Method and apparatus for non-invasive diagnosis of cardiovascular and related disorders

• US 6,135,966 A
• Filed: 05/01/1998
• Issued: 10/24/2000
• Est. Priority Date: 05/01/1998
• Status: Expired due to Fees

• Current Assignee: Sarrus Biomedical Limited
• Original Assignee: Sarrus Biomedical Limited
• Inventors: Ko, Gary Kam-Yuen
• Primary Examiner(s): O'Connor, Cary
• Assistant Examiner(s): ASTORINO, MICHAEL C
• Application Number:
  US09/071,348
• Time in Patent Office: 907 Days
• Field of Search: 600/300, 600/800, 600/481-485, 600/501-509, 128/897-899, 128/920-924, 128/925
• US Class Current: 600/481
• CPC Class Codes: A61B 5/024 Detecting, measuring or recording pulse rate or heart rate A61B5/0205, A61B5/021 take precedence
• CPC Class Codes: A61B 5/411 Detecting or monitoring allergy or intolerance reactions to an allergenic agent or substance
• CPC Class Codes: A61B 5/7225 Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation A61B5/30, A61B5/316 take precedence; transformers for specific applications H01F38/00; amplifiers H03F
• CPC Class Codes: A61B 5/726 using Wavelet transforms
• CPC Class Codes: A61B 5/7267 involving training the classification device
• CPC Class Codes: A61B 7/04 Electric stethoscopes microphones, acoustic transducers therefor H04R
• CPC Class Codes: G16H 50/70 for mining of medical data, e.g. analysing previous cases of other patients
• CPC Class Codes: Y10S 128/03 Heart-lung
• CPC Class Codes: Y10S 128/925 Neural network

Abstract

Apparatus and method for non-invasive diagnosis of cardiovascular and related disorders. The system establishes a correspondence between the dynamics of the wave contour of the arterial pressure pulse and the associated disease states. The system comprises an input module, a contour signal receiver, and a processing module. The input module utilizes a pressure transducer for taking a non-invasive measurement of the arterial pulse. The contour signal receiver amplifies, digitizes and normalizes the arterial pressure pulse signal. In the processing module, the normalized arterial pressure contour is subjected to wavelet analysis, which transforms the dynamics of the time series of arterial blood pressure contour into multi-resolution wavelet coefficients or signatures. The processing module includes a neural network which is trained to associate the diagnostic features of the transformed arterial pressure contour embedded in the coefficients with a disease condition. After the learning phase, the system is capable of diagnosing known cardiovascular conditions in patients.

Claims

1. A method for diagnosing cardiovascular related illness, said method comprising the steps of:

(a) non-invasively obtaining a signal representative of a contour of an arterial pressure pulse over a single cardiac cycle and without disturbing the natural state of the artery;

(b) extracting frequency localization information and temporal localization information from said signal;

(c) providing the extracted information as an input to a neural network, the neural network having been trained with a plurality of training sets, each training set correlating an arterial pressure pulse contour with a known cardiovascular or related illness; and

(d) generating an illness identification output from the neural network.

2. The method as claimed in claim 1, wherein said step of extracting comprises eliminating noise from the extracted information by performing a multi-resolution decomposition of the signal.

3. The method as claimed in claim 1, wherein said step of extracting comprises performing a wavelet transform on the signal.

4. The method as claimed in claim 1, wherein said wavelet transform comprises a Daubechies family wavelet transform.

5. The method as claimed in claim 4, wherein said wavelet transform comprises a discrete wavelet transform, and said step of extracting comprises generating an approximation function and a plurality of detail functions including performing a multi-resolution decomposition of the signal.

6. The method as claimed in claim 1, wherein said training sets are obtained from a plurality of different patients under a plurality of different conditions.

7. The method as claimed in claim 1, wherein said illness identification output comprises a plurality of illness indicators, each of the known cardiovascular or related illnesses included in the training set being associated with a respective illness indicator, and each illness indicator indicating an occurrence of the respective illness.

8. A system for detecting symptoms associated with cardiovascular related illnesses in a patient said system comprising:

input means for receiving a signal representative of a contour of an arterial pressure pulse, said signal being obtained in a non-invasive manner from the patient and over a single cardiac cycle and without disturbing the natural state of the artery;

information extraction means coupled to said input means for extracting frequency localization information and temporal localization information from the signal and including means for producing a localization information output signal representing a portion of the extracted information; and

a neural network training with a plurality of training sets, each training set correlating an arterial pressure pulse contour with a known cardiovascular or related illness, said neural network including means for receiving said localization information output signal and means for producing an illness identification output signal representing a correlation between the extracted information and a cardiovascular related symptom derived from said training sets.

9. The system as claimed in claim 8, wherein said means for receiving is selectively coupled to said information extraction means for receiving only said localization information output signal.

10. The system as claimed in claim 8, wherein said information extraction means includes signal decomposition means for performing a multi-resolution decomposition of the received signal.

11. The system as claimed in claim 10, wherein said signal decomposition means comprises wavelet transformation means.

12. The system as claimed in claim 10, wherein said signal decomposition means comprises Daubechies wavelet transformation means.

13. The system as claimed in claim 8, wherein said receiving means comprises an analog-to-digital converter for generating a digitized representation of the arterial pressure pulse, said analog-to-digital converter including an input port for receiving an output signal of an arterial pressure pulse transducer, and wherein said information extraction means includes discrete wavelet transformation means coupled to said analog-to-digital converter for extracting approximation and detail function data from said received signal through a multi-resolution decomposition of the received signal, vector generation means coupled to the discrete wavelet transformation means for generating, as the localization information output signal, a vector comprising the extracted approximation and detail function data.

14. The system according to claim 8, wherein said neural network comprises a multi-layer perceptron neural network, said neural network being trained through backpropagation learning.

15. The system according to claim 8, wherein said multi-layer perceptron neural network includes a hidden layer comprising sigmoid activation functions, and an output layer comprising linear activation functions.

16. The system according to claim 8, wherein the neural network comprises a radial basis function neural network.

17. The system according to claim 16, wherein the radial basis function neural network includes an output layer comprising linear activation functions, and a hidden layer comprising Gaussian activation functions.