METHOD AND SYSTEM FOR GENERATING TWELVE-LEAD ELECTROCARDIOGRAM SIGNALS USING THREE DIFFERENTIAL VOLTAGES

• US 20160157744A1
• Filed: 06/10/2015
• Published: 06/09/2016
• Est. Priority Date: 12/09/2014
• Status: Active Grant

• Current Assignee: National Chiao Tung University (Government of The Republic of China)
• Original Assignee: National Chiao Tung University (Government of The Republic of China)
• Inventors: Wu, Sau-Hsuan, Hsu, Chih-Hao
• Granted Patent: US 9,622,674 B2
• Time in Patent Office: Days
• CPC Class Codes: A61B 5/0006 ECG or EEG signals
• CPC Class Codes: A61B 5/282 Holders for multiple electrodes for introduction into the body A61B5/287
• CPC Class Codes: A61B 5/327 Generation of artificial ECG signals based on measured signals, e.g. to compensate for missing leads
• CPC Class Codes: A61B 5/6823 Trunk, e.g., chest, back, abdomen, hip
• CPC Class Codes: A61B 5/72 Signal processing specially adapted for physiological signals or for diagnostic purposes character recognition, recognising digital ink, document-oriented image-based pattern recognition G06V30/00; recognition or understanding of postal images G06V30/424

Abstract

In a method and system for generating twelve-lead ECG signals based on three differential voltages, each of which is defined as a potential difference between two myoelectric signals sensed respectively by two corresponding electrodes that cooperatively constitute a corresponding differential electrode pair and that are disposed on a user'"'"'s chest at predetermined specific locations, a signal processor generates the twelve-lead ECG signals using a pre-established dynamic system model that is associated with the specific locations of the electrodes on the user'"'"'s chest and that is configured with temporal and spatial correlations among twelve leads of an ECG.

Claims

1. A method of generating twelve-lead electrocardiogram (ECG) signals to be implemented by a signal processor, said method comprising the steps of:

a) receiving three differential voltages, each of which is defined as a potential difference between two corresponding myoelectric signals sensed respectively by two corresponding electrodes that cooperatively constitute a corresponding differential electrode pair that are disposed on a chest of a user'"'"'s body at predetermined specific locations; and

b) generating, based on the differential voltages, the twelve-lead ECG signals, which include twelve synthesized lead signals, using a pre-established dynamic system model that is associated with the specific locations of the electrodes constituting the differential electrode pairs on the chest of the user'"'"'s body and that is configured with temporal and spatial correlations among twelve leads of an ECG, which include eight independent leads and four non-independent leads.

2. The method as claimed in claim 1, further comprising, prior to step a), the steps of:

i) receiving a plurality of training myoelectric signals, which are sensed respectively by a plurality of test electrodes disposed on the chest of the user'"'"'s body respectively at separate points during a training period, and twelve standard lead signals measured from the user'"'"'s body during the training period;

ii) generating, based on the training myoelectric signals, a plurality of training differential voltages that correspond respectively to a plurality of training differential electrode pairs, each of which consists of a respective adjacent pair of the test electrodes that are disposed at respective distinct locations on the chest of the user'"'"'s body, each of the training differential voltages being defined as a potential difference between two of the training myoelectric signals sensed by two corresponding adjacent ones of the test electrodes;

iii) generating, based on the training differential voltages, a plurality of different training differential voltage subsets, each containing three corresponding ones of the training differential voltages;

iv) for each of the training differential voltage subsets, establishing, based on the three corresponding ones of the training differential voltages and on the twelve standard lead signals, a training dynamic system model that is associated with the three corresponding ones of the training differential electrode pairs and that is configured with temporal and spatial correlations among the twelve leads of the ECG, and generating, based on the three corresponding ones of the training differential voltages and the training dynamic system model, a corresponding set of twelve synthesized lead signals, which contain eight independent synthesized lead signals corresponding respectively to the eight independent leads of the ECG, and four non-independent synthesized lead signals corresponding respectively to the four non-independent leads of the ECG; and

v) comparing the corresponding set of the twelve synthesized lead signals of each of the training differential voltage subsets respectively with the twelve standard lead signals so as to obtain comparison results corresponding respectively to the training differential voltage subsets; and

vi) determining, based on the comparison results and the locations of the test electrodes constituting the training differential electrode pairs, one of the training dynamic system models established in step iv) as the dynamic system model used in step b), and determining the locations at which the test electrodes constituting the three corresponding ones of the training differential electrode pairs associated with said one of the training dynamic system models are disposed as the specific locations of the electrodes constituting the differential electrode pairs in step a).

3. The method as claimed in claim 2, wherein, in step i), the number of the test electrodes is twenty-four, and the test electrodes are two-dimensionally arranged in a manner that a distance between any two adjacent ones of the test electrodes ranges from 10 cm to 15 cm.

4. The method as claimed in claim 2, wherein, in step iv), for each of the training differential voltage subsets, the training dynamic system model is defined by the following expressions

x[n]={tilde over (H)}{tilde over (s)}[n]+w[n]

{tilde over (s)}[n]=[s[n]^T,s[n−

1]^T, . . . ,s[n−

L+1]^T]^T

{tilde over (H)}=[H<sub>3×

8</sub>0<sub>3×

8(L-1)</sub>],

where x[n] is a 3×

1 vector and represents the three corresponding ones of the training differential voltages at time n, {tilde over (s)}[n] is an 8L×

1 vector and represents signals at time n that correspond respectively to the eight independent leads, that characterize the temporal and spatial correlations among the eight independent leads of the twelve leads and that conform to an L^th-order autoregression (AR(L)) model, L is a predetermined order number, {tilde over (H)} is a 3×

8L matrix and is associated with a transformation matrix H<sub>3×

8</sub> obtained from the three corresponding ones of the training differential voltages and the twelve standard lead signals, and w[n] is a 3×

1 noise vector at time n and is obtained from the three corresponding ones of the training differential voltages and the twelve standard lead signals.

5. The method as claimed in claim 4, wherein {tilde over (s)}[n] is dynamically represented, using the AR(L) model, by

{tilde over (s)}[n]=F{tilde over (s)}[n−

1]+Bv[n],

where F is an 8L×

8L matrix and characterizes the temporal and spatial correlations among the eight independent leads of the twelve leads, B is an 8L×

8 noise power matrix, and v[n] is an 8×

1 noise vector at time n.

6. The method as claimed in claim 5, wherein, in step iv), for each of the individual training differential voltage subsets, the corresponding set of the twelve synthesized lead signals is generated based on x[n], {tilde over (s)}[n] and {tilde over (H)} using filtering processing.

7. The method as claimed in claim 6, wherein, in step iv), for each of the individual training differential voltage subsets, eight independent synthesized lead signals ŝ

[n] of the corresponding set of the twelve synthesized lead signals is represented by

ŝ

[n]=Fŝ

[n−

1]+K[n](x[n]−

{tilde over (H)}Fŝ

[n−

1]),

where K[n] is an 8L×

3 filter gain matrix at time n, and four non-independent synthesized lead signals of the corresponding set of the twelve synthesized lead signals are generated based on ŝ

[n].

8. The method as claimed in claim 2, wherein, in step vi):

correlation coefficients between the corresponding set of the twelve synthesized lead signals associated with said one of the training dynamic system models and the twelve standard lead signals are relatively large; and

the specific locations of the electrodes constituting the differential electrode pairs are around the heart of the user'"'"'s body.

9. The method as claimed in claim 8, wherein the specific locations of the electrodes constituting a first one of the differential electrode pairs are between the upper right side edge of the heart of the user'"'"'s body and a collarbone of the user'"'"'s body, the specific locations of the electrodes constituting a second one of the differential electrode pairs are on the lower left side edge of the heart of the user'"'"'s body, and the specific locations of the electrodes constituting a third one of the differential electrode pairs are on the right side of the heat of the user'"'"'s body.

11. The system as claimed in claim 1Q, wherein, for each of said signal sensing devices, a distance between said electrodes ranges from 10 cm to 15 cm.

10. A system for generating twelve-lead electrocardiogram (ECG) signals, comprising:

three signal sensing devices attachable to a chest of a user'"'"'s body, each of said signal sensing devices including two spaced apart electrodes that cooperatively constitute a corresponding differential electrode pair and that are to be disposed on the chest of the user'"'"'s body at predetermined specific locations, and being configured to sense two myoelectric signals through said electrodes thereof so as to generate a corresponding differential voltage that is defined as a potential difference between the myoelectric signals; and

a signal processing terminal for receiving the differential voltages respectively therefrom in a wireless manner, said signal processing terminal being operable to generate, based on the differential voltages, the twelve-lead ECG signals, which include twelve synthesized lead signals, using a pre-established dynamic system model that is associated with the specific locations of said electrodes of said signal sensing devices on the chest of the user'"'"'s body and that is configured with temporal and spatial correlations among twelve leads of an ECG, which include eight independent leads and four non-independent leads.

12. The system as claimed in claim 10, wherein the dynamic system model is established in advance based on training data that includes three training differential voltages, which have been generated respectively by said signal sensing devices disposed respectively at the specific locations during a training period, and twelve standard lead signals measured from the user'"'"'s body during the training period.

13. The system as claimed in claim 12, wherein the dynamic system model is defined by the following expressions

x[n]={tilde over (H)}{tilde over (s)}[n]+w[n]

{tilde over (s)}[n]=[s[n]^T,s[n−

1]^T, . . . ,s[n−

L+1]^T]^T

{tilde over (H)}≡

[H<sub>3×

8</sub>0<sub>3×

8(L-1)</sub>],

where x[n] is a 3×

1 vector and represents the three training differential voltages at time n, {tilde over (s)}[n] is an 8L×

1 vector and represents signals at time n that correspond respectively to the eight independent leads, that characterize the temporal and spatial correlations among the eight independent leads and that conforms to an L^th-order autoregression (AR(L)) model, L is a predetermined order number, {tilde over (H)} is a 3×

8L matrix and is associated with a transformation matrix H<sub>3×

8</sub> obtained from the training data, and w[n] is a 3×

1 noise vector at time n and is obtained from the training data.

14. The system as claimed in claim 13, wherein {tilde over (s)}[n] is dynamically represented, using the AR(L) model, by

{tilde over (s)}[n]=F{tilde over (s)}[n−

1]+Bv[n],

where F is an 8L×

8L matrix and characterizes the temporal and spatial correlations among the eight independent leads of the twelve leads, B is an 8L×

8 noise power matrix, and v[n] is an 8×

1 noise vector at time n.

15. The system as claimed in claim 14, wherein the twelve synthesized lead signals are generated based on x[n], {tilde over (s)}[n] and {tilde over (H)} using filtering processing.

16. The system as claimed in claim 15, wherein eight ŝ

[n] of the twelve synthesized lead signals corresponding respectively to the eight independent leads are represented by

ŝ

[n]=Fŝ

[n−

1]+K[n](x[n]−

{tilde over (H)}Fŝ

[n−

1])

where K[n] is an 8L×

3 filter gain matrix at time n, and the remaining four of the twelve synthesized lead signals corresponding respectively to the four non-independent leads are generated based on ŝ

[n].

17. The system as claimed in claim 10, wherein the specific locations of said electrodes of said signal sensing devices on the chest of the user'"'"'s body are disposed around a heart of the user'"'"'s body.

18. The method as claimed in claim 17, wherein the specific locations of said electrodes of a first one of said signal sensing devices are at a position between the upper right side edge of the heart of the user'"'"'s body and a collarbone of the user'"'"'s body, the specific locations of said electrodes of a second one of said signal sensing devices are on the lower left side edge of the heart of the user'"'"'s body and the specific locations of said electrodes of a third one of said signal sensing devices are on the right side of the heart of the user'"'"'s body.

19. The system as claimed in claim 10, wherein said signal processing terminal is a mobile device that is carried on the user'"'"'s body and that receives the differential voltages from said signal sensing devices via short distance wireless communication.

20. The system as claimed in claim 19, further comprising a cloud server that communicates with said mobile device through a communication network;

wherein said mobile device transmits the twelve-lead ECG signals generated thereby to said cloud server via said communication network such that said cloud server collects and records the twelve-lead ECG signals so as to evaluate cardiac function of the user'"'"'s body.

21. The system as claimed in claim 10, further comprising a mobile device that receives the differential voltages from said signal sensing devices via short distance wireless communication and that is connected to said signal processing terminal through a communication network for transmitting the differential voltages thereto;

wherein said signal processing terminal is a cloud server that further stores the twelve-lead ECG signals generated thereby so as to evaluate cardiac function of the user'"'"'s body.