Innovative techniques for atraumatic and precise electrode array insertion in cochlear implantation

Abstract

Cochlear implant provides hearing perception to people with severe to profound hearing loss. The electrode array inserted during the surgery directly stimulates the hearing nerve according to the frequency bands. The complications during the cochlear implant insertion may cause trauma leading to infection, residual hearing loss and poor speech perception. The motivation for this work is to reduce the trauma induced during electrode array insertion process by carefully designing a sensing method, an actuation system and data driven control strategy to guide electrode array in scala tympani. Due to limited intra-operative feedback during the insertion process, complex bipolar electrical impedance is used as a sensing element. For this, custom impedance meter is implemented along with multiplexers to scan the electrodes to record their bipolar impedance magnitude, phase, and their resistive and reactive components. A 3-DoF actuation system is used for automated insertion and machine learning algorithms are employed for feedback control to steer electrode array atraumatically. This work is mainly focused on electrode array insertion direction, trajectory, and depth. Insertion trajectory is thought to influence trauma induced, final array placement and insertion failure. Initial experiments were performed to see the changes in complex impedance when certain electrode rub against the scala typani wall. It has been concluded that when the electrode array slide along the wall, the electrodes involved show high impedance magnitude and less negative impedance phase. Further several experiments were performed when array was inserted from three different directions; medial, middle and lateral. Complex impedance data recorded during insertion from different directions has the potential to discriminate different trajectories. Supervised machine learning approach is used to train and test the models for the prediction of different insertion trajectories and insertion depth. This method has shown the efficacy for not only achieving high prediction accuracy on full insertion complex impedance data but also on sub-sequence data. Further, prediction based on shorter data windows are utilized in feedback control loop for real-time control strategy that helps insert the array without any damage to its surroundings. Finally, linear insertion depth estimation of electrode array has been carried out using complex impedance data. A hybrid convolution and recurrent neural network regression model is utilized to predict insertion depth at every millimeter. It has been found that our complex impedance data is consistent and reliable and sensitive to the contact of electrode array with scala tympani walls during insertion. The contribution of this work is to predict different insertion direction, electrode array path and insertion depth using complex impedance data and machine learning tools. Furthermore, a preliminary autonomous system is demonstrated for real time classification and correction of electrode array path during insertion using closed loop control strategy.