 |
Description  |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to impedance pneumography and more
particularly is directed toward suppressing cardiovascular artifact within
a respiration signal obtained through impedance pneumography.
2. Description of the Prior Art
A respiration signal is a measure of a patient's transthoracic impedance,
that is, the impedance across a patient's chest which varies primarily due
to the expansion and contraction of the lungs during breathing. Heart and
blood motion also cause a change in the chest size, and thus, a change in
the respiration signal. Thus, the respiration signal really comprises both
a breath component and a component due to heart and blood motion referred
to hereinafter as to cardiovascular artifact.
Therefore, in determining a condition of apnea in the patient, that is,
whether the patient has ceased breathing it is highly desirable to
identify and suppress those components in the respiration signal which are
due to heart and blood motion so that the time duration between breaths
can be measured. Otherwise, cardiovascular artifacts can be mistakenly
interpreted as breath events when, in fact, a condition of apnea exists.
On the other hand, if the breath component of the respiration signal is
suppressed in order to remove cardiovascular artifacts, the filtered
respiration signal may be incorrectly interpreted as representing a
condition of apnea.
One general solution for suppressing cardiovascular artifacts from the
respiration signal is based on the fact that cardiovascular artifacts
normally have frequencies near or above the heart rate. Accordingly, as
long as the heart rate is greater than the breath rate, the cardiovascular
artifact within the respiration signal can be greatly reduced, that is,
suppressed by removing those components of the respiration signal having
frequencies at or above the heart rate. The resulting filtered respiration
signal will contain basically only the breath component. The removal, that
is, the filtering of such selected frequencies based on another time
variant parameter such as heart rate is commonly referred to as adaptive
filtering.
Prior art adaptive filters for suppressing cardiovascular artifacts from a
respiration signal, commonly referred to as cardiovascular artifact (CVA)
filters, typically implement a scheme in which a signal sample is added to
previous samples which have been multiplied by one of a number of
different coefficients. The choice of coefficients which vary in value
based on the heart rate determines the filter's characteristics.
In today's computer era, CVA filtering schemes are typically implemented by
employing one or more microprocessors. These microprocessors besides
processing the CVA filtering scheme are used for undertaking a number of
other tasks which are unrelated to the filtering scheme. In light of these
other tasks, the processing time required to execute the above adaptive
filtering scheme has taken on added importance. In this regard, the above
CVA filtering scheme is considered less than optimal since the frequent
change of coefficients requires a relatively large amount of execution
time. Additionally such an adaptive filtering scheme requires an
undesirable amount of hardware, that is, memory for storing these
coefficients. The additional execution time and memory required, of
course, result in a more expensive microprocessor system.
OBJECTS AND SUMMARY OF THE INVENTION
Accordingly, it is an object of the present invention to provide a filter
for suppressing cardiovascular artifacts from a respiration signal which
avoids the drawbacks of the prior art.
More specifically, it is an object of the present invention to provide a
new and improved filter for suppressing cardiovascular artifacts from a
respiration signal employing a filtering scheme which requires less
processing time.
It is another object of the present invention to provide a filter for
suppressing cardiovascular artifacts from a respiration signal which
requires less microprocessor hardware.
In accordance with an aspect of this invention, a filtering device having a
cutoff frequency for suppressing cardiovascular artifacts derived from a
patient comprises heart beat detecting means for detecting the heart beats
of the patient; processing means for determining the heart rate of the
patient in response to the detected heart beats; and filtering means for
attenuating a portion of the frequency spectrum of the respiration signal
based on the rate at which the filtering means samples the respiration
signal wherein the sampling rate is proportional to the heart rate and
whereby the cutoff frequency varies in proportion to the sampling rate.
It is a feature of the present invention that the hear beat detecting means
comprises a QRS detecting means for detecting the QRS complex of each
heart beat waveform produced by the patient. It is another feature of the
present invention to provide conversion means for converting from an
analog to a digital representation of the respiration signal at a
converting rate proportional to the heart rate and wherein the filtering
means samples the digital representation of the respiration signal. Other
features of the present invention provide two timers one of which clocks
the elapsed time between hear beats of the patient and the other of which
controls the rate at which the conversion means converts from the analog
to a digital representation of the respiration signal. It is still another
feature of the present invention to provide a successive approximating
analog-to-digital converter as the conversion means.
In accordance with another aspect of the present invention, a method for
suppressing cardiovascular artifact from a respiration signal derived from
a patient comprises detecting the heart beats of the patient, determining
the heart rate of the patient in response to the detected heart beats, and
attenuating a portion of the frequency spectrum of the respiration signal
wherein each component of the respiration signal within this portion is
attenuated in accordance with the ratio of the component's frequency
relative to the heart rate.
In regard to this latter aspect of the present invention, features of the
invention provide that the method further comprises converting from analog
to digital values of the respiration signal at a conversion rate
proportional to the heart rate. Another feature of this latter aspect of
the present invention provides that the rate at which attenuated
respiration signal values are computed is proportional to the converting
rate.
The above and other objects, features, and advantages of this invention
will become apparent from the following detailed description which is to
be read in conjunction with the accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a schematic drawing, partially in block form, illustrating a low
pass filter;
FIG. 2 is a block diagram of the present invention;
FIG. 3 is a Bode plot showing the frequency response of a preferred
embodiment of the present invention;
FIG. 4 is a detailed block diagram, partially in schematic form, of a
filter in accordance with a preferred embodiment of the present invention;
and
FIG. 5 illustrates flow charts of a heart beat routine and of a respiration
routine in accordance with the embodiment shown in FIG. 3.
DETAILED DESCRIPTION OF A PREFERRED EMBODIMENT
Shown in FIG. 1 is a single stage, two pole filter 10 for use in
suppressing cardiovascular artifacts within a respiration signal. The
respiration signal input is derived from a patient based on the patient's
transthoracic impedance. Filter 10 comprises a summer 11 connected to both
a summer 12 and a storage device 13a commonly referred to as a Z.sup.-N.
Two amplifiers 14a and 14b, which multiply their respective input signals
by factors K.sub.1 and K.sub.2, are connected at their respective input
terminals to output terminals of storage device 13a. The output terminals
of amplifiers 14a and 14b are connected to input terminals of summers 11
and 12, respectively. An additional storage device (Z.sup.-N) 13b is also
connected to an output terminal of storage device 13a. Two additional
amplifiers 14c and 14d, which multiply their respective input signals by
factors K.sub.3 and K.sub.4, are connected at their input terminals to
output terminals of storage device 13b. The output terminals of amplifiers
14c and 14d are connected to two additional input terminals of summers 11
and 12, respectively. Operation of filter 10 can be most simply described
according to the following equation:
Signal Out (t)=S.sub.(t) +(K.sub.1 +K.sub.2)Z.sup.-1.sub.(t-1) +(K.sub.3
+K.sub.4)Z.sup.-2.sub.(t-1) (eq. 1)
where:
S.sub.(t) is the respiration sample at time t
K.sub.1 is the multiplying factor of amplifier 14a
K.sub.2 is the multiplying factor of amplifier 14b
K.sub.3 is the multiplying factor of amplifier 14c
K.sub.4 is the multiplying factor of amplifier 14d
Z.sup.-1.sub.(t) =S.sub.(t) +K.sub.1 Z.sup.-1.sub.(t-1) +K.sub.3
Z.sup.-2.sub.(t-1)
Z.sup.-2 (t)=Z.sup.-1.sub.(t-1)
Filter 10 has a filtering characteristic dependent upon the choice of
values selected for factors K.sub.1, K.sub.2, K.sub.3 and K.sub.4 and the
rate at which the respiration samples are clocked through filter 10. By
dynamically varying the values of K.sub.1, K.sub.2, K.sub.3 and K.sub.4
based on the heart rate, filter 10 becomes an adaptive filter typically
used in the prior art for suppressing cardiovascular artifacts and thereby
for providing a respiration signal containing only the breath component.
The respiration signal is being sampled by filter 10 at a constant rate.
The above adaptive filtering scheme, however, has two important drawbacks.
More specifically, filter 10 if implemented, in part, by employing a
microprocessor requires less than an optimal amount of execution time due
to the frequent changes of coefficients (K.sub.1 -K.sub.4) required.
Additionally, such an adaptive filter requires far too much microprocessor
hardware due to the memory necessary to store all of these coefficients.
The present invention avoids such prior art drawbacks by employing an
adaptive filtering scheme which is based on the well known premise that
the artifact present in the respiration signal is comprised of frequencies
equal to or at multiples of the heart rate. More specifically, however,
rather than varying coefficients K.sub.1 -K.sub.4 while maintaining a
constant rate at which the rspiration signal is being sampled as in the
prior art, the present invention advantageously varies the rate at which
the respiration signal is being sampled while maintaining coefficients
K.sub.1 -K.sub.4 constant. In other words, the filtering characteristics
of the present invention are varied by changing the rate at which data is
clocked therethrough.
Referring now to the present invention as shown in FIG. 2, an
electrocardiogram (ECG) signal obtained from a patient is fed into an ECG
amplifier 15 and the respiration signal, which includes cardiovascular
artifact, is supplied to respiration amplifier 16 by a transthoracic
sensing device 17. Amplifier 15 has a gain of 1000 millivolts (mv) per 1
mv of ECG signal and a bandwidth of 1 to 100 hertz (Hz). Amplifier 16 has
a gain of 0.4 volts per ohm of transthoracic impedance variation and a
bandwidth of 0.2 to 2.5 Hz. Transthoracic impedance sensing devices, such
as device 17, are well known in the art of respiration monitoring and are
essentially merely a.c. ohmmeters. Filter 20 shown within dashed lines,
comprises a heart beat detector 21, a microcontroller 22 and an
analog-to-digital converter 23. Heart beat detectors such as detector 21
normally detect the QRS complex of the waveform of the patient and are
well known in the field and need not be described further. Detector 21 is
connected to the microcontroller 22 and supplies to the latter a signal
each time a heart beat is detected. Microcontroller 22, which comprises a
microprocessor, an erasable-programmable-read-only-memory (EPROM), a
latch, an input/output (I/O) buffer and two timers, processes, that is,
determines the heart rate of the patient based on these detected heart
beats and supplies control signals to converter 23 so as to control the
rate at which the respiration signal is converted from an
analog-to-digital representation by converter 23. The respiration data
supplied by converter 23 is then filtered by microcontroller 22 to remove
the CVA and then processed to determine the time of occurrence of each
breath. Based on the breath events, microcontroller 22 triggers an alarm
indicating the lack of respiration commonly referred to as apnea and a
breath event indicator.
The conversion rate of converter 23, that is, the rate at which the analog
signal is being sampled by converter 23 is directly proportional to the
heart rate. In the preferred embodiment, the converting rate has been set
at ten times the heart rate. As is well known in the art based on the
Nyquist (Sampling) theorem, a sampling rate must be at least twice the
highest frequency component in order to completely characterize a signal.
The minimum value of the sample rate is 5 Hz which is ten times the
minimum computed heart rate of 0.5 Hz (30 beats per minute). As previously
noted, the respiration signal amplifier 16 limits the respiration signal
to those components less than 2.5 Hz. Thus the 5 Hz minimum sample rate is
twice the highest respiratory component of 2.5 Hz.
The cutoff frequency of filter 20 is directly proportional to the
conversion rate and thus proportional to the heart rate. Consequently, by
employing the adaptive filtering scheme of FIG. 2, portions of the
respiration signal having a frequency content at or above the heart rate
are significantly attenuated while those components of the respiration
signal having a frequency content at half or less the heart rate are
substantially unattenuated. In other words, the adaptive filtering scheme
of the present invention provides a low pass filter. Guidelines for
attentuation of the respiration signal will be discussed in greater detail
below.
A Bode plot of the CVA filter of the present invention is shown in FIG. 3
wherein the abscissa contains all the frequency components of the
respiration signal expressed as a percentage of the patient's heart beat
frequency and wherein the ordinate represents the amplitude of each
frequency component of the filter output expressed in decibels (db).
The present invention operates as a low pass filter which preferably has a
cutoff frequency, that is, a -3 db point, equal to approximately 78.4% of
the fundamental heart beat frequency. Accordingly, the present invention
provides a filtering device for attenuating a portion of the frequency
spectrum of the respiration signal such that the cutoff frequency varies
in proportion to the heart rate. The filter provides attenuations of
approximately 17 db, 10 db, and 1.5 db for those components of the sampled
respiration signal's frequency spectrum at 100%, 90% and 75% of the heart
rate, respectively. Additionally, the filter provides approximately zero
attenuation for those components of the frequency spectrum of the sampled
respiration signal at 50% or less of the heart rate.
As now can be readily appreciated, since the respiration rate is normally a
small percentage of the heart rate, the present invention substantially
suppresses the cardiovascular component relative to the breath component
of the respiration. For example, assume for illustrative purposes only
that the respiration signal comprises a breath component at a single
frequency of 0.25 Hz (15 breaths per minute) and a cardiovascular
component at a single frequency of 1 Hz (60 beats per minute). The breath
rate represents 25% of the heart beat frequency and the cardiovascular
rate is shown only at the fundamental heart beat frequency. Therefore, the
breath component will not be attenuated at all whereas the CVA component
will be attenuated by approximately 17 db by the present invention.
Referring now to FIG. 4, a preferred embodiment of the present invention is
shown in which elements similar to those discussed in connection with FIG.
2 are identified by the same reference numerals. An adaptive filter 20a
comprises heart detecting means 21 connected to a microprocessor 25. A 6
MHz system clock for microprocessor 25 is connected to terminals X.sub.1
and X.sub.2 of the microprocessor. The 6 MHz system clock is divided by
microprocessor 25 for providing various clock signals including a clock
signal CLOCK of 400 KHz (a 2.5 microsecond period) which is supplied to
chip 31 on line 26. Chip 31 comprises a random access memory (RAM), an
input/output (I/O) buffer and two timers. Microprocessor 25 is an 80C39, 8
bit microprocessor and may be obtained from a number of manufacturers such
as the Intel Corporation. Chip 31 is an NSC810 manufactured by the
National Semiconductor Corporation.
An 8 bit low order address/data bus 35a is connected between microprocessor
25 and integrated package 31 and is also connected to a latch 36, a
digital-to-analog converter (DAC) 37 and an
erasable-programmable-read-only-memory (EPROM) 40. A separate low order
address bus 35b is connected between latch 36, a decoder 45 and EPROM 40.
A high order 8 bit address bus 41 is connected between EPROM 40 and
microprocessor 25. Latch 36 is an 8 bit latch commonly identified within
the industry as a 74HC373 which is used to store the addresses of data
processed by microprocessor 25. Decoder 45 has eight output terminals
three of which are shown as WCS0, WCS1 and RCS0. Terminals WCSO and RCSO
are connected to the write and read terminals of chip 31, respectively.
Terminal WCS1 is connected to the write terminal (WR) of DAC 37. Decoder
45 is used for selecting among microprocessor 25, latch 36, EPROM 40 and
chip 31 by sending control signals on a control bus 50.
Chip 31 and microprocessor 25 are also connected together by line QRS
Sample F and by line RESP Sample F. As will be discussed below, these
lines are used for determining the elapsed time between heart beats and
for triggering/beginning the analog-to-digital conversion of the
respiration signal, respectively. Connected to the output of DAC 37 is a
current-to-voltage converter 51 which converts the current signal supplied
by DAC 37 into an analog voltage. The output of converter 51 is connected
to the inverting input of a comparator 60. Supplied to the non-inverting
input of comparator 60 is the analog representation of the sampled
respiration signal provided by respiration amplifier 16 and identified as
RESP.sub.A. The output of comparator 60 is supplied to microprocessor 25.
Device 20a operates as follows: Initially microprocessor 25 loads a
starting value into timer 32 of integrated package 31 according to
instructions stored in EPROM 40, which contains all the programs for
operating device 20a. Timer 32, which is a down counter, decrements from
this starting value every 2.5 microseconds based on the CLOCK signal
received from microprocessor 25 on clock line 26. Once counter 32
decrements to a value of zero, a pulse is sent along line QRS Sample F to
microprocessor 25. Thus each pulse sent along line QRS Sample F represents
a predetermined elapsed period of time. Microprocessor 25 keeps a running
tab of the number of pulses received along line QRS Sample F between heart
beats determined by heart beat detector 21. Based on the number of pulses
received along QRS Sample F line between heart beats, microprocessor 25 is
able to determine the heart beat rate. For example, if each pulse supplied
by QRS Sample F line is equal to an elapsed time of 25 milliseconds, then
if 40 pulses are received by microprocessor 25 between two heart beats,
the heart rate would be 60 beats per minute. Once timer 32 reaches a value
of zero it will automatically reload to the starting value set by EPROM 40
and continues to recycle through this counting scheme thereafter. Based on
the computed heart rate, microprocessor 25 will be directed by EPROM 40 to
load one of over one hundred fifty different starting values in a second
timer 33 of chip 31. These starting values are stored in a table within
EPROM 40 and correspond to the computed heart rate. Timer 33 is also
decremented based on a clock signal derived from CLOCK signal, that is,
timer 33 is decremented every 10 microseconds. Once timer 33 reaches a
value of zero, a pulse is sent along RESP Sample F line and received by
microprocessor 25 which triggers, that is, begins the start of conversion
of the analog-to-digital representation of the respiration signal. Higher
or lower initial values are used for resetting timer 33 when a slower or
faster conversion rate is desired, respectively. As can now be readily
appreciated, the rate at which the respiration signal will be sampled and
converted from an analog to digital representation is determined based on
the heart rate.
In converting the respiration signal from an analog-to-digital form, a
successive approximation technique is employed which utilizes
microprocessor 25, DAC 37, current-to-voltage converter 51, and comparator
60. This technique is well known in the art and will therefore only be
briefly described herein. More specifically, microprocessor 25 will supply
a first approximating ten bit signal having a value midway between the
highest and lowest signals which can be produced by respiration amplifier
16. The first approximating signal produced by microprocessor 25 is
nothing more than an intelligent guess as to the actual value of the
respiration signal (RESP.sub.A) produced by respiration amplifier 16. DAC
37 will convert this first approximating signal to an equivalent analog
current value which is then converted to an equivalent voltage by
current-to-voltage converter 51. The analog output from converter 51 is
supplied to the inverting input of comparator 60 and compared to the
sampled respiration signal (RESP.sub.A) supplied by respiration amplifier
16 to the noninverting input of comparator 60. The output signal from
comparator 60 is supplied to and used by microprocessor 25 for determining
the next, that is, successive approximating signal produced by
microprocessor 25. Through this repetitive process, which is repeated ten
times, microprocessor 25 "zeros-in" on the digital value of RESP.sub.A.
The foregoing successive approximation technique can be easily defined in
terms of an algorithm by one of ordinary skill in the art and more
particularly is defined by a computer program which is used by filtering
device 20a and which is found at the end of this detailed description of a
preferred embodiment. The computer program also discloses all of the
foregoing steps involved in determining the heart rate and the rate at
which the respiration signal is sampled for purposes of conversion from an
analog to digital representation. As now can be readily appreciated, the
ADC 23 of FIG. 2 is comparable to microprocessor 25, DAC 37, converter 51
and comparator 60 of FIG. 4.
Once the digital representation of the sample representation signal has
been determined EPROM 40 is notified of the same through latch 36 by
requesting the next instruction from EPROM 40 as to what to do with this
digital value. Consequently, EPROM 40 will instruct microprocessor 25 to
filter this digital sample respiration signal according to the filtering
scheme of FIG. 1 and as described by eq. 1 above wherein the values of
K.sub.1, K.sub.2, K.sub.3 and K.sub.4 are not dynamically varied based on
the heart rate as in the prior art but rather are maintained at fixed
values while the rate at which the sampled signal is filtered is varied
based on the heart rate. Eq. 1 can be readily implemented by one of
ordinary skill in the art of computer software using a number of different
computer programs. One such computer listing is disclosed below in the
computer program.
Microprocessor 25 then uses the filtered sample of the respiration signal,
which comprises the unattentuated breath component and substantially
suppressed cardiovascular artifacts, in a breath detection algorithm to
determine when each breath occurs. This algorithm is also disclosed as
part of the computer program.
Light emitting diodes 70 and 71, which are connected to chip 31, are used
to alert a user each time a breath and a condition of apnea occurs,
respectively. More particularly, the computer program provides a series of
instructions for directing diode 70 to light each time a breath occurs and
for diode 71 to light when a condition of apnea occurs based on the breath
detection algorithm.
Referring now to FIG. 5 the heart beat routine and respiration routine
described above and implemented by the computer program are shown in the
form of flow charts. The first step (101) in the heart beat routine
comprises waiting for the heart beat which is detected by heart beat
dectector 21. The beat-to-beat interval between heart beats (step 102) is
counted by microprocessor 25. In this regard, the number of pulses
supplied along QRS Sample F line between the detected heart beats are
counted by microprocessor 25 to determine the elapsed time between the
heat beats. In step 103, the heart rate based on the elapsed time between
heart beats is computed. Finally, step 104 computes the rate, at which the
respiration signal should be sampled based as the heart rate. In filtering
device 20a this computation is performed by picking a starting value for
timer 33 based on the computed heart rate. Once step 104 has been
completed, the heart beat routine returns to step 101.
The respiration routine begins with step 110 which samples the respiration
signal. In filtering device 20a step 110 is performed by converting the
respiration signal from an analog to digital representation. The next step
111 of the respiration routine is to filter the digital respiration signal
using either a single or multi stage filter. The number of stages may be
varied in order to achieve the filtering characteristics desired. In the
present invention, the filtering scheme comprises four cascaded single
stage, two pole filters one of which is shown in FIG. 1 and is implemented
according to the computer program. The filtered respiration signal is then
used by microprocessor 24 to detect each breath event as identified by
step 112. A number of different algorithms including, but not limited to,
the algorithm in the computer program can be used for detecting each
breath. The final step 113 of the respiration routine counts the time
until the next respiration signal is sampled. In the present invention,
this counting is done by timer 33. Normally, microprocessor 25 performs
the respiration routine at a rate proportional to the last computed heart
rate. Periodically, the respiration routine is interrupted and the heart
beat routine is performed after which the respiration routine resumes.
As can now be readily appreciated, the present invention provides a CVA
filter which effectively suppresses the cardiovascular artifacts within
the respiration signal and thereby provides a respiration signal which
includes basically only the breath component. Furthermore, as compared to
prior art CVA filters, the present invention does not require as much
execution time and therefore need not be as sophisticated or as costly as
microprocessors presently used for CVA filtering purposes.
Having specifically described an illustrative embodiment of the invention
with reference to the accompanying drawings, it is to be understood that
the invention is not limited to this precise embodiment and that various
changes and modifications may be effected by one skilled in the art
without departing from the scope and spirit of the invention as defined by
the appended claims. For example, the present invention need not be
limited to employing analog to digital converters and may instead directly
filter the analog representation of the respiration signal at a sampling
rate proportional to the heart rate.
##SPC1##
* * * * *
|
|
 |
|
 |
Description |
|
