Non-invasive method and system for characterizing cardiovascular systems

• US 10,905,345 B2
• Filed: 02/13/2019
• Issued: 02/02/2021
• Est. Priority Date: 08/17/2012
• Status: Active Grant

• Current Assignee: Analytics For Life, Inc.
• Original Assignee: Analytics For Life, Inc.
• Inventors: Gupta, Sunny, Yazdi, Mohsen Najafi, Burton, Timothy William Fawcett, Ramchandani, Shyamlal, Exner, Derek Vincent
• Primary Examiner(s): Mahmood, Nadia A
• Application Number:
  US16/274,803
• Publication Number:
  US 20190313926A1
• Time in Patent Office: 720 Days
• Field of Search: 600523
• CPC Class Codes: A61B 2576/023 for the heart
• CPC Class Codes: A61B 5/0205 Simultaneously evaluating both cardiovascular conditions and different types of body conditions, e.g. heart and respiratory condition
• CPC Class Codes: A61B 5/021 Measuring pressure in heart or blood vessels A61B5/0205 takes precedence
• CPC Class Codes: A61B 5/14551 for measuring blood gases
• CPC Class Codes: A61B 5/316 Modalities, i.e. specific diagnostic methods
• CPC Class Codes: A61B 5/318 Heart-related electrical modalities, e.g. electrocardiography [ECG]
• CPC Class Codes: A61B 5/339 Displays specially adapted therefor
• CPC Class Codes: A61B 5/341 Vectorcardiography [VCG]
• CPC Class Codes: A61B 5/7235 Details of waveform analysis A61B5/349 takes precedence
• CPC Class Codes: G16H 30/40 for processing medical images, e.g. editing

Abstract

The present disclosure uses physiological data, ECG signals as an example, to evaluate cardiac structure and function in mammals. Two approaches are presented, e.g., a model-based analysis and a space-time analysis. The first method uses a modified Matching Pursuit (MMP) algorithm to find a noiseless model of the ECG data that is sparse and does not assume periodicity of the signal. After the model is derived, various metrics and subspaces are extracted to image and characterize cardiovascular tissues using complex-sub-harmonic-frequencies (CSF) quasi-periodic and other mathematical methods. In the second method, space-time domain is divided into a number of regions, the density of the ECG signal is computed in each region and inputted into a learning algorithm to image and characterize the tissues.

Claims

1. A method of extracting variable for use in a machine learning operation to diagnose a pathology, the method comprising:

receiving, via a processor, a biopotential signal data set associated with a subject, said biopotential signal data set being associated with a biopotential signal collected from one or more electrical leads;

determining, via the processor, a plurality of parameters of dynamical features of the received biopotential signal data set for a dictionary that estimates a model of the received biopotential signal data set, wherein the plurality of dynamical features includes Lyapunov exponent and correlation dimension features; and

linking, via the processor, the plurality of determined parameters of the dynamical features to a genetic algorithm for producing outputs that correlate with clinical parameters describing tissue architecture, structure and/or function.

2. The method of claim 1, wherein the model comprises a subspace selected from the group consisting of a CSF trajectory subspace of the received biopotential signal data set, a quasi-periodic and chaotic subspace of the received biopotential signal data set, and a low/high-energy subspace of the received biopotential signal data set.

3. The method of claim 1, wherein the biopotential signal data set comprises electrocardiographic signals.

4. The method of claim 1 further comprising:

removing, by the processor, baseline wander from the received biopotential signal data set.

5. The method of claim 4, wherein the baseline wander is removed using a modified moving average filter.

6. The method of claim 1 comprising:

detecting, by the processor, a disease or pathophysiological abnormality selected from the group consisting of hypertrophy, ischemia, scar, and abnormal electrical channel function.

7. The method of claim 1, wherein the outputs of the genetic algorithms are linked to a 17-myocardial segment model of the ventricle.

8. The method of claim 6 further comprising:

identifying, by the processor, using the determined parameters of the dynamical features, segments of the 17-myocardial segment model with altered architectural features and/or function.

9. The method of claim 8, wherein one or more segments of the 17-myocardial segment model are assigned a probability value of a corresponding tissue having a pathophysiological abnormality.

10. A system to extract variable for use in machine learning operation to diagnose a pathology, the system comprising:

a processor;

a memory having instructions stored thereon, wherein execution of the instructions by the processor causes the processor to;

receive a biopotential signal data set associated with a subject, said biopotential signal data set being associated with a biopotential signal collected from one or more electrical leads;

determine a plurality of parameters of dynamical features of the received biopotential signal data set for a dictionary that estimates a model of the received biopotential signal data set, wherein the plurality of dynamical features includes Lyapunov exponent and correlation dimension; and

link the plurality of determined parameters of the dynamical features to a genetic algorithm for producing outputs that correlate with clinical parameters describing tissue architecture, structure and/or function.

11. The system of claim 10, wherein the model comprises a subspace selected from the group consisting of a CSF trajectory subspace of the received biopotential signal data set, a quasi-periodic and chaotic subspace of the received biopotential signal data set, and a low/high-energy subspace of the received biopotential signal data set.

12. The system of claim 10, wherein the biopotential signal data set comprises electrocardiographic signals.

13. The system of claim 10, wherein execution of the instructions by the processor further causes the processor to remove baseline wander from the received biopotential signal data set.

14. The system of claim 13, wherein the baseline wander is removed using a modified moving average filter.

15. The system of claim 10, wherein execution of the instructions by the processor further causes the processor to detect a disease or pathophysiological abnormality selected from the group consisting of hypertrophy, ischemia, scar, and abnormal electrical channel function.

16. The system of claim 10, wherein the outputs of the genetic algorithms are linked a 17-myocardial segment model of the ventricle.

17. The system of claim 16, wherein execution of the instructions by the processor further causes the processor to identify segments of the 17-myocardial segment model with altered architectural features and/or function.

18. The system of claim 17, wherein one or more segments of the 17-myocardial segment model are assigned a probability of a corresponding tissue having a pathophysiological abnormality.

19. A non-transitory computer readable medium having instructions stored thereon, wherein execution of the instructions by a processor causes the processor to:

receive a biopotential signal data set associated with a subject, said biopotential signal data set being associated with a biopotential signal collected from one or more electrical leads;

determine a plurality of parameters of dynamical features of the received biopotential signal data set for a dictionary that estimates a model of the received biopotential signal data set, wherein the plurality of dynamical features includes Lyapunov exponent and correlation dimension features; and

link the plurality of determined parameters of the dynamical features to a genetic algorithm for producing outputs that correlate with clinical parameters describing tissue architecture, structure and/or function.

20. The computer readable medium of claim 19, wherein the model comprises a subspace selected from the group consisting of a CSF trajectory subspace of the received biopotential signal data set, a quasi-periodic and chaotic subspace of the received biopotential signal data set, and a low/high-energy subspace of the received biopotential signal data set.