US20120245441A1

US20120245441A1 - Signal demodulation

Info

Publication number: US20120245441A1
Application number: US13/488,072
Authority: US
Inventors: Ethan Petersen
Original assignee: Nellcor Puritan Bennett LLC
Current assignee: Covidien LP
Priority date: 2007-12-13
Filing date: 2012-06-04
Publication date: 2012-09-27
Also published as: US8204567B2; US20090154573A1

Abstract

A method for processing an analog composite signal in a system has the steps of receiving a composite signal with at least one first signal component and at least one interfering signal component; filtering the composite signal with a filter having a transfer function H(s); sampling the filtered composite signal in periodic intervals wherein each periodic interval has n samples; forming a matrix equation representing the composite signal wherein the matrix equation has a signal vector with the at least first one signal component and the at least one interfering signal component and a matrix comprising weighted coefficients; solving the matrix equation to determine the at least one signal component; outputting the at least one signal component.

Description

CROSS-REFERENCE TO RELATED APPLICATION

This application is a continuation of application Ser. No. 11/955,824, filed Dec. 13, 2007, entitled “Signal Demodulation” in the name of Ethan Petersen and assigned to Nellcor Puritan Bennett LLC, which is incorporated by reference herein in its entirety.

TECHNICAL FIELD

The technical field of the present application relates to oximeter signal processing.

BACKGROUND

Pulse oximeters are used to indirectly measures the amount of oxygen in a patient's blood and for measuring the pulse of a patient. Furthermore, they can be used to measure changes in blood volume in the skin, producing a photoplethysmograph. Pulse oximeters are usually attached to a medical monitor so staff can see a patient's oxygenation at all times. Most monitors also display in addition the heart rate.
A pulse oximeter is a particularly convenient non-invasive measurement instrument. Typically it has a pair of small light-emitting diodes (LEDs) facing a photodiode through a translucent part of the patient's body, usually a fingertip or an earlobe. One LED is red, with wavelength of approximately 660 nm, and the other is infrared, using a wavelength of approximately 905, 910, or 940 nm. Absorption at these wavelengths differs significantly between oxyhemoglobin and its deoxygenated form, therefore from the ratio of the absorption of the red and infrared light the oxy/deoxyhemoglobin ratio can be calculated.
The monitored signal is modulated by the heart beat because the arterial blood vessels expand and contract with each heartbeat. Oximeters are furthermore subject to various interferences. For example, ambient light, in particular light emitted from fluorescent lighting, can introduce a significant interfering signal. Capacitive coupling in the patient cable between the LED wires and the detector wires is also a large source of additional errors. Generally on the rising and falling edges of the LED voltage an impulse current appears in the detector lines due to this capacitive coupling.

SUMMARY

According to an embodiment, a method for processing an analog composite signal in a system, may comprise the steps of; receiving a composite signal comprising at least one first signal component and at least one interfering signal component; filtering the composite signal with a filter having a transfer function H(s); sampling the filtered composite signal in periodic intervals wherein each periodic interval comprises n samples; forming a matrix equation representing the composite signal wherein the matrix equation comprises a signal vector comprising the at least first one signal component and the at least one interfering signal component and a matrix comprising weighted coefficients; solving the matrix equation to determine the at least one signal component; and outputting the at least one signal component.
According to a further embodiment, an interfering signal component can be approximated by a linear approximation between a first and last sample of each periodic interval. According to a further embodiment, the weighted coefficients for a transient interfering signal component can be determined by the transfer function and a sample position within a periodic interval. According to a further embodiment, the system can be an oximeter system comprising an oximeter sensor generating a Red signal component and an Infrared (IR) signal component as signal components, wherein an ambient light signal component and cable transients may be interfering signal components. According to a further embodiment, the ambient light signal component may be approximated by a linear approximation between a first and last sample of each periodic interval. According to a further embodiment, the weighted coefficients for the cable transients may be determined by the transfer function and a sample position within a periodic interval. According to a further embodiment, the Red signal component and the IR signal component may be timely separated within each periodic interval and the Red signal component and the IR signal component each may comprise a predetermined signal length having an on and off transient. According to a further embodiment, the ambient light signal component may be approximated by a linear approximation between a first and last sample of each periodic interval and the weighted coefficients for the cable transients are determined by the transfer function and a sample position within a periodic interval, and wherein a coefficient matrix may comprise first and second coefficients for the linear approximation, switch on and switch off coefficients for the cable transients, a Red coefficient, and an IR coefficient.
According to another embodiment, a system for processing an analog composite signal comprising at least one first signal component and at least one interfering signal component, may comprise: a filter having a transfer function H(s) receiving the composite signal and outputting a filtered composite signal; an analog-to-digital converter receiving the filtered composite signal and sampling the filtered composite signal in periodic intervals wherein each periodic interval comprises n samples; and a signal processor receiving the sampled filtered composite signal, wherein the signal processor forms a matrix equation representing the composite signal wherein the matrix equation comprises a signal vector comprising the at least first one signal component and the at least one interfering signal component and a matrix comprising weighted coefficients, wherein the signal processor is furthermore operable to solve the matrix equation to calculate the at least one signal component and to output the at least one signal component.
According to a further embodiment, an interfering signal may be approximated by a linear approximation between a first and last sample of each periodic interval. According to a further embodiment, the weighted coefficients for a transient interfering signal component may be determined by the transfer function and a sample position within a periodic interval. According to a further embodiment, the system can be an oximeter system comprising an oximeter sensor generating a Red signal component and an Infrared (IR) signal component as signal components, wherein an ambient light signal component and cable transients are interfering signal components. According to a further embodiment, the ambient light signal component may be approximated by a linear approximation between a first and last sample of each periodic interval. According to a further embodiment, the weighted coefficients for the cable transients may be determined by the transfer function and a sample position within a periodic interval. According to a further embodiment, the Red signal component and the IR signal component may be timely separated within each periodic interval and the Red signal component and the IR signal component each may comprise a predetermined signal length having an on and off transient. According to a further embodiment, the ambient light signal component may be approximated by a linear approximation between a first and last sample of each periodic interval and the weighted coefficients for the cable transients may be determined by the transfer function and a sample position within a periodic interval, wherein a coefficient matrix may comprise first and second coefficients for the linear approximation, switch on and switch off coefficients for the cable transients, a Red coefficient, and an IR coefficient.
According to yet another embodiment, an oximeter system may comprise an oximeter sensor generating an output signal with a Red signal component and an Infrared (IR) signal component which are timely separated within a periodic interval wherein the Red signal component and the IR signal component each comprise a predetermined signal length having an on and off transient, a filter having a transfer function H(s) receiving a composite signal consisting of the oximeter sensor output signal and at least one interfering signal component, wherein the filter outputs a filtered composite signal; an analog-to-digital converter receiving the filtered composite signal and sampling the filtered composite signal in periodic intervals wherein each periodic interval comprises n samples; and a signal processor receiving the sampled filtered composite signal, wherein the signal processor forms a matrix equation representing the composite signal wherein the matrix equation comprises a signal vector comprising the Red and IR signal components and the at least one interfering signal component and a matrix comprising weighted coefficients, wherein the signal processor is furthermore operable to solve the matrix equation to calculate the Red and IR signal components and to output the Red and IR signal components.
According to a further embodiment, an ambient light signal component and cable transients may be interfering signal components. According to a further embodiment, the ambient light signal component may be approximated by a linear approximation between a first and last sample of each periodic interval. According to a further embodiment, the weighted coefficients for the cable transients may be determined by the transfer function and a sample position within a periodic interval. According to a further embodiment, the ambient light component may be approximated by a linear approximation between a first and last sample of each periodic interval and the weighted coefficients for the cable transients are determined by the transfer function and a sample position within a periodic interval, wherein a coefficient matrix may comprise first and second coefficients for the linear approximation, switch on and switch off coefficients for the cable transients, a Red coefficient, and an IR coefficient.
According to yet another embodiment, a method for processing an analog composite signal in an oximeter system, may comprise the steps of: receiving a composite signal comprising at Red signal component and an infrared (IR) signal component from an oximeter sensor and at least one interfering signal component; filtering the composite signal with a filter having a transfer function H(s); sampling the filtered composite signal in periodic intervals wherein each periodic interval comprises n samples; forming a matrix equation representing the system wherein the matrix equation comprises a signal vector comprising the Red and IR signal component and the at least one interfering signal component and a matrix comprising weighted coefficients; solving the matrix equation to calculate the Red and IR signal components; and outputting the Red and IR signal components.
According to a further embodiment, an ambient light signal component and cable transients may be interfering signal components, wherein the ambient light can be approximated by a linear approximation between a first and last sample of each periodic interval and wherein the weighted coefficients for the cable transients may be determined by the transfer function and a sample position within a periodic interval. According to a further embodiment, the Red signal component and the IR signal component can be timely separated within each periodic interval and the Red signal component and the IR signal component each may comprise a predetermined signal length having an on and off transient. According to a further embodiment, the ambient light can be approximated by a linear approximation between a first and last sample of each periodic interval and the weighted coefficients for the cable transients are determined by the transfer function and a sample position within a periodic interval, and wherein a coefficient matrix comprises first and second coefficients for the linear approximation, switch on and switch off coefficients for the cable transients, a Red coefficient, and an IR coefficient.
Other technical advantages of the present disclosure will be readily apparent to one skilled in the art from the following figures, descriptions, and claims. Various embodiments of the present application obtain only a subset of the advantages set forth. No one advantage is critical to the embodiments.

BRIEF DESCRIPTION OF THE DRAWINGS

A more complete understanding of the present disclosure and advantages thereof may be acquired by referring to the following description taken in conjunction with the accompanying drawings, in which like reference numbers indicate like features, and wherein:

FIG. 1 is a block diagram of a typical oximeter arrangement, and the sources of interfering signals;

FIG. 2 depicts the various components of an oximeter input signal;

FIG. 3 an example of the impulse response of a filter to a chain of impulses.

FIG. 4 shows how a piecewise linear approximation is applied to a signal representing ambient light;

FIG. 5 shows how the signal is processed after it is digitized.

DETAILED DESCRIPTION

As stated above, oximeter detectors are subject to a variety of interfering signals. Currently the largest source of error in the electronics of a pulse oximeter arises from capacitive coupling in the patient cable between the LED wires and the detector wires. On the rising and falling edges of the LED voltage an impulse current appears in the detector lines due to this capacitive coupling. For example, it has been determined that the error in measured photo current due to such impulse currents can be up to around 71 pA, for example, out of a batch of 17 new cables. Older cables that have been worn will have compromised shields that could result in a much larger error.
FIG. 1 shows a block diagram explaining the influences of the main interfering sources in oximeter systems. Generally, an oximeter sensor comprises a red LED and IR LED whose emitted light is passed through a patient's tissue. A detector receives these signals but also receives some ambient light as shown on the left side of FIG. 1. Node 110 simulates a summing point within the photo detector or detectors of an oximeter system. Thus, the photo detector produces a signal 120 which comprises the RED component, the IR component, and an ambient light component. Node 130 simulates the summing point of capacitive cable transient signals introduced into the detector signal. Thus, output signal 140 now comprises in addition to the signals mentioned above, the cable transient signals. Signal 140 is then fed into filter 150 comprising a transfer function H(s). The output signal of filter 150 is then fed to an analog-to-digital-converter 160.
FIG. 2 shows exemplary signal curves for each signal component as shown in FIG. 1 as well as the composite signal. For each Red and IR signal pulse, according to an embodiment, 8 samples P1 . . . 8 are taken as indicated on the bottom x-axis. During the time frame P1 . . . 8, the ambient light, shown as the dotted line which can be dominated by components of the 50 Hz/60 Hz power line signals, is approximated by a linear line as shown in the top curve between points X1 and X2. The transient pulses caused by the rising and falling edges of the Red and IR signals are shown as signals W1, W2, W3, and W4. Next follows the Red signal and then the IR signal. The bottom curve represents the composite signal as it is fed to the filter 150. This composite signal represents a sum of the above signals.
This signal is then sampled by an analog-to-digital converter 160 as indicated at the bottom line of FIG. 2. As shown in FIG. 2, 8 samples are produced for each Red and IR pulse. However, according to other embodiments, more than eight samples can be generated which will improve performance. The composite signal which is filtered by filter 150 and sampled by analog-to-digital-converter 160 comprises the component signals as discussed with respect to FIG. 3. Thus, each component signal is first filtered before it is sampled by analog-to-digital converter 160. The filter is used for anti-aliasing and to help eliminate out of band noise. Thus, the filter 150 has a transfer function of H(s) that spreads out the composite signal in the time domain. Since the filter 150 is a linear system, each of the components can be analyzed by assuming they have all gone through the filter independently. The result is that an impulse will have energy spread across all the sample periods.
FIG. 3 shows an exemplary output signal from signal filter 150 to which a series of periodic pulses W1 is fed. The respective sample points P1 . . . 8 produced by the analog-to-digital-converter 160 resulting from the pulses W1 fed to filter 150 are shown in FIG. 3 by the vertical lines ending with a crossbar. The magnitude of the sample at each sampling point is, thus, a function of the magnitude of the impulse W1 and the impulse response of the system. Since the time between the impulse W1 and the sample time is constant, the size of the sample at P1 is a constant times the magnitude of the impulse. This results in:
$\begin{matrix} P 1 = k 1 \cdot W 1 \\ P 2 = k 2 \cdot W 1 \\ P 3 = k 3 \cdot W 1 \\ ⋮ \\ and so on . \end{matrix}$
The results for the Red and IR components of the composite signal can be represented in a similar way, as a constant representing the impulse response at that time multiplied by the current. This results in:


	Red component	IR component

	P1 = c1 · R	P1 = b5 · I
	P2 = c2 · R	P2 = b6 · I
	P3 = c3 · R	P3 = b7 · I
	$⋮$	$⋮$
	and so on.

The component of the signal representing the ambient light can be approximated for a sample period (P1 . . . P8) by a linear approximation A_nbetween points X1 and X2 as shown in FIG. 4, wherein point X1 is associated with sample time P1 and X2 is associated with sample time P8. A new approximation A_n+1follows for the next eight samples as indicated in FIG. 4. The terms for ambient light only can, thus, be represented as:
$\begin{matrix} P 1 = X 1 \\ P 2 = \frac{6}{7} X 1 + \frac{1}{7} X 2 \\ P 3 = \frac{5}{7} X 1 + \frac{2}{7} X 2 \\ P 4 = \frac{4}{7} X 1 + \frac{2}{7} X 2 \\ ⋮ \\ and so on . \end{matrix}$
The magnitude of the sample for the composite signal is the sum of all components. For instance:
P1=1·X1+0·X2+k1·W1+k7·W2+k5·W3+k3·W4+c1·R+b5·I
The whole system can, thus, be represented in matrix form as:
$[\begin{matrix} P 1 \\ P 2 \\ P 3 \\ P 4 \\ P 5 \\ P 6 \\ P 7 \\ P 8 \end{matrix}] = [\begin{matrix} \frac{7}{7} & \frac{0}{7} & k 1 & k 7 & k 5 & k 3 & c 1 & b 5 \\ \frac{6}{7} & \frac{1}{7} & k 2 & k 8 & k 6 & k 4 & c 2 & b 6 \\ \frac{5}{7} & \frac{2}{7} & k 3 & k 1 & k 7 & k 5 & c 3 & b 7 \\ \frac{4}{7} & \frac{3}{7} & k 4 & k 2 & k 8 & k 6 & c 4 & b 8 \\ \frac{3}{7} & \frac{4}{7} & k 5 & k 3 & k 1 & k 7 & c 5 & b 1 \\ \frac{2}{7} & \frac{5}{7} & k 6 & k 4 & k 2 & k 8 & c 6 & b 2 \\ \frac{1}{7} & \frac{6}{7} & k 7 & k 5 & k 3 & k 1 & c 7 & b 3 \\ \frac{0}{7} & \frac{7}{7} & k 8 & k 6 & k 4 & k 2 & c 8 & b 4 \end{matrix}] \cdot [\begin{matrix} X 1 \\ X 2 \\ W 1 \\ W 2 \\ W 3 \\ W 4 \\ R \\ I \end{matrix}]$
or as a matrix equation as:
{circumflex over (P)}={circumflex over (K)}·{circumflex over (L)}
After measuring samples P1, P2, P3 . . . P8, the individual components of the composite signal can be isolated by solving the system of equations.
{circumflex over (L)}={circumflex over (K)} ⁻¹ ·{circumflex over (P)}
In practice only the Red and IR components need to be solved as the other components are usually of no interest. This can be done by only computing the results for the bottom two rows of the system. The matrix of coefficients is a constant determined by the impulse response of the system. To solve the matrix for the Red and IR components, the inverse of the matrix only needs to be computed once for a particular front end filter 150, which can be done at start-up if a variable filter design is used or during the design of the system if the system uses a constant filter. Also an adaptive filter might be used. Then, the computation has to be performed after each adaptation.
As a result, the cable transients W can be eliminated from the signal on a real time basis. Stray capacitances in the cable will no longer be an issue. This also allows a front end to be designed with a much tighter anti-aliasing filter which will reduce noise and interference.
As mentioned above, a better performance can be achieved by increasing the number of samples per Red and IR measuring period. This oversampling will result in an over determined system that can be solved by using a pseudo-inverse to the constant matrix which gives a result that is a least squares fit to the sampled data. In general more over sampling will result in a more accurate measurement.
According to a further embodiment, the same technique can be used for more than two wavelength signals. This may also result in an over determined system that can be solved with a pseudo-inverse.
The above described concept is not limited to the error signals discussed, i.e., the ambient light signal and the cable transients. Other known error sources can be included in the matrix as discussed above.
FIG. 5 shows an example of a system for solving the matrix equations. The data stream generated by the analog-to-digital converter 160 is fed to a matrix 410. Separate equations 420 and 430 for the Red signal and for the IR signal are computed to solve the matrix and generate the respective component signals for the Red and IR signals without the external error signals introduced to the signal fed to the analog-to-digital converter 160. The system shown can be easily implemented in a digital signal processor, microcontroller, or application specific integrated circuit (ASIC).
The invention, therefore, is well adapted to carry out the objects and attain the ends and advantages mentioned, as well as others inherent therein. While the invention has been depicted, described, and is defined by reference to particular preferred embodiments of the invention, such references do not imply a limitation on the invention, and no such limitation is to be inferred. The invention is capable of considerable modification, alteration, and equivalents in form and function, as will occur to those ordinarily skilled in the pertinent arts. The depicted and described preferred embodiments of the invention are exemplary only, and are not exhaustive of the scope of the invention. Consequently, the invention is intended to be limited only by the spirit and scope of the appended claims, giving full cognizance to equivalents in all respects.

Claims

1. A system, comprising:

a filter configured to receive a signal from a physiological sensor, wherein the signal comprises a first signal component and a second signal component comprising an interference component, and wherein the filter is configured to output a filtered signal;

an analog-to-digital converter configured to receive the filtered signal and sample the filtered signal in periodic intervals to generate a sampled filtered signal, wherein each periodic interval comprises n samples; and

a processor configured to:

receive the sampled filtered signal;

solve a matrix equation representing the sampled filtered signal for the first signal component, wherein the matrix equation comprises a signal vector comprising the first and second components and a matrix comprising weighted coefficients; and

determine a physiological parameter based on the first signal component and not the second signal component.

2. The system of claim 1, wherein the processor is configured to determine the interference component by a linear approximation between a first and last sample of each periodic interval.

3. The system of claim 1, wherein the filter comprises a variable filter and wherein the processor is configured to determine an inverse of the matrix of coefficients during a start-up phase of the system

4. The system of claim 1, wherein the filter comprises a constant filter and wherein the processor is configured to determine an inverse of matrix of coefficients.

5. The system of claim 1, wherein the filter comprises an adaptive filter and wherein the processor is configured to determine an inverse of the matrix of coefficients after each adaptation of the adaptive filter.

6. The system of claim 1, wherein the physiological sensor comprises a pulse oximetry sensor.

7. The system of claim 1, wherein the second signal component comprises an ambient light signal component.

8. The system of claim 7, wherein the ambient light signal component is approximated by a linear approximation between a first and last sample of each periodic interval.

9. The system of claim 1, wherein the second signal component comprises cable transient components.

10. The system of claim 9, wherein the weighted coefficients of the matrix equation for the cable transients are determined by a transfer function of the filter and a sample position within a periodic interval

11. The system of claim 1, wherein the first signal component comprises a red signal component.

12. The system of claim 1, wherein the first signal component comprises an infrared signal component and a red signal component.

13. The system of claim 1, wherein the red signal component and the IR signal component are timely separated within each periodic interval and the red signal component and the IR signal component each comprise a predetermined signal length having an on and off transient.

14. A method, comprising:

receiving a composite signal comprising a red component, an infrared component, and at least one interfering signal component;

filtering the composite signal with a filter having a transfer function H(s);

sampling the filtered composite signal with an analog to digital converter in periodic intervals wherein each periodic interval comprises n samples; and

using a processor:

solving a matrix equation representing the composite signal for the red component and the infrared component, wherein the matrix equation comprises a signal vector comprising the red component, the infrared component, and the at least one interfering signal component and a matrix comprising weighted coefficients, wherein weighted coefficients for the at least one interfering signal component are based on an impulse response and a magnitude of an impulse;

generating an output based on the red component and the infrared component.

15. The method of claim 14, wherein the red component and the infrared component are represented by an impulse response at a periodic interval multiplied by a measured current.

16. The method of claim 14, wherein the interfering signal component comprises an ambient light signal component and cable transients.

17. The method of claim 16, wherein the weighted coefficients for the cable transients are determined by the transfer function and a sample position within a periodic interval.

18. A system, comprising:

a sensor comprising one or more light emitters and a detector configured to detect light emitted by the one or more light emitter and generate a signal, wherein the signal comprises a primary signal component and at least one of an ambient light component or a cable cross-talk component;

a patient monitor comprising:

a filter configured to receive the signal and outputting a filtered signal;

an analog-to-digital converter configured to receive the filtered signal and sample the filtered composite signal in periodic intervals wherein each periodic interval comprises n samples;

a signal processor configured to:

receive the sampled filtered signal;

solve a matrix equation for the primary signal component, wherein the matrix equation comprises a signal vector comprising the primary signal component and at least one of the ambient signal component or the cable cross-talk component and a matrix comprising weighted coefficients; and

determine a physiological parameter based at least in part on the primary signal component.

19. The system of claim 18, wherein the at least one of an ambient light component or a cable cross-talk component is approximated by a linear approximation between a first and last sample of each periodic interval.

20. The system of claim 18, wherein n is determined based on an accuracy of the physiological parameter.