Method and apparatus for processing a physiological signal

• US 6,339,715 B1
• Filed: 09/30/1999
• Issued: 01/15/2002
• Est. Priority Date: 09/30/1999
• Status: Expired due to Term

• Current Assignee: Prothia
• Original Assignee: OB Scientific Incorporated
• Inventors: Bahr, Dennis E., Reuss, James L.
• Primary Examiner(s): Winakur, Eric F.
• Assistant Examiner(s): KREMER, MATTHEW J
• Application Number:
  US09/410,887
• Time in Patent Office: 838 Days
• Field of Search: 600/323, 600/330, 600/336, 600/473, 600/476, 356/39-41
• US Class Current: 600/323
• CPC Class Codes: A61B 5/14542 for measuring blood gases A61B5/14551 takes precedence
• CPC Class Codes: A61B 5/1455 using optical sensors, e.g. spectral photometrical oximeters
• CPC Class Codes: A61B 5/1464 specially adapted for foetal tissue
• CPC Class Codes: A61B 5/7257 using Fourier transforms

Abstract

A signal processing method, preferably for extracting a fundamental period from a noisy, low-frequency signal, is disclosed. The signal processing method generally comprises calculating a numerical transform for a number of selected periods by multiplying signal data by discrete points of a sine and a cosine wave of varying period and summing the results. The period of the sine and cosine waves are preferably selected to have a period substantially equivalent to the period of interest when performing the transform.

Claims

1. A method for determining the estimated power values of periodic components of a sampled input signal, with periods of the periodic components of the sampled input signal defined at a resolution equal to a sampling rate of the input signal, the method comprising the following steps:

(a) selecting a sampling period equal to the inverse of the sampling rate of the input signal;

(b) selecting a minimum period and a maximum period representing a content of the sampled input signal over a range of the minimum period and the maximum period;

(c) selecting a first periodic waveform and a second periodic waveform, the first periodic waveform and the second periodic waveform being substantially orthogonal to each other, each periodic waveform having a periodic waveform period equal to the maximum period;

(d) varying an incremental step over the range of the minimum period and the maximum period in the sampled input signal, yielding a plurality of incremental steps, having a difference between consecutive incremental steps corresponding to the sampling period, and;

(e) determining a transform at each of the plurality of incremental steps by;

(i) calculating an index into the first and second periodic waveforms, the index computed from an indexing increment inversely proportional to the incremental step, yielding a first decimated periodic waveform and a second decimated periodic waveform, each decimated periodic waveform having a decimated periodic waveform period equal to the incremental step;

(ii) summing the values resultant from multiplying the sampled input signal by a plurality of points spaced equidistantly on the first decimated periodic waveform, yielding first periodic waveform summed values;

(iii) summing the values resultant from multiplying the sampled input signal by a plurality of points spaced equidistantly on the second decimated periodic waveform, yielding second periodic waveform summed values;

(iv) applying a scaling factor to the periodic waveform summed values, yielding scaled sums, the scaling factor selected providing equal weighting of the periodic waveform summed values, and;

(v) combining the scaled sums to calculate a relative power at each incremental step, yielding an array of estimated power values representing a power spectrum of the periodic components in the sampled input signal over the range of incremental steps from the minimum period to the maximum period.

2. The method as defined in claim 1, further comprising the step of reducing the computational overhead by performing the scaling and calculation of the array of estimated power values at a lower frequency than the input signal sampling rate.

3. The method as defined in claim 1, further comprising the step of identifying and tracking a period of interest corresponding to a dominant period of the sampled input data through time, the method comprising the following steps:

(a) searching for a peak power period by scanning the array of estimated power values for a local peak of maximum amplitude;

(b) accepting or rejecting a period corresponding to the local peak of maximum amplitude by utilizing qualifying information;

(c) continuing to search for the peak power period if rejected, else tracking the period of interest in the array of estimated power values if accepted; and

(d) reverting to searching for a peak power period if the tracking of the period of interest is lost.

4. The method as defined in claim 3, where the qualifying information is based upon time domain analysis of the input signal.

5. The method as defined in claim 3, where the qualifying information is based upon a priori knowledge of the signal sources.

6. The method as defined in claim 3, where the tracking of the period of interest is lost if the distribution of power in the array of estimated power values is too diffuse.

7. The method as defined in claim 3, where the tracking of the period of interest is lost if the power around the period of interest in the array of estimated power values indicates no strong peak is present.

8. The method as defined in claim 3, where the qualifying information is further utilized to find other possible tracking candidates if the peak is not qualified for use.

9. The method as defined in claim 3, further comprising the step of averaging the results of tracking the period of interest to reduce jitter and improve the resolution of the period of interest.

10. The method as defined in claim 3, further comprising the step of performing a rate calculation in order to derive a pulse rate based upon the period of interest tracked in the input data.

11. The method as defined in claim 3, further comprising the step of calculating an ensemble average based upon the period of interest tracked in the input data in order to extract a signal of a particular period from a combination of signal and noise in the input signal.

12. The method as defined in claim 11, wherein the ensemble average is calculated as a running average.

13. The method as defined in claim 1, wherein the sampled input signal is a plethysmographic wave.

14. The method as defined in claim 1, wherein the sampled input signal is a plethysmographic signal from a pulse oximetry sensor.

15. The method as defined in claim 1, wherein the sampled input signal is a plethysmographic signal from a fetal pulse oximetry sensor.

16. The method as defined in claim 1, wherein, prior to the step of applying a scaling factor, the method further comprising the step of sliding the sampled input signal through the transform such that a new data point is added and an old data point is subtracted for each iteration.

17. A method for calculating a digital numerical transform for a low frequency periodic signal at a resolution equal to a sampling rate of the periodic signal thereby deriving power versus period in real-time, the method comprising the following steps:

obtaining input data by sampling data points from a continuous periodic waveform at a predetermined time interval;

selecting a period, wherein each selected period is between a pre-selected minimum period and a pre-selected maximum period, wherein each selected period is separated from the adjacent period by the predetermined time interval;

for each selected period;

(a) accessing stored points of a sine wave and a cosine wave of the pre-selected maximum period taken at the time interval;

calculating an index to the sine and cosine waves, where the index is inversely proportional to the selected period; and

indexing the sine wave and the cosine wave to selectively vary the period of the sine and cosine waves;

(b) calculating an imaginary transform component value equal to input data times a selected point on a sine wave having a period equal to the selected period;

(c) adding to a sum of previous imaginary transform component values for the same selected period;

(d) calculating a real transform component value equal to input data times a selected point on a cosine wave having a period equal to the selected period;

(e) adding to a sum of previous real transform component values for the same selected period;

(f) scaling the real and the imaginary transform component sums to give all selected periods equal weighting and adding the squares of the sums of real and imaginary transform components for a given selected period to arrive at a relative power corresponding to that selected period in the input signal; and

(g) incrementing the selected point on the sine and cosine waves;

repeating while calculations are desired on a successive sample of the input data.

18. The method as defined in claim 17, wherein the input data is initially calculated as a difference between the oldest data sample in a buffer and the newest data sample in the buffer and the input data is continually updated by adding the newest data sample and subtracting the oldest data sample.

19. An oximeter comprising:

an input port for receiving a plethysmographic signal from a sensor;

an analog to digital converter;

a processor programmed to calculate a numerical transform of the plethysmographic signal to determine a fundamental period of a cardiac activity, wherein the numerical transform is calculated at each of a plurality of incremental periods by multiplying samples of the plethysmographic signal by first and second periodic orthogonal waveforms of period equal to the selected incremental period, wherein the processor is further programmed to index through a buffer storing the first periodic waveform and a buffer storing the second periodic waveform, where the index is inversely proportional to a selected period to selectively decimate the length of the period; and

wherein the processor is further programmed to scale and to sum the calculated numerical transforms for a given incremental period to yield the fundamental period of the cardiac activity in the plethysmographic signal.

20. The oximeter as defined in claim 19, wherein the sensor is a fetal sensor.

21. An oximeter comprising:

an input port for receiving a plethysmographic signal from a sensor;

an analog to digital converter;

a processor programmed to calculate a numerical transform of the plethysmographic signal to determine a fundamental period of a cardiac activity, wherein the numerical transform is calculated at each of a plurality of incremental periods by multiplying samples of the plethysmographic signal by first and second periodic orthogonal waveforms of period equal to the selected incremental period,

wherein the processor is further programmed to calculate a cardiac rate based on a counted number of pulsatile components found in a blood oxygenation calculation, and

wherein the processor is programmed to calculate the numerical transform only when the pulsatile components of the cardiac rate determined by the oxygenation calculation is not significantly greater than background noise.

22. The oximeter as defined in claim 21, wherein the sensor is a fetal sensor.

23. A method for determining a cardiac signal from a noisy plethysmographic input signal, the method comprising the following steps:

(a) sampling the input plethysmographic signal at a sampling rate having a predetermined time interval;

(b) calculating a numerical transform of the plethysmographic signal for each of a plurality of periods up to a maximum period where each of the periods are separated by the predetermined time interval;

by;

(i) accessing successive stored sampled points of a first periodic waveform and successive stored sampled points of a second periodic waveform, where the first periodic waveform and the second periodic waveform each have a period equal to the maximum period, where the first and second periodic waveforms are orthogonal to each other and where the sampled points are separated by a time interval equivalent to the sampling rate;

(ii) calculating an index to the stored sampled points of the first and second periodic waveforms where the index is inversely proportional to the period such that for small periods the index into the stored data is large and for large periods the index is small;

(iii) calculating a transform component value equal to the sampled plethysmographic signal times a selected point on the first periodic waveform having a period equal to the period;

(iv) adding to a sum of previous transform component values for the same period for the first periodic waveform;

(v) calculating a transform component value equal to the sampled plethysmographic signal times a selected point on the second periodic waveform having a period equal to the period;

(vi) adding to a sum of previous transform component values for the same period for the second periodic waveform;

(vii) incrementing the selected point on the first and second periodic waveforms; and

(viii) scaling the sums of the transform components to give all selected period equal weighting;

(ix) repeating steps (i) through (viii) while calculations are desired;

(c) adding the squares of the sums of the transform component values for a given period to arrive at a relative power corresponding to that period in the input plethysmographic signal; and

(d) calculating a relative strength for each period as the power of each period.

24. The method as defined in claim 23, further comprising the step of adding a new data point and subtracting an old data point at each time interval.

25. A method for calculating a digital numerical transform for a low frequency periodic signal at a resolution equal to a sampling rate of the periodic signal thereby deriving power versus period in real-time, the method comprising the following steps:

obtaining input data by sampling data points from a continuous periodic waveform at a predetermined time interval;

selecting a period, wherein each selected period is between a pre-selected minimum period and a pre-selected maximum period, wherein each selected period is separated from the adjacent period by the predetermined time interval;

for each selected period;

(a) accessing stored points of a sine wave and a cosine wave of the pre-selected maximum period taken at the time interval;

calculating an index to the sine and cosine waves, where the index is inversely proportional to the selected period; and

indexing the sine wave and the cosine wave to selectively vary the period of the sine and cosine waves;

(b) calculating a real transform component value equal to input data times a selected point on a cosine wave having a period equal to the selected period;

(c) adding to a sum of previous real transform component values for the same selected period;

(d) calculating an imaginary transform component value equal to input data times a selected point on a sine wave having a period equal to the selected period;

(e) adding to a sum of previous imaginary transform component values for the same selected period;

(f) incrementing the selected point on the sine and cosine waves;

(g) scaling the real and the imaginary transform component sums to give all selected periods equal weighting and adding the squares of the sums of real and imaginary transform components for a given selected period to arrive at a relative power corresponding to that selected period in the input signal

repeating while calculations are desired on a successive sample of the input data.