SYSTEM AND METHOD FOR EVALUATING AN ELECTROPHYSIOLOGICAL SIGNAL

• US 20140194758A1
• Filed: 08/05/2013
• Published: 07/10/2014
• Est. Priority Date: 02/04/2011
• Status: Active Grant

• Current Assignee: 8825319 Canada Limited
• Original Assignee: 8825319 Canada Limited
• Inventors: KORENBERG, Michael
• Granted Patent: US 9,131,864 B2
• Time in Patent Office: Days
• US Class Current: 600/509
• CPC Class Codes: A61B 5/02028 Determining haemodynamic parameters not otherwise provided for, e.g. cardiac contractility or left ventricular ejection fraction
• CPC Class Codes: A61B 5/0245 by using sensing means generating electric signals, i.e. ECG signals
• CPC Class Codes: A61B 5/055 involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
• CPC Class Codes: A61B 5/316 Modalities, i.e. specific diagnostic methods
• CPC Class Codes: A61B 5/339 Displays specially adapted therefor
• CPC Class Codes: A61B 5/341 Vectorcardiography [VCG]
• CPC Class Codes: A61B 5/349 Detecting specific parameters of the electrocardiograph cycle
• CPC Class Codes: A61B 5/35 by template matching
• CPC Class Codes: A61B 5/361 Detecting fibrillation
• CPC Class Codes: A61B 5/363 Detecting tachycardia or bradycardia
• CPC Class Codes: A61B 5/4848 Monitoring or testing the effects of treatment, e.g. of medication
• CPC Class Codes: A61B 5/4884 inducing physiological or psychological stress, e.g. applications for stress testing
• CPC Class Codes: A61B 5/7203 for noise prevention, reduction or removal
• CPC Class Codes: A61B 5/7232 involving compression of the physiological signal, e.g. to extend the signal recording period
• CPC Class Codes: A61B 5/7239 using differentiation including higher order derivatives
• CPC Class Codes: A61B 5/7253 characterised by using transforms
• CPC Class Codes: A61B 6/032 Transmission computed tomography [CT]
• CPC Class Codes: G06F 2218/04 Denoising
• CPC Class Codes: G06F 2218/12 Classification; Matching
• CPC Class Codes: G16H 50/50 for simulation or modelling of medical disorders

Abstract

A method of evaluating an electrophysical signal. The method utilizes a model-derived reconstruction over at least one cycle of the electrophysical signal to identify a pathological event, whereby at least one term in the model is differentiable. The invention is also directed to a non-transitory computer readable medium having instructions stored thereon for identifying a pathological event from a model-derived reconstruction of an electrophysical signal. The invention is further directed to a system for evaluating an electrophysical signal. The system includes a processor configured to identify a pathological event from a model-derived reconstruction of the electrophysical signal, a data input coupled to the processor for providing the processor with the electrophysical signal, and a user interface.

Claims

1. A method of detecting an abnormality in a patient from an electrophysiological signal, comprising:

a) obtaining an electrophysiological signal corresponding to the patient;

b) finding a model corresponding to the electrophysiological signal wherein at least one term in the model is differentiable;

c) obtaining the derivative of the at least one term; and

d) using the derivative over at least one cycle of the electrophysiological signal in order to detect the abnormality.

2. The method of claim 1, wherein the derivative has an order, and wherein the order is a real number.

3. The method of claim 2, wherein the order of the derivative is a positive number.

4. The method of claim 3, wherein the order of the derivative is a positive integer.

5. The method of claim 1, wherein the at least one model term has a derivative of order a and also of order a-1, and wherein a ratio of the derivative of order a to the derivative of order a-1 exists over a plurality of points over at least one cycle of the electrophysiological signal.

6. The method of claim 5, wherein the order a is a positive integer, and wherein the ratio is used over at least one cycle of the electrophysiological signal in order to detect the abnormality.

7. The method of claim 6, wherein the ratio is displayed over at least one cycle of the electrophysiological signal to form a plot, and wherein one or more highlighted indicators are used in the plot to indicate those portions where the ratio is positive.

8. The method of claim 7, wherein the electrophysiological signal is a portion of an electrocardiogram, and wherein the plot is placed over a portion of a 3-D outline of the heart, and wherein one or more highlighted indicators are used to indicate the areas in the interior or exterior of the heart where damaged areas are in the heart.

13. The method of claim 1, wherein the electrophysiological signal comprises a portion of an electrocardiogram including portions of at least two lead signals, wherein an orthogonalization process is used to create at least two mutually orthogonal signals from the portions of the at least two lead signals, and wherein the at least two mutually orthogonal signals are used in place of the portions of the at least two lead signals.

14. The method of claim 1, wherein the electrophysiological signal comprises portions of at least two signals measured at corresponding instants of time, wherein the derivative is used in a mathematical expression that is applied to each of the at least two signals at corresponding instants in time to obtain positive or negative indications of the abnormality, wherein the expression is used to form a plot corresponding to different instants of time, and wherein one or more highlighted indicators are used to indicate segments of the plot when the expression detects the abnormality, said highlighted indicators comprising at least one of:

color coded indicators, including use of a first and a second color such that the first color dominates at instants when negative indications of the abnormality dominate, the second color dominates at instants when positive indications of the abnormality dominate, and the highlighted indicator is closer to neutral in hue at instants when positive indications of the abnormality are offset by negative indications of the abnormality; and

shaded indicators, one or more type of broken line, and one or more type of line thickness.

15. The method of claim 8, wherein the highlighted indicators comprise at least one of color coded indicators, shaded indicators, one or more type of broken line, and one or more type of line thickness.

17. The method of claim 2, wherein a function of the derivative is defined, wherein the function of the derivative assumes at least one value over a plurality of points over at least one cycle of the electrophysiological signal, and wherein at least one representation of the at least one value is displayed over the plurality of points.

18. The method of claim 3, wherein a function of the derivative is defined,

wherein the function of the derivative assumes at least one value over a plurality of points over at least one cycle of the electrophysiological signal, and wherein at least one representation of the at least one value is displayed over the plurality of points.

9. A method for building a model approximating an electrophysiological signal, wherein the method includes choosing a measure of approximating the signal, wherein the method involves a search through candidate terms to select terms to add to the model at successive stages, and wherein at least two distinct candidate terms are selected at the same stage at least once, where one of the selected terms causes, out of the candidate terms searched, the greatest reduction in the measure of approximating the signal, and another of the selected terms causes, out of the candidate terms searched, at least one of:

a relative maximum of the reduction of the measure of approximating the signal; and

a reduction of the measure of approximating the signal above a specified threshold level.

10. The method of claim 9, further comprising:

a) receiving the electrophysiological signal at a computing device;

b) selecting, at the computing device, a plurality of terms for the model approximating the electrophysiological signal;

c) separating a noise component from the plurality of terms selected for the model; and

d) forming a reconstructed electrophysiological signal whereby the noise component is removed by using a subset of the plurality of terms selected for the model.

11. The method of claim 9, wherein the at least two candidate terms that are selected at the same stage are complex exponentials.

12. The method of claim 9, further comprising:

applying a modified fast orthogonal search (MFOS) transform to the electrophysiological signal to generate a plurality of terms corresponding to the electrophysiological signal; and

forming a reconstructed electrophysiological signal whereby a noise component is separated from the electrophysiological signal using a subset of the plurality of terms, corresponding to the electrophysiological signal, which have at least a desired compression ratio.

16. The method of claim 9, wherein the measure of approximating the signal involves at least one of the error, the square error, the mean square error, the weighted mean square error, the maximum square error, and the perpendicular distance to a hyperplane.

19. A method of using a model-building procedure to select a concise subset of electrocardiogram (ECG) leads to use out of a larger set of ECG leads, comprising:

a. for each lead in the larger set, using the corresponding lead signal to define a candidate signal;

b. for each lead in the larger set, using the corresponding lead signal to define a desired target output signal from an experiment; and

c. selecting the same subset of lead signals, out of the candidate signals, to approximate all the target signals from all the experiments by linear combinations of the selected lead signals, while allowing the coefficients of the selected lead signals to vary in approximating the target signals from one experiment to another.

20. The method of claim 19, wherein selecting the same subset of lead signals uses at least one of fast orthogonal search (FOS), modified fast orthogonal search (MFOS), least angle regression (LARS), LASSO, and iterative fast orthogonal search (IFOS).