Authors: François Guérit1, Simone R. de Rijk2, Manohar L. Bance2,Robert P. Carlyon1
1Cambridge Hearing Group, MRC Cognition and Brain Sciences Unit, University of Cambridge
2Cambridge Hearing Group, Department of Clinical Neurosciences, University of Cambridge
Background: Current spread within the cochlea limits speech outcomes with cochlear implants (CIs). One way to quantify the spread is to stimulate with one electrode while recording the scaled voltage at the other electrodes (also known as EFI, Electrical Field Interactions, TIM, Trans-Impedance Matrix or IFT, Impedance Field Telemetry). Interpreting TIM/EFI/IFTs in terms of current pathways is, however, not straightforward. A proposed solution has been to model TIM/EFI/IFTs with a ladder network of resistors (Vanpoucke et al. 2004). Although quite a simple model, it may still allow the localization of the main current pathways in the cochlea. However, it has only been validated using human recordings where the true current pathways are unknown.
Methods: We replicate the ladder model of Vanpoucke et al. 2004, but validate it by using 16-contact EFI recordings obtained with an implant inserted in a plastic tube (cf. Figure). This allows us to compare the modelled transversal current flow with openings of different sizes at different positions in the tube (A to D in Figure). We then investigate how the model fits data from cadaver and live human recordings, where the true current flow is unknown.
Results: The ladder network needed to be heavily constrained in order to fit the artificial current-shunt dataset. The constrained model could detect large changes in current pathways, such as there being a hole in a plastic tube at a basal vs middle vs apical location. Although the simplest model (order 1) could fit the artificial current-shunt dataset very well, increasing the model complexity improved the fit to human recordings, particularly near the stimulating electrode.
Conclusions: The constrained model showed great potential to detect transversal current shunts in a simple bench model of the cochlea. A comparison of the model output in different setups (bench model, cadaver, intra-operative and regular clinical routine) might allow for a better understanding of the current pathways in CI listeners.