Super-Resolution Classification Improves Facial Nerve Segmentation from CBCT Imaging

• Cochlear implantation (CI) is a treatment for patient suffering from severe to profound hearing loss. • A surgical planning software, OtoPlan, has been developed at the University of Bern for minimally invasive drilling of a cochlear access. • One of the constraints of the surgical planning during trajectory definition is to avoid the facial nerve with sufficient safety margin. • Cone-beam computed tomography (CBCT) is used for surgical planning, however, its relatively low resolution renders the identification of the nerve difficult. • In this work, we hypothesize that supervised-learning techniques can be used to segment the facial nerve from CBCT for more precise surgical planning.