SPARC: Method for Overcoming Temporal Dispersion in Unmyelinated Nerves for Imaging C Fibres with Electrical Impedance Tomography (EIT)

Abstract

Fast neural EIT is a novel technique that enables imaging of evoked myelinated fibre activity in fascicles in rat sciatic nerve in vivo. However, imaging unmyelinated fibres is challenging due to temporal dispersion of the neural activity, caused by variability in fibre conduction velocities. This reduces the signal to below noise several centimetres from onset. The purpose of this work was to develop a method to overcome dispersion and allow EIT recordings of phasic activity at any distance from onset along the nerve.MethodsCompound impedance changes (dZ) were simulated using an electrically stimulated CAPs in 1D FEM model of a C fibre [1], with statistical analysis for 5000 fibres with conduction velocities 0.6±0.07 m/s (mean±SD) [2]. This enabled simulation of dZ up to 20 cm from the stimulus with stimulus trains of 5/10/20/50 Hz lasting 0.25–8 s with 3 s intervals between trains (Fig. 1a). The total duration of each simulation was 100 s.dZ was simulated 1) By coherent averaging of responses of supramaximal activity of single spikes [3] with demodulation bandwidth of 50–300 Hz and 2) By bandpass filtering of entire evoked spike trains with bandwidths of 0.25–5 Hz which allowed significant reduction of the noise without the need for coherent spike averaging.Random gaussian noise (0.015 μV RMS) was added to the signal to simulate realistic recording conditions. The SNR in the utilised approaches was computed for all tested stimulation and recording paradigms and compared.ResultsWhile coherent spike averaging was superior at short distances, its SNR fell exponentially with the distance from the onset (Fig. 1b, solid lines). The SNR with the compound demodulation was more consistent across distances (~2 at the optimal stimulation frequency of 20 Hz, Fig. 1b, dashed lines) allowing recording at ≥ 20 cm from stimulation. The estimated minimal averaging time required to obtain an SNR of 4 at 20 cm from the stimulus was equal to approximately 30 minutes at 20 Hz.Conclusion and future workThe developed stimulation and signal processing paradigm demonstrated the feasibility of dZ recording at long distances from the stimulus. Future work is to validate it on the ex vivo crustacean leg nerve model and subsequently an in vivo mammalian model with the final objective of imaging C fibres in the vagus nerve in humans. If successful, this technique will be applied for improvement of the emerging field of bioelectronic medicines [4].Support or Funding InformationThe work was supported by NIH SPARC grant no: 3OT2OD026545‐01S2.(a) Example of AP and dZ responses of the C fibre subjected to repetitive 50 Hz stimuli; (b) SNR computed for the single and compound recording paradigms.Figure 1