Histological examination was then conducted to assess the cross-sectional positions of the electrode array at different insertion angles in relation to the cochlea.

The Nautilus reconstructions were coherent with the histological analysis for all cases (k = 1.0, n = 39 histology measurements). Nautilus' prediction of the basilar membrane and electrode position with respect to the walls of the bony labyrinth yielded average absolute errors of 0.09 mm (r = 0.54) and 0.23 mm (r = 0.40), respectively (n = 26 cross-sections). Additionally, Nautilus yielded mean dice scores of 0.90, 0.83, and 0.90 (n = 26 cross-sections) and average surface distance errors of 0.05, 0.07, and 0.06 mm (n = 26 cross-sections) for the scala tympani, vestibuli, and bony labyrinth, respectively.

The identification of intra-cochlear structures from CT images is challenging due to the limited resolution and metallic artifacts. Nautilus provides an easily accessible research tool for automated cochlear analysis using pre- and postoperative CT images.

This study, conducted on cadaver bones, suggests that Nautilus may be a reliable tool for inferring the position of the basilar membrane and electrode array in CT scans.