U.S. patent application number 10/508833 was filed with the patent office on 2006-03-23 for non-invasive blood component value measuring instrument and method.
This patent application is currently assigned to Ken-ichi YAMAKOSHI. Invention is credited to Ken-ichi Yamakoshi.
Application Number | 20060063983 10/508833 |
Document ID | / |
Family ID | 28449186 |
Filed Date | 2006-03-23 |
United States Patent
Application |
20060063983 |
Kind Code |
A1 |
Yamakoshi; Ken-ichi |
March 23, 2006 |
Non-invasive blood component value measuring instrument and
method
Abstract
A non-invasive blood constituents measuring instrument measures
blood constituent values including blood glucose concentration in a
living body. The instrument is composed of a light source 11 to
irradiate a light containing plural wavelengths to a living body
13, a light detector 14 to detect the light transmitted through a
living body or reflected thereon, an instantaneous spectrum
analyzer 15 to analyze the light transmitted through the living
body or reflected thereon at different times, a subtraction
processor 18 to generate spectrum subtraction from the spectrum of
the light at the different times measured by the spectrum analyzer
15, and a blood glucose concentration predictor 21 into which the
output data of the subtraction processor 18 are input and which
outputs a blood constituent value.
Inventors: |
Yamakoshi; Ken-ichi;
(Kanazawa-shi, JP) |
Correspondence
Address: |
BANNER & WITCOFF
1001 G STREET N W
SUITE 1100
WASHINGTON
DC
20001
US
|
Assignee: |
Ken-ichi YAMAKOSHI
5-57, Wakunami 3-chome
Kanazawa-shi, Ishikawa
JP
920-0953
TYT Institute of Technology Corp
1-143-602, Morinosato
Kanazawa-shi, Ishikawa
JP
920-1167
|
Family ID: |
28449186 |
Appl. No.: |
10/508833 |
Filed: |
March 25, 2003 |
PCT Filed: |
March 25, 2003 |
PCT NO: |
PCT/JP03/03587 |
371 Date: |
April 14, 2005 |
Current U.S.
Class: |
600/310 ;
600/316 |
Current CPC
Class: |
A61B 5/14546 20130101;
A61B 5/1455 20130101; G01N 2201/129 20130101; G01N 2201/1228
20130101; G01N 2201/1232 20130101; G01N 21/359 20130101; A61B
5/14532 20130101; G01N 33/491 20130101 |
Class at
Publication: |
600/310 ;
600/316 |
International
Class: |
A61B 5/00 20060101
A61B005/00 |
Foreign Application Data
Date |
Code |
Application Number |
Mar 25, 2002 |
JP |
2002-083587 |
Claims
1. A non-invasive blood constituents measuring instrument
comprising: a light source for irradiating light including plural
waveforms to a living body; a light detector to detect light
transmitted through the living body or reflecting thereon; a
spectrum analyzer to which the output signal of the light detector
is supplied and which analyzes spectrum of the light transmitted
through the living body or reflected therefrom at different times;
a spectrum subtraction generator to generate spectrum subtraction
from the spectrum of the light measured by the spectrum analyzer at
different times; and a blood constituents predictor into which the
output data of the spectrum subtraction generator is input and
which outputs the blood constituents.
2. A non-invasive blood constituents measuring instrument claimed
in claim 1, wherein the blood constituents predictor is provided
with a multi-regression analyzing model using plural whole blood
sample spectrum data having known blood constituent as an
explanatory variable and blood constituent values as an objective
variable, wherein the spectrum subtraction data obtained from
bloods having known blood constituents are input into the
multi-regression analyzing model as an explanatory variable, the
objective variable is computed from the multi-regression analyzing
model and output as a blood constituent value.
3. A non-invasive blood constituents measuring instrument claimed
in claim 2, wherein the multi-regression analyzing model is a
regression analysis model using the PLS or PCR method.
4. A non-invasive blood constituent measuring instrument claimed in
claim 3, wherein blood constituents of the plural whole blood
samples are arranged at a specified interval within the range of
concentration including the physiological concentration range.
5. A non-invasive blood constituent measuring instrument claimed in
claim 4, wherein the light including the plural wavelengths is the
light in the near infrared range.
6. A non-invasive blood constituent measuring instrument claimed in
claim 5, wherein the light includes plural wavelengths in the
wavelength band of 800.about.2400 nm arranged at an interval of
about 3 nm.
7. A non-invasive blood constituent measuring instrument claimed in
claim 6, wherein the light including the plural wavelengths is
generated by separating the light from the light source by the
active spectroscope.
8. A non-invasive blood constituent measuring instrument claimed in
claim 7, wherein the active spectroscope separates the light in the
near infrared range at an interval of about 50 ms period of
time.
9. A non-invasive blood glucose concentration measuring instrument
comprising: a light source to irradiate a light containing plural
wavelengths; a light detector to detect the light transmitted
through a living body or reflected therefrom; a spectrum analyzer
to which the output signal of the light detector is supplied and
which analyzes spectrum of the light transmitted through the living
body or reflected therefrom at different times; a spectrum
subtraction generator to generate spectrum subtraction from the
spectrum of the light measured by the spectrum analyzer at
different times; and a blood glucose concentration predictor into
which the output data of the spectrum subtraction generator is
input and which outputs the blood glucose concentration.
10. A non-invasive blood glucose concentration measuring instrument
claimed in claim 9, wherein the multi-regression analyzing model is
a regression analysis model using the PLS or PCR method.
11. A non-invasive blood glucose concentration measuring instrument
claimed in claim 10, wherein blood constituents of the plural whole
blood samples are arranged at a specified interval within the range
of concentration including the physiological concentration
range.
12. A non-invasive blood glucose concentration measuring instrument
claimed in claim 11, wherein the light including the plural
wavelengths is the light in the near infrared range.
13. A non-invasive blood glucose concentration measuring instrument
claimed in claim 12, wherein the light includes plural wavelengths
in the wavelength band of 800.about.2400 nm arranged at an interval
of about 3 nm.
14. A non-invasive blood glucose concentration measuring instrument
claimed in claim 13, wherein the light including the plural
wavelengths is generated by separating the light from the light
source by the active spectroscope.
15. A non-invasive blood glucose concentration measuring instrument
as set force in claim 14, wherein the active spectroscope separates
the light in the near infrared range at an interval of about 50 ms
period of time.
17. A non-invasive blood glucose concentration measuring instrument
claimed in claim 12, wherein the plural whole blood samples include
protein containing albumin.
18. A non-invasive blood glucose concentration measuring instrument
claimed in claim 17, wherein the concentration of the albumin is
about 3.0.about.6.0 g/dl.
19. A non-invasive blood glucose concentration measuring instrument
claimed in claim 18, wherein the plural whole blood samples contain
blood having different hematocrit values.
20. A method for non-invasively measuring blood constituents
comprising the steps of: irradiating a light containing plural
wavelengths to a living body; detecting light transmitted through
or reflected from the living body and converting it into an
electric signal; analyzing spectrum of the light transmitted
through the living body or reflected therefrom at different times
using the converted electric signal; generating spectrum
subtraction from the spectrum of the light at different times; and
predicting corresponding blood constituents from the spectrum
subtraction.
21. A method for non-invasively measuring blood constituents
claimed in claim 20, wherein the step of predicting the blood
constituents further comprises; preparing a multi-regression
analyzing model, to which spectrum data of plural whole blood
samples of which blood constituents are known are input as
explanatory variables and outputs blood constituents as objective
variables, inputting the spectrum subtraction data obtained from
blood of which blood constituents is not known as explanatory
variables, and outputting the blood constituents as the objective
variables.
22. A method for non-invasively measuring blood constituents
claimed in claim 21, wherein the multi-regression analyzing model
is constructed using the PLS or PCR method.
23. A method for non-invasively measuring blood glucose
concentrations comprising the steps of: irradiating a light
containing plural wavelengths to a living body; detecting light
transmitted through or reflected from the living body and
converting it into an electric signal; analyzing spectrum of the
light transmitted through the living body or reflected therefrom at
different times using the converted electric signal; generating
spectrum subtraction from the spectrum of the light at the
different times; and predicting corresponding blood glucose
concentration from the spectrum subtraction.
24. A method for non-invasively measuring blood glucose
concentrations claimed in claim 23, wherein the step of predicting
the blood glucose concentration further comprises; preparing a
multi-regression analyzing model, to which spectrum data of plural
whole blood samples of which blood glucose concentrations are known
are input as explanatory variables and outputs blood glucose
concentrations as objective variables, inputting the spectrum
subtraction data obtained from blood of which blood glucose
concentration is not known as explanatory variables, and outputting
the blood glucose concentration as the objective variables.
25. A method for non-invasively measuring blood glucose
concentrations claimed in claim 24, wherein the multi-regression
analyzing model is constructed using the PLS or PCR method.
Description
FIELD OF THE INVENTION
[0001] The present invention relates to an instrument and a method
for measuring blood biochemical constituent including blood glucose
concentration and, more particularly, to a non-invasive blood
constituent measuring instrument and a method for measuring blood
glucose concentration without sampling blood from a living
body.
[0002] In the modern era of the expanding aged society as well as
the change in life-style, there has been a considerable increase in
the number of diabetes which is growing social concern as a
representative of lifestyle-related diseases. It is so far a
general practice to measure blood glucose concentration by sampling
a small amount of blood. However, it is strongly desired to reduce
pain and botheration associated with the blood sampling. In
addition, there is no other method available than the blood
examination for measuring blood biochemical constituent.
[0003] On the other hand, a non-invasive measurement using near
infrared light is collecting attentions for extremely low risk to
living bodies and the possibility for measurement of items so far
impossible by existing measuring methods. For example, glucose has
an inherent absorption band derived from its constituents in this
wavelength band and various methods are reported (Reference
Literature: Ozaki Yukihiro, Practical Spectroscopy Series No. 4
"Medical Application of Spectroscopy", IPC (Industrial Publishing
Consulting, Inc.)
[0004] For example, according to the reference literature, a method
to obtain blood glucose concentration by irradiating an infrared
light to a fingertip and through the computation of its transmitted
light by a computer is proposed. For this method, however, it is
very difficult to estimate glucose concentration in blood from the
transmitted light obtained and thus a method to estimate glucose
concentration using a multi-regression analysis is also
proposed.
[0005] However, the absorption band inherent to glucose in the near
infrared range overlaps on other constituent absorption ranges of
protein materials, etc. and it is difficult to separate an
absorption characteristic coming from glucose only and absorption
characteristic of other material and therefore, there is a question
in measuring accuracy and reproducibility of measurement and the
proposed method is not yet put in practical use.
[0006] Further, a glucose measuring method using the
above-mentioned multi-regression analysis is reported in the
above-mentioned reference literature as shown below. That is, this
method is to measure glucose in blood serum using the PLS method
(partial least squares analysis) that is one of chemometrics by
measuring infrared spectrum with lights in two wavelength ranges of
1325.about.1800 nm and 2035.about.2375 nm applied to glucose sample
melted in blood serum.
[0007] However, as reported that a near infrared spectroscope made
by NIR System Corp. according to the transmission penetration
method using a quartz photocell in 0.5 mm light path length in the
measurement, a quartz photocell was used in the measurement and is
not a non-invasive measurement by irradiating light to living
bodies.
[0008] In a non-invasive blood glucose concentration measuring
method using a conventional absorption analysis method, the glucose
absorption band overlaps the absorption ranges of other biological
tissues in living bodies such as bones, veins, muscles and it is
difficult to separate the ranges and the accurate measurement is
not feasible and is therefore not put in practical use.
[0009] Accordingly, an objective of the present invention is to
provide a non-invasive blood glucose measuring instrument and a
measuring method by solving the above-mentioned problems to allow
the blood glucose concentration measurement with a simple way as
well as with high accuracy.
SUMMARY OF THE INVENTION
[0010] A non-invasive blood constituent measuring instrument
according to an embodiment of the present invention includes a
light source to irradiate a light having plural wavelengths to a
living body; a light detector to detect the light transmitted
through a living body or reflected therefrom; an instantaneous
spectrum analyzer to analyze spectrum of light transmitted through
or reflected on the living body at different times when the output
signal of the light receiver is supplied; a spectrum subtraction
generator to generate spectrum subtraction from light spectrum at
the different times measured by the spectrum analyzer; and a blood
constituent predictor into which output data of the spectrum
subtraction is input and blood constituent is output.
[0011] Further, in the non-invasive blood constituent measuring
instrument according to the embodiment of the present invention, a
blood constituent predictor is provided with a multi-regression
analyzing model using plural spectrum data of whole blood
constituent of which is known as an explanatory variable and using
the blood constituent as an objective variable, wherein being input
the spectrum subtraction data obtained from the blood of which
blood constituent is known as the explanatory variable, the
multi-regression analyzing model computes the object variable and
outputs this objective variable as a blood constituent.
[0012] Further, the non-invasive blood glucose concentration
measuring instrument according to the embodiment of the present
invention is composed of a light source to irradiate a light
containing plural wavelengths; a light detector to detect the light
transmitted through a living body or reflected therefrom; an
instantaneous spectrum analyzer to which the output signal of the
light receiver is supplied and which analyzes spectrum of the light
transmitted through the living body or reflected therefrom at
different times; a spectrum subtraction generator to generate
spectrum subtraction from the spectrum of the light measured by the
spectrum analyzer at the different times; and a blood glucose
concentration predictor into which the output data of the spectrum
subtraction generator is input and which outputs the blood glucose
concentration.
[0013] Further, in the non-invasive blood glucose concentration
measuring instrument according to the embodiment of the present
invention, the blood glucose concentration predictor is constructed
with a multi-regression analyzing model into which spectrum
subtraction data of plural whole blood samples of known blood
constituent is input as the explanatory variable and in which the
blood glucose concentration is computed as an objective variable
and output as blood glucose concentration.
[0014] A non-invasive blood constituent measuring method according
to an embodiment of the present invention includes the steps of
irradiating a light containing plural wavelengths to a living body;
detecting light transmitted through or reflected from the living
body and converting it into an electric signal; analyzing spectrum
of the light transmitted through the living body or reflected
therefrom at different times using the converted electric signal;
generating spectrum subtraction from the spectrum of the light at
the different times; and predicting corresponding blood
constituents from the spectrum subtraction.
[0015] Further, in the steps of the non-invasive blood constituent
measuring method according to the embodiment of the present
invention, the blood constituent predicting step further includes
the steps of preparing a multi-regression analyzing model, into
which spectrum data of plural whole blood samples having known
blood constituent is input as an explanatory variable and blood
constituent is output as an objective variable, inputting the
spectrum subtraction data obtained from blood of which blood
constituent is not known as an explanatory variable, and outputting
the blood constituent as an objective variable.
BRIEF DESCRIPTION OF THE DRAWINGS
[0016] FIG. 1 is a block diagram according to the embodiment of the
present invention;
[0017] FIG. 2 is a flowchart showing a construction method of an
analytical prediction model used in the blood concentration
prediction instrument shown in FIG. 1;
[0018] FIG. 3 is a diagram showing an arterial pulsatile volume
waveform in a living body;
[0019] FIG. 4 is a waveform diagram showing examples of spectrums
output from an instantaneous spectrum analyzer in FIG. 1;
[0020] FIG. 5 is a diagram for explaining the operation of a blood
glucose prediction instrument shown in FIG. 1;
[0021] FIG. 6 is a diagram showing properties of the light passed
through a living body for explaining the principle of the present
invention; and
[0022] FIG. 7 is a diagram showing another embodiment of the
non-invasive blood glucose concentration measuring instrument
according to the present invention.
DETAILED DESCRIPTION OF THE INVENTION
[0023] An embodiment of the present invention will be described
below in detail referring to the attached drawings. In the
embodiment shown below, the concentration measurement of blood
glucose as one of blood constituents will be explained. However,
the present invention is also applicable to the concentration
measurement of such other materials as glycol-albumin, hemoglobin
A.sub.1C(HbA.sub.1C), cholesterol and so on, which are blood
constituents other than blood glucose existing in the arterial
blood having light absorption characteristics and scattering as
well as reflecting characteristics.
[0024] FIG. 1 is a block diagram showing a non-invasive blood
glucose concentration measuring instrument of the present
invention.
[0025] As shown in FIG. 1, the light source 11 to emit a light
having a near infrared wavelength range of, for example,
800.about.2400 nm wavelength has been installed in a non-invasive
blood glucose concentration measuring instrument. The light emitted
from the light source 11 is irradiated to a living body 13 such as
a fingertip, an ear lobule, etc. through an active spectroscope 12.
The active spectroscope 12 separates light emitted from the light
source 11 sequentially over its whole wavelength range at an
interval of, for example, 3 nm and sequentially outputs about 530
number of lights having a different wavelength. The scanning of the
wavelength by the active spectroscope 12 in the above-mentioned
wavelength range is executed repeatedly about 20 times or more in
one cycle time of the arterial volume waveform in the living body
13. In other words, the active spectroscope 12 transmits the lights
in a near infrared range sequentially at an interval of about 50 ms
or less and irradiates them to the living body 13. The light passed
through the living body 13 is detected by a light detector arranged
at the opposite side of the light source 11 and is converted into
an electric signal.
[0026] An output signal of the light detector 14 is supplied to an
instantaneous spectrum analyzer 15, wherein an absorption spectrum
obtained as an output of the light detector 14 for each wavelength
of the light source 11 is produced. That is, the output from a
sensor 16 that detects an intensity of the light incident to the
living body 13 from the light source 11, that is, an intensity of
the incident light I.sup..lamda..sub.o with each wavelength
(.lamda.) is supplied with the output signal of the light detector
14 to the spectrum analyzer 15. As described later, the intensity
of light (I.sup..lamda.) with each wavelength (.lamda.) passed
through the living body 13, that is, an absorbance (OD.sup..lamda.)
which is a ratio of the logarithmic intensity of passed light
I.sup..lamda. to that of the incident light I.sup.80 .sub.o
(OD.sup..lamda.=log I.sup..lamda..sub.o/I.sup..lamda.) is computed
here and an absorption spectrum is produced. Twenty (20) number of
the absorption spectrums are produced per second by twenty (20)
times of scanning per second of the active spectroscope 12 as
described above.
[0027] The absorption spectrum data obtained by the spectrum
analyzer 15 is stored in a spectrum data memory 17. The spectrum
data memory 17 stores and maintains output data for several seconds
of the spectrum analyzer 15 sequentially on the first-in first-out
basis.
[0028] Spectrum data read from the spectrum data memory 17 is
supplied to a subtraction processor 18 and a spectrum subtraction
(OD.sup..lamda..sub.ti'-OD.sup..lamda..sub.ti=.DELTA.OD.sup..lamda..sub.t-
i), which is composed of a difference in absorbance in
corresponding wavelengths (.lamda.) between the absorption
spectrums (OD.sup..lamda..sub.ti) at different times (ti) is
produced as described later. From this subtraction technique, the
subtraction data (.DELTA.OD.sup..lamda..sub.ti) include only
information of arterial blood without the other biological tissue
components such as skin, bones, muscles etc, as also described
later.
[0029] The spectrum analyzer 15, the spectrum data memory 17 and
the subtraction processor 18 are operated in sync with the 20 times
scanning per second of the active analyzer 12. The synchronization
between these units is made by a timing device 19 to supply a
synchronizing signal to them.
[0030] The spectrum subtraction data (.DELTA.OD.sup..lamda..sub.ti)
produced by the subtraction processor 18 is stored in a spectrum
subtraction memory 20. The spectrum subtraction memory 20 also
stores the output data of the subtraction processor 18 for several
seconds sequentially on the first-in first-out basis. It is noted,
as described later, that the subtraction data
.DELTA.OD.sup..lamda..sub.ti is mathematically derived to be equal
to the change in the intensity of the transmitted light
(I.sup..lamda..sub.ti'-I.sup..lamda..sub.ti=.DELTA.I.sup..lamda..sub.ti)
divided by the intensity of the light at the time ti
(I.sup..lamda..sub.ti), provided that .DELTA.I.sup..lamda..sub.ti
is very small, that is,
.DELTA.OD.sup..lamda..sub.ti=.DELTA.I.sup..lamda..sub.ti/I.sup..lamda..su-
b.ti. This means that the light intensity and its change are needed
in practical use without detection of the intensity of the incident
light (I.sup..lamda..sub.o).
[0031] The spectrum subtraction data read out of this spectrum
subtraction memory 20 is input into a blood glucose predictor 21.
The blood glucose predictor 21 is a device to predict blood glucose
concentration through the multi-regression analysis using the PLS
(Partial Least Squares Regression) method that is one of
multi-regression analyses from input spectrum subtraction data.
That is, the blood glucose predictor 21 is constructed as a
software model to compute the blood glucose concentration according
to the PLS method using whole blood samples that have many known
blood glucose concentrations.
[0032] FIG. 2 is a flow chart showing a method for constructing the
blood glucose predictor 21 as the software model shown in FIG. 1.
Known blood glucose concentration samples 31 are the whole blood
samples filled in plural quartz photo-cells whose glucose
concentrations are known and are slightly-different from each
other. These samples 31 were taken directly from, for example,
seven healthy adult males and were made the plural whole blood
samples having different albumin or hematocrit concentrations from
other blood samples by 18 mg/dl like 36, 54, . . . 486 mg/dl in the
glucose concentration range 30.about.450 mg. These samples 31 are
analyzed by a spectroscopic analyzer composed of the light source
11, the spectroscope 12, the light detector 14 and the spectrum
analyzer 15, and thus a absorption spectrum 32 is prepared. A PLS
regression analysis prediction model 34 is determined by data X
consisting of these absorption spectrum 32, together with
corresponding known n number of blood glucose concentrations (yn)
33. That is, data X consisting of the absorption spectrum 32 is an
absorbance for different m (about 530 waves) number of the
spectroscopic waveforms. Expressing these absorbance with x.sub.1,
x.sub.2, . . . , x.sub.m, the known n number of blood glucose
concentrations y1, y2, . . . , yn are approximated by the following
determinant using these variables: ( y1 y2 yn ) = ( a11 a1n a1n ann
) .times. ( X1 Xn ) Formula .times. .times. 1 ##EQU1##
[0033] A coefficient of this determinant is determined using the
PLS method by substituting the absorption spectrum data using the
above-mentioned sample solution into the determinant. A blood
glucose prediction model formula is thus obtained. Here, the PLS
method is a technique to consider the correlation of potential
variables T.sub.PLS as explanatory variables and to utilize data
contained in X as many as possible. y = Tq + f X = TPt + E S = ytT
} Formula .times. .times. 2 ##EQU2##
[0034] where, T: Potential variable
[0035] q: Potential variable regression coefficient
[0036] E, f: Residual of X, y
[0037] P: Loading matrix
[0038] S: Covariance of y and T
[0039] P of the determinant 2 and regression coefficient q of
potential variable T are determined by inputting blood glucose
y.sub.1, y.sub.2, . . . y.sub.n of n known blood glucose samples
into a regression analytical computer application software (for
example, Trade Name: MATLAB) according to the PLS method available
in the market. Thus, the regression analysis prediction model
(blood glucose computing model) according to the PLS method is
obtained. Then, a new T is computed based on P that is determined
when a model is prepared, when new absorbance of respective
spectroscopic wavelengths x.sub.1, x.sub.2, . . . , x.sub.m
obtained from blood of which blood glucose concentration is unknown
are input as data. These new absorbance of respective spectroscopic
wavelengths x.sub.1, x.sub.2, . . . , x.sub.m are input as spectrum
subtraction data read from the above-mentioned spectrum subtraction
memory 20. Using this new T and q determined when a model was
prepared, a blood glucose prediction value y.sub.i is obtained.
[0040] Next, the operations of the non-invasive blood glucose
concentration measuring instrument thus constructed according to
the embodiment of the present invention and the blood glucose
measuring procedures will be explained referring to FIG. 3 and FIG.
4.
[0041] As shown in FIG. 1, the light emitted from the light source
11 is spectroscopically scanned over the wavelength range by the
active spectroscope 12 at a rate of, for example, 20 times per
second and is irradiated to the living body 13. The light
transmitted through the living body 13 is detected by the light
detector 14 and each absorption spectrum is measured by the
spectrum analyzer 15 at an intervals of 40.about.50 ms. The
spectrum data thus measured is stored in the spectrum memory 20
until the next spectrum measuring time. FIG. 3 shows the arterial
pulsatile volume waveform in the living body 13, the horizontal
axis shows time and the vertical axis shows arterial blood volume
change (pulsatile volume waveform). Time t.sub.1, t.sub.2, . . . ,
t.sub.n in FIG. 3 show the time when the scanning of the wavelength
starts by the active spectroscope 12, where n is 20 in this case.
Absorption spectrum at the time t.sub.1, t.sub.2, . . . , t.sub.n
thus obtained are shown in FIG. 4, where the horizontal axis shows
the wavelength (.lamda.) and the vertical axis shows absorbance
(OD.sup..lamda..sub.ti; ti=t1, t2, - - - , tn).
[0042] Next, the spectrum subtraction processor 18 shown in FIG. 1
produces a spectrum subtraction from absorption spectrums at two
any optional times, for example, a time t.sub.1 and a peak time
t.sub.m in the arterial pulsatile volume waveform selected from the
times t.sub.1, t.sub.2, . . . , t.sub.n.
[0043] FIG. 5 is a diagram for explaining an operation of the blood
glucose predictor 21 shown in FIG. 1. One example of the
above-mentioned spectrum subtraction is shown in FIG. 5(a). The
horizontal axis in FIG. 5 shows the wavelength (.lamda.) and the
vertical axis shows a difference in the absorbance
(.DELTA.OD.sup..lamda..sub.ti). The curved line indicating the
spectrum subtraction is a plotted difference in the absorbance at
respective wavelengths of absorption spectrum, for example, at
t.sub.3 and t.sub.6 in this case.
[0044] Graphes (S1), (S2), . . . , (Sm) in FIG. 5 show absorption
spectrums of m number of whole blood samples of known blood glucose
concentration.
[0045] Spectrum subtraction data shown in FIG. 5 (a) are input to
the blood glucose concentration predictor 21. Further, a PLS
regression analytical model is incorporated in the blood glucose
concentration predictor 21. The PLS regression analytical model is
a numerical expression showing the relation between absorption
spectrums of m number of whole blood samples (S1), (S2), . . . ,
(Sm) shown in FIG. 5 each having known blood glucose concentration
and the known blood glucose concentrations corresponding to the
samples. The blood glucose concentration predictor 21 compares the
spectrum subtraction given from the spectrum subtraction memory 20
as input data with each of the absorption spectrums of the sample
solutions and outputs the blood glucose concentration of the sample
solution having the most similar absorption spectrum as a predicted
blood glucose concentration.
[0046] Thus, it is revealed that a blood glucose concentration can
be predicted at a high level of accuracy when spectrum subtraction
is used as input data to the blood glucose concentration predictor
21. The reason will be explained referring to FIG. 6. FIG. 6 is a
schematic diagram showing the relation of the intensity of incident
light I.sup..lamda..sub.o, the intensities of transmitted lights
I.sup..lamda..sub.1, I.sup..lamda..sub.2 and absorption amount in
the living body 13 at the wavelength .lamda.. The arterial blood
volume waveform P as shown in FIG. 3 is also shown in FIG. 6. In
FIG. 6, for example, the transmitted light intensity
I.sup..lamda..sub.1 (Incident light intensity I.sup..lamda..sub.o)
at t=t.sub.1 where the arterial blood volume waveform P becomes
minimal is (Incident light intensity
I.sup..lamda..sub.o)-(Absorption light intensity in the arterial
blood layer at the minimum volume change
I.sup..lamda..sub.3)-(Absorption light intensity in the venous
blood layer I.sup..lamda..sub.4)-(Absorption light intensity in the
biological tissues excluding blood I.sup..lamda..sub.5); that is,
I.sup..lamda..sub.1=I.sup..lamda..sub.o-(I.sup..lamda..sub.3+I.sup..lamda-
..sub.4+I.sup..lamda..sub.5) at t=t1. Further, the transmitted
light intensity I.sup..lamda..sub.2 at t=t.sub.m where the volume
change in the artery becomes maximal is (Incident light intensity
I.sup..lamda..sub.o)-(Absorption light intensity in the arterial
blood layer of the maximum volume change
I.sup..lamda..sub.6)-(Absorption light intensity in the venous
blood layer I.sup..lamda..sub.4)-(Absorption light intensity in the
biological tissues excluding blood I.sup..lamda..sub.5), that is,
I.sup..lamda..sub.2=I.sup..lamda..sub.o-(I.sup..lamda..sub.6+I.sup..lamda-
..sub.4+I.sup..lamda..sub.5) at t=tm. The differences of these two
transmitted light intensities
(I.sup..lamda..sub.1-I.sup..lamda..sub.2) extract the spectrum of
pulsative element .DELTA.I.sup..lamda. that is the pulsating
absorption intensity of the artery
(I.sup..lamda..sub.1-I.sup..lamda..sub.2=I.sup..lamda..sub.6-I.sup..lamda-
..sub.3=.DELTA.I.sup..lamda.). Although the absorption light
spectrum in the spectrum analyzer 15 or the spectrum data memory l
shown in FIG. 1 contains the absorption light element in the venous
blood and biological tissues excluding blood, the spectrum
subtraction (.DELTA.OD.sup..lamda.) generated in the spectrum
subtraction processor 18 becomes the light absorption spectrum
depending on the light absorption element of arterial blood
absorption element only. Because the subtraction
(.DELTA.OD.sup..lamda.) from the absorbance at t=t1
(OD.sup..lamda.1=log I.sup..lamda..sub.o/I.sup..lamda..sub.1) to
that at t=tm (OD.sup..lamda.2=log
I.sup..lamda..sub.o/I.sup..lamda..sub.2) is equal to log
I.sup..lamda..sub.2/I.sup..lamda..sub.1(=log(I.sup..lamda..sub.1-.DEL-
TA.I.sup..lamda.)/I.sup..lamda..sub.1=log(1-.DELTA.I.sup..lamda./I.sup..la-
mda..sub.1), and thus .DELTA.OD.sup..lamda. is nearly equal to
-.DELTA.I.sup..lamda./I.sup..lamda..sub.1 when
.DELTA.I.sup..lamda.<<I.sup..lamda..sub.1
(.DELTA.OD.sup..lamda..apprxeq.-.DELTA.I.sup..lamda./I.sup..lamda..sub.1)-
. Accordingly, this subtraction does not contain the absorption
element by the venous blood and biological tissues excluding blood.
Therefore, it becomes possible to eliminate influence of these
interfering factors and to put into practical use of a highly
precise non-invasive blood glucose concentration measuring
instrument.
[0047] By the way, in producing the spectrum subtraction
(.DELTA.OD.sup..lamda.) by the measurement of living body 13
described above, when a difference in arterial spectrum waveforms
that become the maximum and minimum volume changes in one heart
beat, the blood glucose concentration is computed at one time per
one heart beat and the blood glucose concentration is output at one
time per one heart beat. However, as spectrum data is measured
repetitively nearly 20 times in one heart beat, it is possible to
take out spectrum subtraction at two adjacent times as continuous
spectrum subtractions while shifting times sequentially and compute
blood glucose concentrations using these continuous spectrum
subtractions. In this case, it is expected that a change in
spectrums at adjacent times is very little, signal noise ratio of
spectrum subtraction drops and a fluctuation (a residual error) of
the result of blood glucose concentration computation may become
large. Accordingly, it is also possible to display the measured
result easy to look by inputting the result into the blood glucose
concentration predictor 21 by executing the time series average of
these spectrum subtractions or by smoothing successively computed
blood glucose concentrations through the statistical procedure such
as the time average or moving average by the blood glucose
concentration predictor 21.
[0048] Further, in the embodiment described above, the transmitted
light spectrum from the living body 13 is measured but the
reflected light from the living body 13 may be measured other than
the transmitted light.
[0049] FIG. 7 is a partial explanatory diagram showing this
embodiment, in which the same composed elements as those in FIG. 1
are assigned with the same reference numerals and the detailed
explanation thereof will be omitted. In this embodiment, the light
detector 14 is arranged at the same side as the light source 11 to
the living body 13 as illustrated and detects the reflected light
from the living body 13. By supplying the output signal of the
light detector 14 to the spectrum analyzer 15 shown in FIG. 1, it
is possible to measure blood glucose concentration likewise the
embodiment described above.
[0050] Further, in the embodiments shown in FIG. 1 and FIG. 7, the
light from the light source 11 is separated by the active
spectroscope 12 and then irradiated to the living body 13. However,
the transmitted light or reflected light may be separated for
spectrum analysis after the light from the light source 11 is
irradiated to the living body 13. For example, the light can be
separated by an array of plural light detectors each having a
sensitivity only for specific wavelengths (.lamda.).
[0051] Further, in the embodiments described above, a model applied
with the PLS method is used as the blood glucose concentration
predictor 21. However, a model according to the principal
constituents regression shown in Formula 3, which is one of
multi-regression analyses may be used. The regression analysis
blood glucose concentration computing model that is constructed
using the PCR method is expressed by the following Formula 3. Y =
.times. Tb + f = .times. t 1 .times. b 1 + t 2 .times. b 2 + + t n
.times. b n } Formula .times. .times. 3 ##EQU3##
[0052] where, T: Principal constituents score [0053] b: Principal
constituents score regression coefficient
[0054] That is, a multi-regression analysis blood glucose
concentration computing model is constructed by corresponding a
known blood glucose concentration of the whole blood sample 31 to
an objective variable y, applying spectrum data of the whole blood
sample 31 to an explanatory variable x and deciding a
multi-regression analysis blood glucose concentration computing
model. When spectrum subtraction data of an unknown blood glucose
concentration is input into the blood glucose concentration
predictor 21 in which this principal constituents score regression
coefficient b is set, a blood glucose concentration predict value
ya is computed and output.
[0055] Further, when developing a regression analysis prediction
model (a blood glucose concentration computing model) in the
above-mentioned embodiment, the sample 31 having a known blood
glucose concentration is filled in plural quartz cells and
absorption spectrum data X1, X2, . . . Xm are developed with a
spectroscopic analyzer comprising the light source 11, the
spectroscope 12, the light receiver 14 and the spectrum analyzer
15. However, for these absorption spectrum data X1, X2, . . . , Xm,
spectrum subtraction data obtained with units ranging from the
light source 11 to the spectrum subtraction memory shown in FIG. 1
using plural living bodies of which blood glucose concentrations
are known can be used.
[0056] Further, in the embodiment mentioned above, the measurement
of blood glucose concentration is shown. However, regarding the
measurement of concentration of another material having absorption
characteristics and scatter reflection characteristics existing in
the arterial blood, it is possible to predict and compute the
concentration of that material existing in the arterial blood
similarly. That is, it is possible to predict and compute the
concentration by measuring spectrum of wavelength band
corresponding to the absorption characteristics or the reflecting
characteristics of the material and deciding the regression
coefficient of the multi-regression analyzing model using the PLS
method or the PCR method referring to a concentration of a sample
of that is the standard of that material using the same system and
procedures shown in the above embodiment.
[0057] As described above, with the non-invasive blood constituent
measuring instrument and the method according to the embodiment of
the present invention, it is possible to measure blood constituents
in a living body by irradiating near infrared light to a finger
tip, etc. quickly and highly precisely without feeling pain and
burden involved in the blood drawing.
[0058] Further, according to the embodiment of the present
invention, spectrum subtraction using the arterial blood beat is
used as described above. However, the spectrum subtraction analysis
may be made by generating the venous blood volume change in the
biological tissues using such a method as the venous occlusion
method, for example. Thus, the adverse effect of other biological
tissue constituents is eliminated and blood constituent can be
measured at a highly precise and sensitive level.
* * * * *