Note: Descriptions are shown in the official language in which they were submitted.
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
METHODS FOR NONINVASIVELY MEASURING
ANALYTE LEVELS IN A SUBJECT
CROSS REFERENCE TO RELATED APPLICATIONS
This application claims the benefit of U.S. Provisional Application No.
60/686,721 entitled "Method for Noninvasively Measuring Blood Glucose" filed
June 2,
2005, the entire contents of which is hereby incorporated herein by reference.
This
application is also related to U.S. Provisional Application No. 60/671,285,
entitled
"Method For Data Reduction And Calibration Of An OCT-Based Blood Glucose
Monitor," filed April 14, 2005; and U.S. Application No. 10/916,236, entitled
"Method
And Apparatus For Monitoring Glucose Levels In A Biological Tissue," filed
August
11, 2004. The entire contents of both applications are hereby incorporated by
reference
herein.
FIELD OF THE INVENTION
The present invention relates generally to methods for noninvasively measuring
blood glucose or other analyte levels in a subject by measuring localized
changes in light
scattering from skin or other biological tissue. For example, such a method
can include
identifying tissue structures where the effect of blood glucose concentrations
or levels
are high, and targeting localized regions within the identified structures to
measure
blood glucose concentrations.
RELATED ART
Monitoring of blood glucose (blood sugar) levels has long been critical to the
treatment of diabetes in humans. Current blood glucose monitors employ a
chemical
reaction between blood serum and a test strip, requiring an invasive
extraction of blood
via a lancet or pinprick to the finger. Although small handheld monitors have
been
developed to enable a patient to perform this procedure anywhere, at any time,
the
inconvenience associated with this procedure -- specifically the blood
extraction and the
need for test strips -- has led to a low level of compliance by diabetic
patients. Such low
compliance can lead to diabetic complications. Thus, a non-invasive method for
monitoring blood glucose is needed.
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-2-
Studies have shown that optical methods can be used to detect small changes in
light scattering from biological tissue related to changes in levels of blood
sugar.
Although highly complex, a first order approximation of transmitting
monochromatic
light through biological tissue can be described by the following simplified
equation:
IR = Io exp[- (ao + fU.JL],
where IR is the intensity of light reflected from the skin, lo is the
intensity of the light
illuminating the skin, a is the absorption coefficient of the skin at the
specific
wavelength of the light, g is the scattering coefficient of the skin at the
specific
wavelength of the light, and L is the total path traversed by the light. From
this
relationship, it can be seen that the intensity of the reflected light decays
exponentially
as either the absorption or the scattering by the tissue increases. The
attenuation of light
can be characterized by an attenuation coefficient, which is the sum of s and
a.
It is well established that there is a difference in the index of refraction
between
blood serum/interstitial fluid (IF) and cell membranes (such as membranes of
blood cells
and skin cells). (See, R.C. Weast, ed., CRC Handbook of Chemistry and Physics,
70th
ed., (CRC Cleveland, Ohio 1989).) This difference can produce characteristic
scattering
of transmitted light. Glucose, in its varying forms, is a major constituent of
blood and IF.
The variation in glucose levels in either blood or IF changes the refractive
index of
blood-perfused tissue, and thus the characteristic of scattering from such
tissue. Further,
glucose-induced changes to the refractive index are substantially greater than
changes
induced by variation of concentrations of other osmolytes in physiologically
relevant
ranges. In the near-infrared (NIR) wavelength range, blood glucose changes the
scattering coefficient, g, more than it changes the absorption coefficient,
a. Thus,
optical scattering of the blood/IF and cell combination varies as the blood
glucose level
changes. Accordingly, there is the potential for non-invasive measurement of
blood
glucose levels.
Current non-invasive optical techniques being explored for blood glucose
applications include polarimetry, Raman spectroscopy, near-infrared
absorption,
scattering spectroscopy, photoacoustics, and optoacoustics. Despite
significant efforts,
these techniques have shortcomings, such as low sensitivity (signal-to-noise
ratio) for
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-3-
the glucose concentrations at clinically-relevant levels, low accuracy (less
than that of
current invasive home monitors), and insufficient specificity of glucose level
measurement within a relevant physiological range of 1.7-27.8 mM/L or 30-500
(mg/dL). For example, diffuse reflectance, or diffuse scattering, has been
explored as a
technique for noninvasively measuring levels of blood glucose. M. Kohl, Optics
Letters,
19(24) 2170-72 (1994); J. S. Maier, et al, Optics Letters, 19(24) 2062-64
(1994), Using
diffuse reflectance, a glucose-induced change of around 0.2% - 0.3% in the
scattering
coefficient per 18 mg/dL (or 1 mM/L) has been measured. This measured change
is too
small to be utilized efficiently for a blood-glucose monitor for home use.
Additionally,
glucose-induced changes to the scattering coefficient can be masked by changes
induced
by temperature, hydration, and/or other osmolytes. Accordingly, there is a
need for a
method to conveniently, accurately, and non-invasively monitor glucose levels
in blood.
Optical coherence tomography, or OCT, is an optical imaging technique that
uses
light waves to produce high-resolution imagery of biological tissue. OCT
produces
images by interferometrically scanning, in depth, a linear succession of spots
and
measuring absorption and/or scattering at different depths at each successive
spot. The
data then is processed to present an image of the linear cross section. The
key benefits
of such a system in imaging applications include the ability to achieve a high
resolution,
e.g., better than 10 micrometers, and the ability to select the depth at which
a sample can
be imaged. For example, blood vessels beneath the surface of the skin can be
imaged
using such a system.
As discussed in U.S. Application No. 10/916,236, and in R. O. Esenaliev, et
al,
Optics Letters, 26(13) 992-94 (2001), the entire disclosure of which is
incorporated by
reference, it has been proposed that OCT might be useful in measuring blood
glucose.
However, difficulties associated with this technique include the large number
of scans
required to reduce optical noise, or speckle, which arises from wavefront
distortion
when coherent light scatters from tissue. While an OCT imaging system can
reduce
speckle by averaging it out over many scans or measurements, this approach is
time-
consuming, which makes the use of a conventional OCT imaging system
impractical for
in-home monitoring of blood glucose levels. Additionally, an OCT imaging
system
requires complex processing to form a workable image and to analyze the image
data
sufficiently in order to determine glucose levels.
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-4-
Accordingly, there is a need for enhanced OCT systems for measuring analytes
such as blood glucose levels.
SUMMARY OF INVENTION
In accordance with the present invention, a method for non-invasively
measuring
glucose levels in blood is presented. Specifically, changes in a scattering
profile
produced from an OCT-based monitor are related to changes in blood glucose
levels by
focusing on highly localized regions of the scattering profile where changes
to the
scattering profile induced by temperature, hydration, and other osmolytes are
negligible.
Glucose-induced changes to the scattering coefficient measured from these
localized
regions range between about 2% and about 20% per I mM/L or 18 mg/dL, with an
average value of about 12% per 18 mg/dL. These percentage values are
significantly
higher than those measured using other methods. Additionally, within the
localized
regions, effects to the scattering coefficient induced by temperature,
hydration, and other
osmolytes are negligible compared to the effects of glucose, and, accordingly,
can be
ignored. The changes in the scattering profile can be related to changes in
glucose
concentrations by one or more mathematical algorithms.
A method for noninvasively measuring blood glucose-levels in biological tissue
is described herein. The method includes the steps of scanning a two-
dimensional area
of skin with a monitor based on non-imaging optical coherence tomography,
collecting
cross-sectional depth measurement data continuously during the scanning step,
and
identifying at least one localized region within the cross-sectional depth
measurement
data, wherein the at least one localized region corresponds to a structure
within the skin
where glucose-induced changes to the cross-sectional depth measurement data
are
prominent. Further, the method includes the step of relating the cross-
sectional depth
measurement data to blood glucose levels.
In one exemplary embodiment, a method for calibrating OCT measurements
using multiple light wavelengths is described to identify a tissue for
measurement. At
least two OCT scattering profiles can be obtained from light attenuated by a
subject's
tissue as a function of tissue depth. Non-limiting types of tissue include
vascular tissue
(e.g., a blood vessel wall), at least one component of blood, dermal tissue
surrounding
the vascular tissue, or some combination of the aforementioned types. The OCT
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-5-
scattering profiles can be obtained at different wavelengths of light such
that the tissue
can exhibit a different attenuation coefficient for each wavelength. The
attenuation of
light can be based at least in part on the presence of an analyte associated
with the tissue
(e.g., water or hemoglobin). The wavelengths can also be chosen such that the
tissue has
a different absorption coefficient at the two wavelengths. The wavelengths can
also be
chosen such that the scattering coefficient is larger than the absorption
coefficient at the
first selected wavelength, and optionally the absorption coefficient at the
second
wavelength is larger than the absorption coefficient at the first wavelength.
A localized
region (e.g., one or more depths) can be identified corresponding to a tissue
location of
OCT measurement calibration. Such calibration can be based upon a differential
comparison of the two OCT scattering profiles. A blood glucose measurement
(e.g.,
some type of chemical blood analysis measurement) can be associated with each
of the
OCT scattering profiles for calibrating other OCT measurements (e.g., using
the OCT
scattering profiles and blood glucose measurements to make a calibration
between
attenuation coefficient and blood glucose concentration). In general, the
localized can
have changing light attenuation coefficients based on the presence of blood
glucose or
other measurable analytes.
With respect to the exemplary method previously described, the OCT scattering
profiles can be normalized prior to differential comparison, with a depth
corresponding
to a tissue location for OCT measurement calibration depending upon a
differential
comparison of normalized OCT profiles (e.g., subtracting one normalized
profile from
another at corresponding depth locations). Normalization can be performed by
dividing
the scattering data of a respective OCT profile by the profile's respective
peak intensity
value. One or more extrema points in the differential comparison of normalized
OCT
profiles can be identified, and subsequently correlated with the depth of the
tissue
location or some other measure of the localized region corresponding with the
tissue
location.
In general, an offset location and an interval can define a localized region
of an
OCT scattering profile that can be correlated with a particular attenuation
coefficient.
The offset can correspond with a depth of a tissue location, and the interval
can be
determined from the offset location and the OCT scattering profile. The offset
location
and interval can be used to define the region of the OCT scattering profile in
which a
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-6-
slope measurement can be correlated with the attenuation coefficient (or the
scattering
coefficient when absorption effects are small).
Another exemplary embodiment is directed to a method of determining an
absorption coefficient in OCT measurements using multiple light wavelengths.
Two or
more OCT scattering profiles can be obtained as a function of subject tissue
depth at
different wavelengths of light such that the tissue has a larger scattering
coefficient than
absorption coefficient at a first selected wavelength (e.g., the scattering
coefficient being
at least about 5 times greater than the absorption coefficient). A scattering
coefficient
can be determined from the first OCT scattering profile (e.g., by locating a
slope in the
first OCT scattering profile). An estimate of a scattering coefficient from
the second
OCT scattering profile can be obtained from the scattering coefficient of the
first OCT
scattering profile. Such an estimate can be obtained using scattering theory
(e.g., Mie
scattering). The absorption coefficient of the second OCT scattering profile
can be
determined using the estimate of the scattering coefficient at the second
selected
wavelength. A similar method can also be used to determine a scattering
coefficient.
Another method consistent with an embodiment of the invention is directed to
calibrating OCT measurements using multiple light wavelengths. Two or more OCT
measurements can be obtained as a function of time using different wavelengths
of light
for each measurement. The wavelengths can be chosen such that the tissue has a
larger
absorption coefficient at a first selected wavelength relative to a second.
Such an
absorption coefficient can also depend upon the presence of an analyte (e.g.,
water) in or
around the tissue. One wavelength can also be chosen such that the scattering
coefficient exceeds the absorption coefficient by at least about a factor of
five. A first
OCT measurement can be converted into an analyte measurement as a function of
time.
The analyte measurement can be used to calibrate a scattering coefficient
measurement
as a function of time.
BRIEF DESCRIPTION OF THE DRAWINGS
The present invention will be more readily understood from the detailed
description of the embodiment(s) presented below considered in conjunction
with the
attached drawings, of which:
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-7-
FIG. 1 illustrates a process flow of a method for measuring blood glucose;
FIG. 2 is a graphical illustration of a typical scattering cross section from
a patch
of human skin measured using an OCT-based monitor;
FIG. 3 is an example of an intensity difference plot, according to an
embodiment of the
present invention;
FIGS. 4A and 4B are graphical illustrations in which scattering
discontinuities
are identified according to an embodiment of the present invention; and
FIG. 5 is a graphical illustration of an absorption effect of water at
multiple
wavelengths, according to an embodiment of the present invention;
FIGS. 6A and 6B are examples of scattering profiles at wavelengths of 1310
nanometers and 1440 nanometers, respectively, according to an embodiment of
the
present invention; and
FIG. 7 is a graphical illustration of a differential data set scattering
profile,
according to an embodiment of the present invention.
DETAILED DESCRIPTION OF THE INVENTION
According to an embodiment of the present invention, a method for measuring
blood glucose levels includes the step of utilizing an OCT-based sensor to
take
scattering cross-sectional depth measurements on a small area of biological
tissue or
skin. The OCT-based sensor can be a non-imaging system such as that described
in
detail in U.S. Application No. 10/916,236. In some embodiments, a two-
dimensional
area of the skin can be scanned, preferably either in a circular pattern,
e.g., with a radius
no greater than about 2 mm, or in a filled disk or filled rectangular pattern
where the
pattern is drawn randomly. As the OCT-based sensor scans the two-dimensional
pattern
continuously, the sensor continuously collects data corresponding to cross-
sectional
depth measurements within the biological tissue. Other embodiments can utilize
an
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-8-
OCT-based sensor to obtain cross-sectional depth measurements with two-
dimensional
scanning.
By continuously acquiring a two-dimensional scan pattern and continuously
taking cross-sectional depth measurements, the noise associated with OCT
sensing,
often referred to as "speckle," is reduced more efficiently than scanning
tissue using a
step-scan method, as described in A. Kholodnykh, et al, Applied Optics, 42(16)
3027-37
(2003), In Kholodnykh, a method proposed for an OCT-based system includes
scanning
a two-dimensional pattern using a step-scan process, where the OCT-based
system light
beam picks a spot on the skin and takes multiple depth scans. The OCT-based
system
then averages these depth scans to reduce speckle, and moves on to another
spot on the
skin, takes multiple depth scans, and averages the depth scans. The OCT-based
system
repeats this process until a two-dimensional pattern has been made.
In accordance with an embodiment of the present invention, the OCT-based
monitor continuously scans an area of skin and continuously collects data.
Using this
method, fewer scans in less time are required to produce sufficient results.
To further
reduce speckle, a number of OCT scans can be averaged to produce an average
OCT
scan result. Thus, data associated with a particular OCT scan at a specific
point in time
is actually an averaged result of a group of OCT scans.
Using the cross-sectional depth measurements, an intensity profile, or
scattering
profile, can be generated. Within the scattering profile, localized regions
where changes
to the scattering profile are dominated by changes in blood glucose can be
identified. To
locate these regions, a second-derivative plot can be generated, as disclosed
in U.S.
Provisional Application No. 60/671,285. Using the second-derivative plot,
discontinuities in the scattering profile are exaggerated and easily
visualized. These
discontinuities represent structures in the skin where changes in blood
glucose levels
dominate the scattering profile. Within these highly localized regions,
changes to the
scattering profile induced by temperature, hydration, and other osmolytes,
such as
sodium, potassium, and urea, are very small compared to the effects of
glucose, and
therefore, can be ignored.
By focusing on the localized regions identified in the intensity profile,
there are
multiple means that can be used to correlate efficiently the scattering
profile to blood
glucose levels. Upon identifying these localized regions, the data of the
scattering
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-9-
profile can be related to blood glucose levels using one or more mathematical
algorithms
such as, for example, an algorithm relating the slope of a portion of the OCT
data curve
to blood glucose levels, where the portion of the OCT data curve corresponds
to a
discontinuity in the scattering profile. An example of such an algorithm is
disclosed in
detail in U.S. Provisional Application No, 60/671,285. Optionally, the
scattering profile
can be related to blood glucose levels by utilizing a magnitude of the glucose-
induced
localized change, either using a straight peak intensity measurement or using
an
integrated intensity measurement where each region integrated corresponds to a
localized region identified in the second-derivative plot. Alternatively, the
scattering
profile can be related to blood glucose levels using a change in full width at
half-
maximum measurement of one or more of the localized regions identified in the
second-
derivative plot. In addition, the scattering profile can be related to blood
glucose levels
using an angle computation, where the angle corresponds to a peak change in a
localized
region and an arbitrary depth.
Another aspect of the embodiment of the present invention includes identifying
localized regions of change in the scattering profile by utilizing an
intensity difference
plot (IDP), which is described in detail in U.S. Provisional Application No.
60/671,285.
Although an IDP requires a significant change in glucose concentrations, such
as, for
example, the change caused by the subject ingesting food during the course of
the testing
time period while OCT scans are taken, one or more localized regions in the
data curve
that correspond to tissue structures where noticeable changes to the
scattering profile
were produced by changes in blood glucose levels can be identified. Once the
localized
regions are identified, the scattering profile from the localized regions can
be related to
blood glucose levels using the algorithms mentioned above.
Yet another aspect of the embodiment of the present invention includes using a
multiple-wavelength method to identify localized regions of the scattering
profile that
correspond to tissue and/or tissue structures, such as blood, blood vessels,
or other
tissue, where changes in the scattering profile due to presence of one or more
analytes,
such as blood glucose levels, are detectable. The term "wavelength" is used
herein to
define a region of the electromagnetic radiation spectrum that is
distinguishable from
other regions. While laser sources with narrow linewidths can be preferable,
other lower
resolution, or even broadband light sources, can also be used. For example,
the
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-10-
invention can be practiced with two wavelengths of light, one of which niight
be a
multimode source spanning several nanometers, e.g., 1308-1312 nm or 1438-1442
nm.
As disclosed in U.S. ApplicationNo. 10/916,236, the OCT-based monitor can be
constructed such that multiple wavelengths of light are employed to illuminate
the skin.
Light from multiple wavelengths is absorbed differently by different
biological
constituents, which differentially reduces the intensity of the scattered
light. Moreover,
light reflected in and around tissue can be partially absorbed by a
constituent for that
wavelength. The constituent, in or around the tissue, for that wavelength
absorbs some
of the light according to the specific wavelength and/or the analyte level in
or around the
tissue. The differences in the scattering and absorption properties produced
by multiple
wavelengths interacting with different constituents provide for a
determination of an
optimal correlation between the scattered signal and a chosen analyte level.
For
example, light reflection and absorption in and around particular tissues and
tissue
structures can be correlated with the presence of glucose to provide a
measurement of
blood glucose levels. Potential tissues and tissue structures, whose light
interaction can
be correlated with blood glucose levels, include (but are not limited to)
vascular tissue
(e.g., blood vessel walls), blood and its components (e.g., cells), derm.al
tissue surround
blood vessels, and any combination of the aforementioned tissues and/or tissue
structures.
The wavelengths can be chosen to provide an optimal contrast between the
absorption and scattering effects of blood and other biological constituents,
such as
water. For instance, the wavelengths can be chosen to accentuate contrast
regarding the
presence of a particular analyte that is a signature of the presence of a
tissue or tissue
structure desired to be targeted by OCT measurements (e.g., water being a
signature of
the presence of blood perfused tissue). A first wavelength of light emitted
from the
OCT-based monitor can be chosen such that there is minimum absorption of the
light by
water compared to the scattering effect, which makes the absorption effects
corresponding to water negligible, i.e., the total attenuation coefficient (p,
+ga) is
dominated by the scattering coefficient contribution. In general, when one of
the
coefficients dominates another in the total attenuation coefficient, we can
assume that
the less dominant coefficient can be ignored. For example, we can say that s
a when
the scattering coefficient is at least about 5 times, or at least about 10
times, greater than
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-11-
the absorption coefficient. If a second wavelength is chosen to provide peak
absorption
of light by water, then the difference in light attenuation between the two
wavelengths
can be used to indicate the position in depth of a blood perfused tissue
structure, such as
a blood vessel. Clearly, three or more wavelengths of light can also be used
to generate
corresponding OCT profiles, with specific wavelength pairs utilized in a
combination to
generate a corresponding light attenuation difference.
According to this aspect of the embodiment of the present invention, OCT scans
are taken at two different wavelengths of light, where the first wavelength is
chosen such
that the scattering effects are dominant over absorption effects of water, and
the second
wavelength is chosen such that there is substantial absorption by water.
Preferably, the
scattering data sets produced by scanning a two-dimensional area of the skin
by the first
and second wavelengths are normalized by finding the peak data point in each
scattering
data set and dividing all data points by the xespective peak data point. Thus,
each
normalized scattering data set is now a set of decimal values with each peak
data point
having a value of 1Ø
The normalized scattering data set of the second wavelength can be subtracted
from the normalized scattering data set of the first wavelength to produce a
differential
scattering data set over the depth of the OCT signal, for a specific point in
time. As
discussed in U.S. Provisional Application No. 60/671,285, there are two
variables or
parameters associated with fitting the OCT data to blood glucose levels in
order to
achieve the best correlation. These variables are an offset and an interval.
An "offset" is
the depth of the OCT data curve at which to begin correlating the OCT data to
the blood
glucose levels. An "interval" is a certain portion or segment of the OCT data
curve that
is measured from the offset. For each OCT data curve there are numerous
potential
combinations or pairs of offsets and intervals. By identifying a peak value in
a
differential scattering data set, an offset (depth) can be obtained that
corresponds to
tissue structures where glucose-induced changes are dominant. If a linear fit
from the
peak value to another data point of the differential data curve is generated,
the linear fit
corresponds to an offset and interval combination where the slope of the
offset and
interval combination is highly correlated to blood glucose levels, i.e., the
offset and
interval can define the localized region of an OCT data curve in which an
appropriate
attenuation coefficient can be identified and correlated with a blood glucose
level. Thus,
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-12-
each linear fit can identify a localized region where glucose-induced changes
to the
scattering profile are predominant. In cases where the OCT data curve is
generated
using a wavelength of light in which p,, a,, the attenuation coefficient
reduces to a
scattering coefficient.
As mentioned above, by focusing on these highly localized regions along the
scattering profile, structures within the skin can be identified where changes
to the
scattering profile induced by glucose are high, which allows effects induced
by
temperature, hydration, and other osmolytes to be ignored.
This method focuses on the relative depths of certain structures within the
skin,
namely, capillaries where blood vessels are prevalent, and identifies the
regions of the
depth scan that correlate to these structures where glucose levels are known
to fluctuate
significantly. By focusing on these highly localized regions of a scattering
profile,
glucose-induced changes to the scattering coefficient of about 2% to about 20%
per 18
mg/dL can be obtained, which is significantly higher than the 0.2% - 0.3%
obtained
using other noninvasive optical scattering methods.
A method for measuring blood glucose levels noninvasively is summarized in the
flow chart presented in FIG. 1. According to an embodiment of the present
invention, at
step S101, a anon-imaging OCT-based monitor, or a "sensing" OCT-based monitor,
can
be utilized to take multiple scattering cross-sectional depth measurements on
an area of
skin. The OCT-based monitor continuously scans a two-dimensional area of skin,
preferably scanning either a circle, a filled disk, or a filled rectangular
pattern, where the
filled pattem is drawn randomly. As the OCT-based monitor scans the skin, the
monitor
continuously collects cross-sectional depth measurements. As discussed above,
continuously scanning a two-dimensional area of skin while continuously
collecting data
reduces speckle faster than previously known methods that use an OCT-based
monitor.
Additionally, fewer scans are required to average out speckle and thus, less
time is
required to take the scans.
At step S102, the cross-sectional depth measurements can be utilized to create
a
scattering profile in which the OCT data curve is plotted over time. FIG. 2
shows a
scattering profile of light scattered from human skin as measured via an OCT-
based
monitor, according to an embodiment of the present invention. If an
appropriate
wavelength of light is chosen (e.g., around 1300 nanometers) where "an
appropriate
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-13-
wavelength of light" is one in which the absorption coefficient of the light,
A, is small
relative to the scattering coefficient, s, of the light by the skin. A change
in the OCT
signal (e.g., a change in the slope of a portion of an OCT profile) likely
will be
dominated by glucose-induced changes in the tissue scattering. Based on the
wavelength of light chosen, the OCT data curve spikes at certain regions of
the surface
of the skin and then falls dramatically within the epidermis region of the
skin. The OCT
data curve also rises and slowly decreases within the dermis region as the
depth of light
in the skin increases. As shown in FIG. 2, the slope of the OCT data curve can
increase
or decrease relative to the blood glucose level. That is, the slope of the OCT
data curve
will change in response to glucose level changes in very small defined
regions. Because
most blood vessels are located in the dermis region, it is this portion of the
OCT data
curve that provides data for measuring blood glucose levels. To identify this
region, one
or more of the graphs described below can be generated.
At step S103, an intensity difference plot (IDP) can be generated to highlight
one
or more regions of the OCT data curve that correspond to tissue structures
where
glucose-induced changes are dominant. An example of an intensity difference
plot is
illustrated in FIG. 3. As described in U.S. Provisional Application
60/671,285, two
OCT scans are selected and the difference in the OCT data between the selected
two
OCT scans is computed. The differential data can then be plotted to produce an
IDP, as
shown in FIG. 3. From the IDP, one or more zero-crossing points can be
identified as
well as localized extrema surrounding the zero-crossing points, respectively.
The IDP in
FIG. 3 has one zero-crossing point, which is located at a depth of about 225
microns. A
local maximum data point is located at around 200 microns and a local minimum
point
is located at around 350 niicrons. The region of the localized extrema
represents a
highly localized region where glucose-induced changes to the scattering
coefficient are
the dominant effect within a tissue structure, and is represented in FIG. 3 by
a shaded
box. To relate the OCT data to blood glucose levels, the highly localized
region can be
focused upon and data falling outside this region can be ignored. Within this
region,
effects due to temperature, hydration, and other osmolytes are negligible.
Optionally,
the box can be expanded to include potential offsets within a variance amount
of the
localized extrema. For example, in FIG. 3, the range of potential offsets
includes offsets
from 175 microns to 400 niicrons.
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-14-
According to another aspect of the embodiment, at step S 103, the scattering
profile can be used to generate a second-derivative plot. As described in U.S.
Provisional Application 60/671,285, discontinuities in the scattering profile
represent
structures in the skin where changes due to variations in blood-glucose levels
are high
relative to changes in the scattering profile induced by other analytes. The
second-
derivative plot enhances these discontinuities to help identify one or more
highly
localized regions where the scattering profile can be related to blood glucose
levels.
FIGS. 4A and 4B graphically illustrate how a second-derivative plot enhances
discontinuities in the scattering profile. In FIG. 4A, a scattering profile is
plotted against
the depth of the scanned area of skin. Discontinuities in the scattering
profile are
identified by circles in the graph, however, these discontinuities typically
are difficult to
visualize. In FIG. 4B, a square of a second derivative of the scattering
profile is plotted
against the depth of the scanned area of skin. The discontinuities in the
scattering
profile are enhanced by the second derivative computation, while calculating
the square
value of the second derivative removes any negative values that can exist. The
discontinuities correspond to structures in the skin where changes in blood
glucose
levels are donzinant, such as, for example, blood vessels. The scattering data
corresponding to the identified localized regions can then be related to blood
glucose
levels.
Another aspect of the embodiment includes utilizing multiple wavelengths to
identify tissue and/or tissue structures with a high degree of hydration or
water content
due to blood perfusion, such as blood vessels where changes in blood glucose
levels are
prevalent, at step S103. The localized regions of the scattering profile that
correspond to
these tissue structures then correlate well to blood glucose levels. As
described above,
the OCT-based monitor can utilize multiple wavelengths of light, where one
wavelength
is chosen that produces a minimum absorption of light by water in the
interstitial fluid,
and another wavelength is chosen that provides a substantial absorption of
light by
water. FIG. 5 illustrates the absorption of light by water at different
wavelengths. For
example, if a first wavelength of light at 1310 nanometers (nm), where the
absorption
effects of water are minimal, and a second wavelength of light at 1440
nanometers (nm),
where the absorption effects of water are maximized, are chosen, the
differential
scattering data set produced from the OCT data of the two wavelengths can be
used to
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-15-
determine tissue structures where hydration is high, such as a blood vessel.
Of course,
other analytes indicative of a tissue or tissue structure can also be detected
by the choice
of appropriate light wavelengths. For example, hemoglobin has a peak
absorption at 660
nm when deoxygenated and 940 zun when oxygenated. Accordingly, either of these
wavelengths can be useful to detect oxygen levels in tissue. It can also be
advantageous
to select light wavelengths such that the scattering due to the presence of a
measured
analyte (e.g., blood glucose or hemoglobin) does not differ a great deal in
the two
wavelengths, i.e., the difference in intensity of the two wavelengths is due
mostly to the
presence of water or some other analyte indicative of the presence of a blood
vessel or
other tissue structure.
The scattering profiles for each wavelength at a particular point in time can
be
plotted, as shown in FIGS. 6A and 6B, which represent exemplary scattering
profiles for
first and second wavelengths of 1310nm and 1440 nm, respectively. In both
FIGS. 6A
and 6B, the scattering data set for each wavelength has been normalized using
the
respective peak intensity value. Thus, the peak intensity value for each
scattering data
set is 1.0, and each data point around the peak is less than 1Ø Because the
sensitivity of
the OCT-based monitor is different at the two wavelengths, the scattering
profiles of the
two wavelengths can not be compared directly. Normalization of the scattering
data sets
allows direct comparison of the scattering data sets from the two wavelengths.
Upon normalizing the data, the normalized scattering data set of the second
wavelength can be subtracted from the normalized scattering data set of the
first
wavelength to produce a differential scattering data set. Using the exemplary
wavelengths of 1310 nm and 1440 nm, a differential data curve plot can be
produced, as
shown in. FIG. 7. The profile of the differential data curve suggests one or
more offset
and interval pairs that correspond to localized regions of the scattering
profile where
variations in blood glucose levels are the predoniinant effect. One or more
peak data
points identified in the differential data curve suggests one or more depths
or offsets at
which to begin correlating the OCT data to blood glucose levels. Using the
peak data
point(s), one or more intervals can be identified by choosing one or more data
points on
either side of the peak data point(s). The combination of the offset(s) and
the one or
more intervals produces offset and interval pairs that can be applied to the
scattering
profile produced by the first wavelength, e.g., 1310 nrn, to identify
localized regions
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-16-
where glucose-induced effects to the scattering profile are predominant.
Upon identifying one or more highly localized regions of the scattering
profile
where glucose-induced changes are prominent, one or more algorithms can be
used to
relate the scattering profile to blood glucose levels, at step S 104. At step
S 104a, the
slope of portions or segments of the IDP data curve that correspond to the
localized
regions can be used to compute predicted blood glucose levels, as described in
U.S.
Provisional Application 60/671,285. Alternately, at step S104b, the scattering
profile
can be related to blood glucose levels using a magnitude value of a localized
change,
either using a straight peak intensity measurement or an integrated intensity
measurement using the entire localized region. Another option is to use a
change in the
full-width at half-maximum measurement of one or more of the localized
regions, at step
S 104c. Yet another option is to use an angle measurement calculation in
relating the
OCT data to blood glucose levels.
The description of using multiple wavelengths to locate tissue or tissue
structures
for glucose monitoring is not intended to limit the use of the technique to
the particular
application exemplified in the description. Indeed, beyond identifying the
presence of
water or hydration content of blood vessels, other analytes such as hemoglobin
at
varying oxygen content can also be utilized as a signature of a particular
tissue or tissue
structure (e.g., oxygenated tissue). As well, the types of tissue and tissue
structures to
which multiple wavelength OCT measurements can be used are not limited to
blood
vessels but can include other vascular tissue, blood (or particular
constituents thereof
such as cells), dermal tissue surround vascular tissue, and combinations of
such
exemplary tissues and tissue structures.
Furthermore, the technique is not limited to detecting blood glucose, but can
be
used to diagnose other conditions unrelated to blood glucose. In one instance,
the
technique of using of multiple wavelengths to determine tissue hydration
levels can be
applied in a variety of contexts including assessment and/or monitoring of
congestive
heart failure, management of fluid therapy for shock or surgery, management of
fluid
load in dialysis patients (e.g., peritoneal dialysis or hemodialysis), and
management of
tissue hydration in pulmonary disease and hypertension. For example, multiple
wavelength OCT measurements can be used to monitoring clotting factors in
blood.
Since the scattering coefficient of blood is affected by hydration, use of the
multiple
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-17-
wavelengths allows one to deterniine the contribution to the scattering
coefficient that is
substantially hydration independent by comparing scattering coefficients at
wavelengths
that absorb water strongly and weakly. The scattering coefficient at low water
absorbing
wavelengths can be related to the viscosity, and eventually the clotting
factors of the
blood. Such a measurement could be useful in post-surgical monitoring of
patients who
are administered blood thinning agents. The scattering coefficient at low
water
absorbing wavelengths can also be adjusted using the measurements at higher
water
absorbing wavelengths. When using any of the calibration methods encompassed
by the
present application, actual samples of the measured analyte (or other non-
chemically
oriented types of analyte measurements) can be utilized to aid in calibration
(e.g., the use
of blood glucose samples as described with reference to glucose monitoring
herein).
In a further aspect of the embodiment, multiple wavelength OCT measurements
can be utilized to provide an improved estimate of a scattering coefficient or
an
absorption coefficient from tissue measurements. Such an aspect can be
utilized in
conjunction with any of the potential applications of the present invention
such as
determining the viscosity of blood. The following description is with
reference to
estimating an absorption coefficient, though estimates of a scattering
coefficient can also
be obtained under analogously consistent conditions.
In one example, a pair of OCT scattering profiles are obtained, each profile
corresponding to a measurement at a particular wavelength of light. With
reference to
the S 10 1 and S 102 of the flowchart of FIG. 1 and the corresponding
description, the
profiles can be obtained by scanning a two-dimensional area of skin to obtain
measurements at a number of cross-sectional depths. In this particular
example, one
profile is obtained using light with a wavelength of about 1310 nm and another
profile is
obtained using 1440 nm light. The intensity of the reflected light at 1310 nm
can be
approximated by the following equation:
13]
~ IR _ __Lr~S310 + P1n310 I
'lo L
where IR is the reflected light intensity at 1310 nm, Io is the initial light
intensity at 1310
nm, L is the total light pathlength, p, 1310 is the scattering coefficient of
the tissue at 1310
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-18-
nm, and Va1310 is the absorption coefficient of the tissue at 1310 nm.
In some instances, a wavelength can be selected such that one of the
scattering or
absorption coefficients is stronger than the other to the extent that the
contribution of the
weaker can be ignored (e.g., when one contribution is at least about 5 times
greater or at
least about 10 times greater than the other). When measuring hydration levels,
the
scattering coefficient Vy1310 is stronger than the absorption coefficient
ILa131 such that the
contribution from a131 can be ignored; this allows the scattering
coefficient P.,1310 to be
determined. Accordingly, a plot of ln(IR/Io) versus depth can yield a line
with a slope
that can be equated with s13111
The scattering coefficient at 1310 nm us1310 can be used to provide a measure
of
the scattering coefficient at 1440 nm, p.,1440 Various scattering theories, as
known to
those skilled in the art, can be used to relate the scattering coefficients at
the two
different wavelengths. For example, under Mie scattering, (0.7)ps131 z P.1440
Using
this estimate for E,cs1440, an estimate of the absorption coefficient at 1440
nm can be found
using:
1440
ln IR = _LI~s440 + P1a4401
Io LLL J
where IR is the reflected light intensity at 1440 nm, Io is the initial light
intensity at 1440
nm, L is the total light pathlength, P., 1440 is the scattering coefficient of
the tissue at 1440
nm, and pa1440 is the absorption coefficient of the tissue at 1440 nm. The OCT
profile at
1440 nm, along with the estimated scattering coefficient P,1440 can allow one
to
determine [61440
As previously mentioned, the outlined technique can also be used to determine
scattering coefficients when a scattering profile utilizes a wavelength in
which an
absorption coefficient dominates (e.g., an absorption coefficient is measured
using a
wavelength where absorption dominates attenuation, followed by estimating an
absorption coefficient at a second wavelength and determining the scattering
coefficient
at the second wavelength). Those skilled in the art will appreciate that the
technique can
also be applied with respect to other analytes besides water (e.g.,
hemoglobin) when
appropriate wavelengths of light are chosen.
CA 02610249 2007-11-29
WO 2006/130847 PCT/US2006/021535
-19-
As discussed in U.S. Application No. 10/916,236, the use of multiple
wavelengths can also provide an additional sensor calibration technique. Using
water
detection as an exemplification of calibrating analyte effects on OCT
measurements, the
scattering coefficient of a first wavelength OCT measurement can drift even
though the
glucose concentration remains static because of the change in the scattering
coefficient
due to hydration changes. Thus, by measuring the skin hydration using a second
wavelength in which the wavelength is selected such that the resulting
scattering profile
tracks hydration changes (e.g., the absorption coefficient at the second
wavelength is
high for water, and much higher relative to the absorption coefficient at the
first
wavelength), this drift can be compensated for and the OCT sensor can maintain
calibration. Clearly, other analytes that can effect scattering coefficient
measurements
can also be compensated for using this technique.
While the present invention has been described with respect to particular
embodiments discussed herein, it is to be understood that the invention is not
limited to
the disclosed embodiments. To the contrary, the invention is intended to cover
various
modifications and equivalent arrangements included within the spirit and scope
of the
appended claims. The scope of the following claims is to be accorded the
broadest
interpretation so as to encompass all such modifications and equivalent
structures and
functions.
What is claimed is:
