Prediction of the Cochlear Implant Electrode Insertion Depth: Clinical Applicability of two Analytical Cochlear Models.
Griet MertensV Van RompaeyP Van de HeyningE GorrisV TopsakalPublished in: Scientific reports (2020)
Although the spiral anatomy of the human cochlea seems evident, measuring the highly inter-variable true dimensions is still challenging. Today, only a few three-dimensional reconstruction models of the inner ear are available. Previously, spiral equations were applied to two-dimensional computed tomography (CT) images to predict the electrode insertion depth prior to cochlear implantation. The study aimed primarily to compare the clinical applicability of two analytical cochlear models using a recently introduced planning software to predict the insertion depth of the electrode array of 46 cochlear implant recipients. One was based upon the Escudé formula, which relies only on the basal turn diameter, and another based upon the Elliptic-Circular Approximation (ECA), using the diameter and width. Each case was measured twice by two ENT surgeons. Secondly, in order to measure the benefit of the new planning software over the use of the existing clinical routine method, the results were compared to the prediction based upon a two-dimensional CT image. The intra -and inter-observer agreement using the planning software was significantly better when the ECA was applied, compared to the Escudé formula (p < 0.01). As a reference, the predicted insertion depth was compared to the actual insertion depth measured on post-operative images. The mean absolute error was |2.36| (|1.11|) mm in case of the Escudé approach and |1.19| (|0.92|) mm in case of the ECA. The use of a new planning software that allows three-dimensional handling, integrating the diameter and width of the basal turn (ECA formula), resulted in the most accurate predictions of the electrode insertion depths.
Keyphrases
- optical coherence tomography
- computed tomography
- deep learning
- optic nerve
- positron emission tomography
- dual energy
- carbon nanotubes
- image quality
- magnetic resonance imaging
- high resolution
- endothelial cells
- convolutional neural network
- fluorescent probe
- preterm infants
- living cells
- machine learning
- liquid chromatography
- hearing loss
- pet ct
- high density