Blind source separation of nerve cuff recordings

Electrical stimulation of peripheral nerves can be used to restore partial motor function to individuals with neurological impairment. Previous work from our lab has shown that the flat interface nerve electrode (FINE) can be used to selectively stimulate fiber populations within a nerve trunk. We expect that selective recording of spatially segregated axon populations may also be possible with the FINE. The purpose of this modeling study was to assess the feasibility of using blind source separation (BSS) of the neurograms recorded with the FINE to distinguish signals from independent fascicles. We show that BSS is useful for identifying independent fascicular signals. Further, we introduce a simple post-BSS processing method that resolves the inherent permutation ambiguity of BSS, and allows the BSS-estimated signals to be deterministically related to the appropriate corresponding fascicles.