Device for analysis of a signal, in particular a physiological signal such as a ECG signal

• US 20040162495A1
• Filed: 11/13/2003
• Published: 08/19/2004
• Est. Priority Date: 11/14/2002
• Status: Active Grant

• Current Assignee: Ela Medical S.A. (LivaNova Plc)
• Original Assignee: Ela Medical S.A. (LivaNova Plc)
• Inventors: Quenet, Brigette, Faisandier, Yves, Dubois, M. Remi
• Granted Patent: US 7,359,749 B2
• Time in Patent Office: Days
• US Class Current: 600/509
• CPC Class Codes: A61B 5/349 Detecting specific parameters of the electrocardiograph cycle
• CPC Class Codes: A61B 5/353 Detecting P-waves
• CPC Class Codes: A61B 5/355 Detecting T-waves
• CPC Class Codes: A61B 5/36 Detecting PQ interval, PR interval or QT interval
• CPC Class Codes: A61B 5/726 using Wavelet transforms
• CPC Class Codes: A61B 5/7264 Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems neural networks per se G06N3/00; expert systems per se G06N5/00
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: Y02A 90/10 Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Abstract

A device for analyzing a signal, in particular a physiological signal such as an electrocardiogram or electrogram. The signal to be analyzed is preferably one that was previously collected, filtered, sampled and digitized. The device memorizes the digitized signal (stores in memory), and analyzes it by decomposing the signal into a plurality of N elementary waves, and classifying each N elementary waves by recognizing at least one parameter characteristic of each wave, and allotting a standardized label, selected among a plurality of predetermined labels, according to one or to more of the characteristic parameters thus recognized. The decomposition of the signal is into N parameterized bump functions (1-5), where each bump function is a continuous function defined by three successive intervals, respectively, a first monotonic parameterized function, an affine function, and a second monotonic parameterized function, with one of the monotonic parameterized functions being increasing and the other decreasing. The parameterized functions are preferably half-Gaussian functions, and the affine function preferably has a null slope.

Claims

1. A device for analyzing a signal having variations that define a monodimensional function, said signal having been collected, filtered, sampled and digitized, comprising:

means for memorizing the digitized signal in a memory, and

processor means for analyzing the memorized signal, comprising;

extracting means for decomposing the signal into a plurality of N elementary waves, each wave comprising a parameterized bump function, wherein each bump function is a continuous function having three successive intervals including a first monotonic parameterized function, an affine function, and a second monotonic parameterized function, one of said first and second monotonic parameterized functions being an increasing function and the other being a decreasing function; and

classifying means for recognizing at least one characteristic parameter of each N elementary wave, and allotting to said wave a standardized label selected from among a plurality of predetermined labels, according to said at least one recognized characteristic parameter.

2. The device of claim 1 wherein said device comprises an active medical device and said signal is a physiological signal collected by said device.

3. The device of claim 1 wherein said monodimensional function comprises a temporal dimension.

4. The device of claim 3 wherein said signal is an electrocardiographic signal forming a wave of the PQRST type.

5. The device of claim 4 wherein said analyzing means further comprises a subtracting means for withdrawing from the memorized signal at least one of the determined N elementary waves carrying an allotted label.

6. The device of claim 4 wherein said N elementary waves further comprise five waves.

7. The device of claim 6 wherein said predetermined labels comprise the P, Q, R, S and T waves of said electrocardiographic signal.

8. The device of claim 7 wherein the device further comprises means for determining a variability over time of at least one specific factor of at least one of the determined N elementary waves.

9. The device of claim 8 wherein said specific factor is selected from among an amplitude of the T wave, a temporal interval between the QRS wave and the T wave, a PR interval, an amplitude of the P wave, and a direction of a significant axis determined by a PCA analysis.

10. The device of claim 7 further comprising means for determining a temporary correlation of a specific factor between at least two of said determined elementary N waves.

11. The device of claim 10 wherein said specific factor is selected from among an amplitude of the T wave, a temporal interval between the QRS wave and the T wave, a PR interval, an amplitude of the P wave, and a direction of a significant axis determined by a PCA analysis.

12. The device of claim 4 wherein said electrocardiographic signal comprises a signal obtained by a PCA analysis and projection of the principal components on a significant axis.

13. The device of claim 12 wherein said significant axis comprises a dynamically computed axis of maximum amplitude.

14. The device of claim 1 wherein said affine function is a function having a null slope.

15. The device of claim 1 wherein each of said parameterized functions comprises a half-Gaussian function.

16. The device of claim 1 wherein said affine function is a function having a null slope and said first and second monotonic parameterized functions are each a half-Gaussian function, and said characteristic parameters further comprise a set of five parameters selected from among the standard deviation of each of the two half-Gaussian functions, a definition interval length of the affine function, an ordinate position of said interval, and a peak amplitude of said half-Gaussian function.

17. The device of claim 1 wherein said extracting means further comprises:

a library containing a plurality of predetermined bump types;

means for selecting from said library, for each of said N elementary waves, a bump type that is the most relevant in regard to the signal to be decomposed; and

means for adapting the parameters of each of the N selected bump-types and minimizing a variation between the signal and the composition of the parameterized N bump-types.

18. The device of claim 17 wherein the selecting means operates by an orthogonalisation of said selected most relevant bump-type.

19. The device of claim 17 wherein said adapting means performs a non-linear optimization under constraints of said parameters.

20. The device of claim 1 wherein said classifying means further comprises means for implementing hidden Markov models.