SYSTEM FOR EXTRACTING RESPIRATORY RATES FROM A PULSE OXIMETER

• US 20130197383A1
• Filed: 10/12/2011
• Published: 08/01/2013
• Est. Priority Date: 10/12/2010
• Status: Active Grant

• Current Assignee: Worcester Polytechnic Institute
• Original Assignee: Worcester Polytechnic Institute
• Inventors: Chon, Ki H., Lee, Jinseok
• Granted Patent: US 9,888,866 B2
• Time in Patent Office: Days
• US Class Current: 600/529
• CPC Class Codes: A61B 5/0816 Measuring devices for examining respiratory frequency measuring frequency of electric signals G01R23/00
• CPC Class Codes: A61B 5/14551 for measuring blood gases
• CPC Class Codes: A61B 5/725 using specific filters therefor, e.g. Kalman or adaptive filters A61B5/316 take precedence

Abstract

Accurate AR method for extracting respiratory rates directly from a pulse oximeter and accurate methods of extracting respiratory rates directly from a pulse oximeter under low signal-to-noise ratio (SNR) conditions.

Claims

1. A method for detecting respiratory rate of a patient, the method comprising the steps of:

obtaining parameters for an autoregressive (AR) representation of a photoplethysmography (PPG) signal obtained from the patient, the parameters being obtained using a projection onto linearly independent non-orthogonal bases;

obtaining poles for a frequency transfer function of the AR representation;

selecting a pole with a highest magnitude in a frequency region of interest;

the respiratory rate being determined by the pole with a highest magnitude.

2. The method of claim 1 wherein the frequency region of interest is between about 0.15 Hz and about 0.9 Hz.

3. The method of claim 1 wherein the frequency region of interest is between about 0.15 Hz and about 1.5 Hz.

4. The method of claim 1 further comprising the step of filtering the PPG signal before obtaining parameters for the AR representation.

5. The method of claim 4 further comprising the step of downsampling the filtered PPG signal.

6. A method for detecting respiratory rate of a patient, the method comprising the steps of:

obtaining, from an autoregressive (AR) representation of a photoplethysmography (PPG) signal obtained from the patient, the respiratory rate of the patient using an iterative Particle Filtering Monte Carlo method, where a distribution of respiratory rate is the distribution approximated by Monte Carlo sampling, measurements are magnitudes and phase of poles for a frequency transfer function of the AR representation, the magnitudes and phase of the poles obtained by;

obtaining parameters for the AR representation using a projection onto linearly independent non-orthogonal bases;

obtaining poles for a frequency transfer function of the AR representation.

7. The method of claim 6 wherein the iterative Particle Filtering Monte Carlo method comprises:

a. selecting an initial group of particles, each particle from said initial group being a different value for the respiratory rate;

each particle from said initial group being assigned a weight equal to an inverse of a number of particles said initial group;

b. obtaining a new group of particles by propagating the initial group of particles;

c. obtaining measurements;

measurements being magnitudes and phase of poles for the frequency transfer function of the AR representation obtained by;

obtaining parameters for the AR representation using a projection onto linearly independent non-orthogonal bases;

obtaining the poles for a frequency transfer function of the AR representation;

d. obtaining a new weight for each particle from said new group by using a predetermined likelihood function;

e. select a candidate respiratory rate of a patient equal to a weighted sum of the particles in the new group of particles; and

f. resampling the new group of particles.

8. The method of claim 7 wherein the, predetermined likelihood function is a strongest neighbor (SN) likelihood function.

9. The method of claim 7 wherein the predetermined likelihood function is a nearest neighbor (NN) likelihood function.

10. The method of claim 7 wherein the predetermined likelihood function is a weighted nearest neighbor (WNN) likelihood function.

11. The method of claim 7 wherein the predetermined likelihood function is a probabilistic data association (PDA) likelihood function.

12. The method of claim 7 wherein the predetermined likelihood function is a weighted probabilistic data association (WPDA) likelihood function.

13. The method of claim 6 wherein a frequency region of interest is between about 0.15 Hz and about 1.5 Hz.

14. A system comprising:

at least one processor;

said at least one processor receiving a photoplethysmography (PPG) signal obtained from the patient, and

a computer usable memory having computer readable code embodied therein, the computer readable code causing said at least one processor to;

obtain parameters for an autoregressive (AR) representation of a photoplethysmography (PPG) signal obtained from the patient, the parameters being obtained using projection onto linearly independent non-orthogonal bases;

obtain poles for a frequency transfer function of the AR representation;

select a pole with a highest magnitude in a frequency region of interest;

a respiratory rate being determined by the pole with a highest magnitude.

15. The system of claim 14 wherein the frequency region of interest is between about 0.15 Hz and about 0.9 Hz.

16. The system of claim 14 wherein the frequency region of interest is between about 0.15 Hz and about 1.5 Hz.

17. The system of claim 14 wherein the computer readable code is also capable of causing said at least one processor to filter the PPG signal before obtaining parameters for the AR representation.

18. The system of claim 17 wherein the computer readable code is also capable of causing said at least one processor to downsample the filtered PPG signal.

19. A system comprising:

at least one processor;

said at least one processor receiving a photoplethysmography (PPG) signal obtained from the patient, and

a computer usable memory having computer readable code embodied therein, the computer readable code causing said at least one processor to;

obtain, from an autoregressive (AR) representation of a photoplethysmography (PPG) signal obtained from the patient, a respiratory rate of the patient using an iterative Particle Filtering Monte Carlo method, where a distribution of respiratory rate is the distribution approximated by Monte Carlo sampling, measurements are magnitudes and phase of poles for a frequency transfer function of the AR representation, the magnitudes and phase of the poles obtained by;

obtaining parameters for the AR representation using a projection onto linearly independent non-orthogonal bases;

obtaining the poles for a frequency transfer function of the AR representation.

20. The system of claim 19 wherein the iterative Particle Filtering Monte Carlo method comprises:

a. selecting an initial group of particles, each particle from said initial group being a different value for the respiratory rate;

each particle from said initial group being assigned a weight equal to an inverse of a number of particles said initial group;

b. obtaining a new group of particles by propagating the initial group of particles;

c. obtaining measurements;

measurements being magnitudes and phase of poles for the frequency transfer function of the AR representation obtained by;

obtaining parameters for the AR representation using a projection onto linearly independent non-orthogonal bases;

obtaining the poles for a frequency transfer function of the AR representation;

d. obtaining a new weight for each particle from said new group by using a predetermined likelihood function;

e. select a candidate respiratory rate of a patient equal to a weighted sum of the particles in the new group of particles; and

f. resampling the new group of particles.

21. The system of claim 20 wherein the predetermined likelihood function is a strongest neighbor (SN) likelihood function.

22. The system of claim 20 wherein the predetermined likelihood function is a nearest neighbor (NN) likelihood function.

23. The system of claim 20 wherein the predetermined likelihood function is a weighted nearest neighbor (WNN) likelihood function.

24. The system of claim 20 wherein the predetermined likelihood function is a probabilistic data association (PDA) likelihood function.

25. The system of claim 20 wherein the predetermined likelihood function is a weighted probabilistic data association (WPDA) likelihood function.

26. The system of claim 19 wherein a frequency region of interest is between about 0.15 Hz and about 1.5 Hz.

27. A computer program product comprising:

a non-transitory computer usable medium having computer readable code embodied therein for detecting respiratory rate of a patient, the computer readable code causing at least one processor to;

obtain parameters for an autoregressive (AR) representation of a photoplethysmography (PPG) signal obtained from the patient, the parameters being obtained using a projection onto linearly independent non-orthogonal bases;

obtain poles for a frequency transfer function of the AR representation;

select a pole with a highest magnitude in a frequency region of interest;

the respiratory rate being determined by the pole with a highest magnitude.

28. The computer program product of claim 27 wherein the computer readable code is also capable of causing said at least one processor to filter the PPG signal before obtaining parameters for the AR representation.

29. The computer program product of claim 28 wherein the computer readable code is also capable of causing said at least one processor to downsample the filtered. PPG signal.

30. A computer program product comprising:

a non-transitory computer usable medium having computer readable code embodied therein for detecting respiratory rate of a patient, the computer readable code causing at least one processor to;

obtain, from an autoregressive (AR) representation of a photoplethysmography (PPG) signal obtained from the patient, the respiratory rate of the patient using an iterative Particle Filtering Monte Carlo method, where a distribution of respiratory rate is the distribution approximated by Monte Carlo sampling, measurements are magnitudes and phase of poles for a frequency transfer function of the AR representation, the magnitudes and phase of the poles obtained by;

obtaining parameters for the AR representation using a projection onto linearly independent non-orthogonal bases; and

obtaining the poles for a frequency transfer function of the AR representation.

31. The computer program product of claim 30 wherein the iterative Particle Filtering Monte Carlo method comprises:

a. selecting an initial group of particles, each particle from said initial group being a different value for the respiratory rate;

each particle from said initial group being assigned a weight equal to an inverse of a number of particles said initial group;

b. obtaining a new group of particles by propagating the initial group of particles;

c. obtaining measurements;

measurements being magnitudes and phase of poles for the frequency transfer function of the AR representation obtained by;

obtaining parameters for the AR representation using a projection onto linearly independent non-orthogonal bases;

obtaining the poles for a frequency transfer function of the AR representation;

d. obtaining a new weight for each particle from said new group by using a predetermined likelihood function;

e. select a candidate respiratory rate of a patient equal to a weighted sum of the particles in the new group of particles; and

f. resampling the new group of particles.

32. The computer program product of claim 31 wherein the predetermined likelihood function is a strongest neighbor (SN) likelihood function.

33. The computer program product of claim 31 wherein the predetermined likelihood function is a nearest neighbor (NN) likelihood function.

34. The computer program product of claim 31 wherein the predetermined likelihood function is a weighted nearest neighbor (WNN) likelihood function.

35. The computer program product of claim 31 wherein the predetermined likelihood function is a probabilistic data association (PDA) likelihood function.

36. The computer program product of claim 31 wherein the predetermined likelihood function is a weighted probabilistic data association (WPDA) likelihood function.