Method and apparatus for identifying constituent signal components from a plurality of evoked physiological composite signals

Abstract

A novel application of independent component analysis (ICA) to data acquired by a single sensor. The technique exploits the unique relationship between multiple physiologic (source) and electronic (artifact) components in surface recorded sensory nerve action potential (SNAP) waveforms that are evoked by different activating magnitudes. A forward model of the SNAP is developed and used to test the approach on a simplified data simulation. The method is applied to experimental data and shown to be effective at separating artifact and source components and reconstructing artifact-free traces. A method of automated reconstruction for use within an expert system is also disclosed.

Description

REFERENCE TO PENDING PRIOR PATENT APPLICATION [0001] This patent application claims benefit of pending prior U.S. Provisional Patent Application Ser. No. 60 / 298,831, filed Jun. 18, 2001 by Martin D. Wells for METHODS FOR EXTRACTING OR SEPARATING MULTIPLE EVOKED PHYSIOLOGICAL SIGNAL COMPONENTS FROM RECORDINGS CONSISTING OF THEIR MIXTURES (Attorney's Docket No. NEURO-3 PROV), which patent application is hereby incorporated herein by reference.FIELD OF THE INVENTION [0002] This invention relates to medical apparatus and methods in general, and more particularly to methods and apparatus for identifying constituent signal components from a plurality of evoked physiological composite signals. BACKGROUND OF THE INVENTION [0003] Non-invasive peripheral nerve conduction studies (NCS) are an important tool in the diagnosis and assessment of neuromuscular injuries and pathologies. Electrical stimulation of a nerve bundle by surface electrodes produces impulses that travel in both the proximal ...

AI Technical Summary

Benefits of technology

[0017] The present invention can be described as single channel independent component analysis (SCICA). SCICA is a technique that allows ICA to be applied to peripheral evoked potential (PEP) signals. Applications of SCICA to peripheral electrodiagnostics include the removal of stimulation artifacts and the deconvolution of overlapping components. Removal of corrupting artifacts can improve the accuracy of electrodiagnostic parameter estimation. Removal of corrupting artifacts also permits improved diagnostic indices to be identified in the independent component domain, which may more closely represent the underlying electrophysiology. SCICA may also be implemented within a fully automated expert system performing waveform analysis for peripheral neuromuscular diagnostics.

Problems solved by technology

As noted above, while independent component analysis (ICA) appears to be a very useful tool for blind source separation and removal of contaminating artifacts from cortical evoked potential and EEG recordings, spatially separated peripheral compound muscle and sensory nerve action potentials do not fit the model of linear mixtures normally required for ICA.

Images