Increasing the resolution of morphological 3D image data sets through image stitching: application to the temporal bone

Abstract

Background and objective: Imaging of soft and hard tissue structures inside the inner ear is necessary for the development of new cochlear implant electrodes and their evaluation through insertion studies using human temporal bone specimens. An automated image stitching is presented to enhance the image resolution and the identification of the inner ear structures. Methods: Thirty human temporal bone specimens were either embedded in white epoxy resin for generating an anatomical atlas or in transparent epoxy resin for evaluation of insertion studies of cochlear implant electrodes. Cross-sectional overview images and the corresponding magnified detail images of these samples were stitched using the Scale Invariant Feature Transform algorithm. The resulting fiducial registration error (FRE) was evaluated. Results: The pixel spacing improved by a factor of 3.5, respectively, 5.6 depending on the magnification used for the detail images. The mean FRE is 0.5 ± 0.3 px based on all 4213 processed image pairs. Conclusion: The integrated image stitching improves the quality and resolution of three-dimensional histology and thus the soft tissue differentiation of membraneous structures inside the cochlea.