Cochlear Implant-based Cortical Activity Recordings
A long-form technical account of the experimental framework, clinical translation, and signal-processing advances behind this research programme.
Project Overview
This work examines how intracochlear stimulation can be used to recover informative cortical responses in patients implanted with cochlear devices. The project is carried out in close collaboration with Prof. Manohar Bance, whose clinical perspective has guided both the experimental design and the translational interpretation of the data.
The programme has also received £10,000 in funding from the RNID, supporting the development of methods that could improve how auditory-evoked activity is measured and interpreted in clinical environments.
Methodology
Automated Electrical Impedance Spectroscopy
A central component of the workflow is an automated electrical impedance spectroscopy pipeline designed to characterize the response of the implant system under controlled stimulation conditions. The procedure enables repeatable measurement of impedance behaviour and supports identification of changes associated with device state and tissue interaction.
By standardizing acquisition and reducing manual intervention, the approach improves consistency across measurement sessions and makes it feasible to compare observations across patients and time points.
Fitting to the Cole-Cole Model
The impedance spectra are fitted to the Cole-Cole model to extract parameters that capture the effective electrical properties of the system. This model provides a compact representation of the underlying behaviour and allows the experimenter to separate broad trends from noise-driven deviations.
These fitted parameters form a useful bridge between raw electrical measurements and biologically relevant interpretation, helping contextualize the recordings within a stable physical model.
Equivalent COMSOL Models
Complementary equivalent COMSOL models were developed to evaluate how the implant geometry and surrounding media influence the recorded signals. These simulations allow the project to test hypotheses about current spread, field distribution, and the relationship between device configuration and measurable responses.
By combining experimental measurements with physics-based simulation, the project creates a more rigorous basis for interpreting both expected and unexpected observations.
Clinical Trials
The clinical component of the work involved a 16-patient study designed to assess the feasibility and reliability of the methodology in a real-world cochlear implant population. The study was structured to test whether the approach could uncover usable cortical responses under controlled conditions.
A key design feature was ipsilateral stimulation, in which the side of stimulation was aligned with the recording context to minimize ambiguity in the interpretation of the observed responses. This choice was intended to strengthen the specificity of the measurements while preserving clinical relevance.
Validation against scalp-EEG was conducted to compare the observed response profiles with a well-established non-invasive benchmark. This comparison was essential for assessing whether the implant-based recordings captured meaningful neural activity beyond what could be inferred from standard surface measurements.
Results
The results demonstrate that the noise removal algorithm substantially improved the clarity of the recordings, increasing signal-to-noise ratios by up to 13 dB. This improvement was particularly valuable in a setting where low-amplitude neural signals are easily obscured by system noise and physiological fluctuation.
These gains helped make the data more robust for downstream analysis and contributed to a stronger basis for comparing the implant-based recordings with scalp-EEG measurements. The outcome suggests that the methodology is not only technically feasible, but also clinically promising for future work in auditory neurophysiology.