Extracting impedance changes from a frequency multiplexed signal during neural activity in sciatic nerve of rat: preliminary study in vitro.

OBJECTIVE: To establish suitable frequency spacing and demodulation steps to use when extracting impedance changes from frequency division multiplexed (FDM) carrier signals in peripheral nerve. APPROACH: Experiments were performed in vitro on cadavers immediately following euthanasia. Neural activity was evoked via stimulation of nerves in the hind paw, while carrier signals were injected, and recordings obtained, with a dual ring nerve cuff implanted on the sciatic nerve. Frequency analysis of recorded compound action potentials (CAPs) and extracted impedance changes, with the latter obtained using established demodulation methods, were used to determine suitable frequency spacing of carrier signals, and bandpass filter (BPF) bandwidth and order, for a frequency multiplexed signal. MAIN RESULTS: CAPs and impedance changes were dominant in the frequency band 200 to 500 Hz and 100 to 200 Hz, respectively. A Tukey window was introduced to remove ringing from Gibbs phenomena. A ±750 Hz BPF bandwidth was selected to encompass 99.99% of the frequency power of the impedance change. Modelling predicted a minimum BPF order of 16 for 2 kHz spacing, and 10 for 4 kHz spacing, were required to avoid ringing from the neighbouring carrier signal, while FDM experiments verified BPF orders of 12 and 8, respectively, were required. With a notch filter centred on the neighbouring signal, a BPF order of at least 6 or 4 was required for 2 and 4 kHz, respectively. SIGNIFICANCE: The results establish drive frequency spacing and demodulation settings for use in FDM electrical impedance tomography (EIT) experiments, as well as a method for their selection, and, for the first time, demonstrates the viability of FDM-EIT of neural activity on peripheral nerve, which will be a central aspect of future real-time neural-EIT systems and EIT-based neural prosthetics interfaces.

Empirical validation of statistical parametric mapping for group imaging of fast neural activity using electrical impedance tomography.

Electrical impedance tomography (EIT) allows for the reconstruction of internal conductivity from surface measurements. A change in conductivity occurs as ion channels open during neural activity, making EIT a potential tool for functional brain imaging. EIT images can have >10 000 voxels, which means statistical analysis of such images presents a substantial multiple testing problem. One way to optimally correct for these issues and still maintain the flexibility of complicated experimental designs is to use random field theory. This parametric method estimates the distribution of peaks one would expect by chance in a smooth random field of a given size. Random field theory has been used in several other neuroimaging techniques but never validated for EIT images of fast neural activity, such validation can be achieved using non-parametric techniques. Both parametric and non-parametric techniques were used to analyze a set of 22 images collected from 8 rats. Significant group activations were detected using both techniques (corrected p < 0.05). Both parametric and non-parametric analyses yielded similar results, although the latter was less conservative. These results demonstrate the first statistical analysis of such an image set and indicate that such an analysis is an approach for EIT images of neural activity.