| Method and apparatus for identifying constituent signal components from a plurality of evoked physiological composite signals -> Monitor Keywords |
|
|
  Method and apparatus for identifying constituent signal components from a plurality of evoked physiological composite signalsUSPTO Application #: 20050283090Title: 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. (end of abstract)
Agent: Mark J. Pandiscio Pandiscio & Pandiscio, P.C. - Waltham, MA, US Inventor: Martin D. Wells USPTO Applicaton #: 20050283090 - Class: 600544000 (USPTO) Related Patent Categories: Surgery, Diagnostic Testing, Detecting Brain Electric Signal The Patent Description & Claims data below is from USPTO Patent Application 20050283090. Brief Patent Description - Full Patent Description - Patent Application Claims
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 and distal directions. Compound signals can be differentially recorded from the muscle or muscle group that is innervated by the stimulated nerve or from a separate location over the nerve itself. The amplitude and latency (or conduction velocity) of these evoked potential (EP) signals are calculated and used clinically to determine the location of nerve lesions and/or to provide an overall characterization of nerve function. More elaborate analysis of both compound muscle action potentials (CMAPs) and sensory nerve action potentials (SNAPs) have also been investigated and are believed to provide more precise diagnoses and assessment by extracting additional information from the complex signals. [0004] Large artifacts due to the electrical stimuli often appear in surface EP traces. Stimulus artifacts can be significant enough in magnitude and duration to contaminate the CMAP or SNAP waveform. Signal contamination can be severe, particularly in SNAP recordings where the evoked potential may only be a few microvolts in amplitude. The causes of stimulus artifacts include actual voltage gradients between the recording electrodes, capacitive coupling between the stimulation and detection hardware, and shaping of the stimulus spike by the detection amplifier and analog filters. The magnitude of the artifacts can generally be reduced through careful hardware design, improved skin preparation, and the use of sample-and-hold amplifiers or delay circuits that exclude the stimulus from the recorded action potential trace. In general, however, the stimulus artifacts cannot be completely eliminated from peripheral evoked potential recordings and may dwarf the EP waveform even after implementing these measures. [0005] Several methods of post-processing to remove stimulus artifacts from EP recordings have been investigated and documented. Inverse filtering to counteract the effects of the detection amplifier, fitting of an artifact to a parameterized function, estimation and subtraction of an artifact from a separate recording, a sub-threshold stimulus or a second stimulus pulse during the refractory period, and non-linear adaptive filtering techniques have all been used. While these methods have proven effective and useful, none are universally applicable and the search for new methods for stimulus artifact removal continues. [0006] Another aspect that complicates evoked potential analysis is the compound nature of the recorded signals. Often, healthy and diseased tissues are both present and are both activated and recorded. The response of healthy tissue, having a normal amplitude and latency, can mask the effect of existing pathology. Ideally, the healthy and diseased tissues could be measured separately, but this is very difficult in practice. Alternatively, it would be useful to be able to separate the healthy and diseases responses from compound signals that contain both. [0007] Independent component analysis (ICA) is a statistical analysis method that has applications in telecommunications, image processing, and biomedical signal analysis. ICA identifies and extracts the contributions of different, non-Gaussian sources given multiple recordings that are linear mixtures of those contributions. The mixtures may be of multiple sources of interest, in which case ICA allows tracking of the amplitude and latency of each separated source, or they may include unwanted signals such as artifacts that can, after being identified with ICA, be removed from the recordings. Often referred to as a method of blind source separation (BSS), ICA can be performed with no a-priori knowledge of the source signals other than their statistical independence, and no a-priori knowledge about the contribution of each source signal to the recorded mixtures. ICA can also be performed when limited knowledge is available or assumed about either the morphology of the source signals or their contributions to the recordings. Several ICA algorithms have been recently developed, including a fast ICA (FICA) Matlab package that is freely available on the World Wide Web. [0008] In biomedical signal analysis, ICA has been used very promisingly in the separation of multiple sources in scalp recordings of somato-sensory, visual, or auditory evoked potentials (SEP, VEP, AEP) or for source separation and the removal of motion and eye-blink artifacts in passive electroencephalography (EEG). These applications lend themselves to ICA because they involve numerous detection electrodes recording combinations of sources from a relatively large distance. The effects of source propagation and of dissimilar filtering by the intervening tissues are neglected for this far-field recording situation and all sources are assumed to contribute an identical, but scaled, component to each recording. [0009] This is not usually the case for peripheral EPs, which are recorded in closer proximity to a larger, more coherent group of sources. Different electrode locations over an activated muscle will produce CMAPs that differ in shape and temporal extent due to the active propagation of the generating sources and their near-field relationship to the detection electrodes. Similarly, even closely spaced detection sites along a nerve will see SNAPs that have different latencies and durations due to propagation of the sources past the electrodes and temporal dispersion among the individual action potentials that compose the compound SNAPs. [0010] 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. SUMMARY OF THE INVENTION [0011] As a result, one object of the present invention is to provide a novel method for identifying constituent signal components from a plurality of evoked physiological composite signals. [0012] And another object of the present invention is to provide novel apparatus for identifying constituent signal components from a plurality of evoked physiological composite signals. [0013] 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. [0014] In accordance with the present invention, it has now been discovered that one way to overcome this problem is to record different linear mixtures of multiple sources from a single recording site. In this case, the contribution from each source is known to have an identical shape and temporal extent in all recordings. Problems associated with the propagation of near-field sources and the differences in intervening tissues, common in multiple site recordings, are eliminated. As a result, independent component analysis (ICA) can be used for blind source separation and the removal of contaminating artifacts from evoked physiological composite signals. [0015] More particularly, the field of evoked, surface-recorded, peripheral neuromuscular electrodiagnostics involves the non-invasive activation and recording of biopotentials directly from nerve and muscle tissues. Surface-recorded, peripheral neuromuscular recordings do not lend themselves to multiple recording site ICA because the dispersive propagation of action potentials through the tissue and the "near-field" nature of the sources lead to violations of the standard assumption of signal congruency among recordings. In general, sources may be considered to be "near-field" when the spatial dimension of the physiologic activity is of the same order of magnitude as the spatial dimensions of the recording electrodes and of the spatial relationship between the electrodes and the active tissue. In accordance with the present invention, it has now been discovered that there are certain situations, however, for which multiple traces recorded from the same electrodes can be modeled as mixtures of the same, statistically independent components with different weighting coefficients--the same model that has been used to apply ICA to traces from multiple recording sites. The use of multiple signals from a single sensor obviates the assumption that different sources contribute similar components to each recording but requires a means, other than spatial separation, to generate multiple and different mixtures of the same signal components. [0016] In one preferred form of the invention, multiple and different mixtures of the same signal components (i.e., a plurality of evoked physiological composite signals) are generated by successively activating the patient's tissue with different stimuli. And in one particularly preferred form of the invention, the composite signals are generated by applying different levels, or grades, of the same stimuli to the tissue. Such stimuli may comprise electrical stimuli, mechanical stimuli, magnetic stimuli, etc.; electrical stimuli is generally most preferred. [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. BRIEF DESCRIPTION OF THE DRAWINGS [0018] These and other objects and features of the present invention will be more fully disclosed or rendered obvious by the following detailed description of the preferred embodiments of the invention, which is to be considered together with accompanying drawings wherein like numbers refer to like parts and further wherein: [0019] FIG. 1 is a multi-panel schematic view illustrating a forward model of compound sensory nerve action potential and stimulus artifact signal generation; Continue reading... Full patent description for Method and apparatus for identifying constituent signal components from a plurality of evoked physiological composite signals Brief Patent Description - Full Patent Description - Patent Application Claims Click on the above for other options relating to this Method and apparatus for identifying constituent signal components from a plurality of evoked physiological composite signals patent application. ###  How KEYWORD MONITOR works... a FREE service from FreshPatents1. Sign up (takes 30 seconds). 2. Fill in the keywords to be monitored. 3. Each week you receive an email with patent applications related to your keywords. Start now! - Receive info on patent apps like Method and apparatus for identifying constituent signal components from a plurality of evoked physiological composite signals or other areas of interest. ### Previous Patent Application: Method and apparatus useful in the diagnosis of obstructive sleep apnea of a patient Next Patent Application: Method and apparatus for determining conditions of a biological tissue Industry Class: Surgery ### FreshPatents.com Support Thank you for viewing the Method and apparatus for identifying constituent signal components from a plurality of evoked physiological composite signals patent info. IP-related news and info Results in 8.29018 seconds Other interesting Feshpatents.com categories: Novartis , Pfizer , Philips , Polaroid , Procter & Gamble , |
||
