U.S. patent application number 12/383930 was filed with the patent office on 2012-01-12 for correlation technique for analysis of clinical condition.
This patent application is currently assigned to Ottawa Hospital Research Institute. Invention is credited to Neil Lagali, Rejean Munger.
Application Number | 20120008130 12/383930 |
Document ID | / |
Family ID | 39229671 |
Filed Date | 2012-01-12 |
United States Patent
Application |
20120008130 |
Kind Code |
A9 |
Munger; Rejean ; et
al. |
January 12, 2012 |
CORRELATION TECHNIQUE FOR ANALYSIS OF CLINICAL CONDITION
Abstract
The present invention provides a correlation technique for
analysis of changes in bodily fluid and/or tissue in order to
identify or monitor appearance, progression or treatment of a
disease or condition in a subject. The disclosed method involves
measuring spectral properties or changes in bodily fluid and/or
tissue of a subject using at least two optical techniques; and
correlating the measured properties or changes to a corresponding
clinical condition or change in clinical condition, respectively.
The measure of spectral changes over time can be used as indicators
of changes in the clinical condition, for example, in disease
treatment and/or disease regulation. This method is particularly
useful for identifying a disease state and for monitoring
efficacies of therapies used to treat different diseases or
disorders, for example, renal dialysis.
Inventors: |
Munger; Rejean; (Orleans,
CA) ; Lagali; Neil; (Ottawa, CA) |
Assignee: |
Ottawa Hospital Research
Institute
|
Prior
Publication: |
|
Document Identifier |
Publication Date |
|
US 20090303462 A1 |
December 10, 2009 |
|
|
Family ID: |
39229671 |
Appl. No.: |
12/383930 |
Filed: |
March 30, 2009 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
PCT/CA2007/001706 |
Sep 21, 2007 |
|
|
|
12383930 |
|
|
|
|
60827605 |
Sep 29, 2006 |
|
|
|
Current U.S.
Class: |
356/39 ;
356/73 |
Current CPC
Class: |
A61B 5/0059 20130101;
G01N 21/31 20130101; A61B 5/72 20130101; A61B 5/14551 20130101;
A61B 5/0075 20130101; G01N 21/65 20130101; G01N 21/64 20130101 |
Class at
Publication: |
356/39 ;
356/73 |
International
Class: |
G01N 33/48 20060101
G01N033/48; G01N 21/00 20060101 G01N021/00 |
Claims
1. A method for investigating a clinical condition of a subject,
comprising the steps of: (a) measuring spectral properties of
bodily fluid or tissue of said subject using at least two optical
techniques; and (b) correlating the spectral properties to a
corresponding clinical condition, wherein said spectral properties
are measured in ultra-violet wavelengths, visible wavelengths,
near-infrared wavelengths, mid-infrared wavelengths or a
combination thereof.
2. The method of claim 1, wherein said correlation step includes
deriving from the measured spectral properties an index or indices
correlating with the clinical condition of said subject at a single
point in time, over a period of time, quasi-continuously or
continuously, in response to at least one effector.
3. The method of claim 2, wherein said at least one effector is a
single treatment, multiple treatments, a single dose of a medicine
or multiple doses of a medicine.
4. The method of claim 2, wherein said clinical condition is
indicative of long-term health of the subject, subjective or
objective clinical outcome in said subject, or a potential future
change in clinical status or state of health of said subject.
5. The method of claim 4, wherein said potential future change in
clinical status or state of health of said subject is a
predisposition to disease.
6. The method of claim 1, further comprising the step of comparing
the measured spectral properties to spectral properties obtained
from a population of subjects using said at least two optical
techniques, wherein said population of subjects has a known average
clinical condition.
7. The method of claim 1, wherein the clinical condition is a
disease state.
8. The method of claim 1, wherein the at least two optical
techniques include a combination of any two of the following
phenomena: diffuse reflection, inelastic (Raman) scattering,
absorption/transmission, or fluorescence.
9. The method of claim 1, wherein the bodily fluid is whole blood
or blood serum.
10. The method of claim 1, further comprising the step of
processing raw spectral data to obtain said measured spectral
properties.
11. The method of claim 10, wherein said processing step comprises
one or more of the following: (a) normalizing raw spectral data
relative to a spectral data obtained from a baseline sample; (b)
normalizing raw spectral data relative to total integrated power;
(c) combining chosen spectral bands and aggregating spectral data
obtained using said at least two optical techniques; (d) applying a
spectral pretreatment correction; and (e) applying a chemometric
algorithm.
12. The method of claim 11, wherein said spectral pretreatment
correction is baseline correction, standard normal variate
transformation, multiple scatter correction, wavelength selection,
smoothing, derivatisation, or any combination thereof.
13. The method of claim 11, wherein said chemometric algorithm is a
principal component analysis.
14. A method of monitoring changes in a clinical condition of a
subject, comprising the steps of: (a) measuring spectral changes in
bodily fluid or tissue of said subject using at least two optical
techniques; and (b) correlating the measured changes to a
corresponding change in clinical condition, wherein said spectral
changes are measured in ultra-violet wavelengths, visible
wavelengths, near-infrared wavelengths, mid-infrared wavelengths or
a combination thereof.
15. The method of claim 14, wherein said correlation step includes
deriving from the measured spectral properties an index or indices
correlating with the clinical condition of said subject.
16. The method of claim 14, further comprising the step of
comparing the measured spectral properties over time.
17. The method of claim 16, wherein said comparison step includes a
comparison to baseline spectral properties obtained at an initial
time point.
18. The method of claim 14, wherein the at least two optical
techniques include a combination of any two of the following
phenomena: diffuse reflection, inelastic (Raman) scattering,
absorption/transmission, or fluorescence.
19. The method of claim 14, wherein the bodily fluid is whole blood
or blood serum.
20. The method of claim 14, further comprising the step of
processing raw spectral data to obtain said measured spectral
properties.
21. The method of claim 20, wherein said processing step comprises
one or more of the following: (a) normalizing raw spectral data
relative to a spectral data obtained from a baseline sample; (b)
normalizing raw spectral data relative to total integrated power;
(c) combining chosen spectral bands and aggregating spectral data
obtained using said at least two optical techniques; (d) applying a
spectral pretreatment correction; and (e) applying a chemometric
algorithm.
22. The method of claim 21, wherein said spectral pretreatment
correction is baseline correction, standard normal variate
transformation, multiple scatter correction, wavelength selection,
smoothing, derivitisation, or any combination thereof.
23. The method of claim 21, wherein said chemometric algorithm is a
principal component analysis.
24. The method of claim 14, wherein the method is used to monitor
disease progression, onset, regulation or treatment in said subject
or to monitor changes in athletic conditioning or performance of
said subject.
25. A method for deriving an index or indices for correlation to an
observed clinical condition of a subject comprising the steps of
(a) obtaining a body of raw spectral data by measuring spectral
properties of bodily fluid or tissue of said subject using at least
two optical techniques; and (b) comparing the raw spectral data
with the clinical condition of said subject.
26. A method for overcoming confounding or interfering influences
on measured optical spectra by obtaining a body of raw spectral
data from measured spectral properties of a bodily fluid or tissue
of a subject using at least two optical techniques.
27. The method of claim 26, wherein said confounding or interfering
influences are oxygen saturation, hematocrit, hemoglobin, heparin,
pH, environmental factors, temperature, humidity, or a combination
thereof.
Description
CROSS REFERENCE TO RELATED APPLICATIONS
[0001] The present application claims priority to and benefit under
35 U.S.C. .sctn.119 and .sctn.365 of International Application
PCT/CA2007/001706, filed Sep. 21, 2007; which in turn, claims
priority to the U.S. Provisional Application Ser. No. 60/827,605,
filed on Sep. 29, 2006, the entire contents of International
Application PCT/CA2007/001706 and U.S. Provisional Patent
Application Ser. No. 60/827,605 are hereby expressly incorporated
herein by reference.
FIELD OF THE INVENTION
[0002] The present invention pertains to the field of investigating
clinical condition, or change in clinical condition and, more
particularly, to a correlation technique investigating clinical
condition of a subject using optical properties of bodily fluid or
tissue.
BACKGROUND
[0003] Determining the concentration of an analyte or a marker of
physical condition in a biological sample has been an important
technique in the field of diagnostics. Markers that have diagnostic
value include nutrients, metabolites, enzymes, immunity entities,
hormones, and pathogens. The physical characteristics of a
biological sample, such as temperature, optical properties,
density, and hardness, are also of interest because they can
provide indications with diagnostic value. Most determination
methods currently in use to detect markers and analytes and many
imaging methods employ signal-enhancing agents
[0004] Current standard blood analysis (laboratory assay) is
performed using blood samples obtained from a subject and assays
are generally based on the identification of measurable features of
the blood that are used to indicate the presence of a specific
known species within the blood. In some instances the measurable
features of the blood can be used to calculate the concentration of
the known species in the blood. The presence of the species, or its
concentration in the blood, is then used as an indicator or a
marker and correlated to a certain state of health within an
individual. Limitations of this approach of blood analysis include
difficulties limitations are the long time needed to perform the
various assays (in vivo and ex vivo) resulting in a historical
snapshot of blood species as an indicator for dynamic and possibly
rapid changing health states, and the reliance upon discrete, known
species as adequate markers for health states within an
individual.
[0005] Much interest has been expressed recently in developing
spectroscopic, in particular visible, infrared (IR) or
near-infrared (NIR) spectroscopic, techniques to non-invasively or
minimally invasively determine blood or tissue chemistry or to
analyze blood samples isolated from the patient. These non-invasive
techniques have the advantage of eliminating or greatly reducing
the need for collection of a blood sample or series of blood
samples from a patient, which, in turn avoids the discomfort and
complications that can be associated with blood collection. In
techniques developed to date, the spectroscopic measurements are
used to specifically identify or quantitate a particular marker or
analyte, or combination thereof. For example, U.S. patent
application Ser. No. 11/091,396 (Publication No. 2005/0222502)
discloses a respiratory monitoring apparatus that detects changes
in physiological parameters relevant to respiration using near
infrared spectroscopy. Similarly, U.S. patent application Ser. No.
11/125,107 (Publication No. 2005/0202567) discloses a spectroscopic
assay arrangement and technique for detection of the presence
and/or concentration of an analyte in a sample of bodily fluid.
[0006] A number of patents and patent applications disclose
spectroscopic methods and devices for non-invasive measurement of
blood or tissue analytes (See, e.g., U.S. patent application Ser.
No. 10/971,447 (Publication No. 2005/0107676), U.S. patent
application Ser. No. 10/943,737 (Publication No. 2005/0075546),
International PCT Application No. WO 01/016577, International PCT
Application No. WO 99/043255, International PCT Application No. WO
93/016629 and U.S. Pat. No. 6,928,311). In each case, the
techniques are used to specifically identify or quantitate a
specific analyte or characteristic.
[0007] There remains a need, therefore, for a reliable, convenient
method that permits measurement of the spectral properties of
bodily fluid or tissue as an indicator of clinical condition,
without the need to use the spectroscopic data to first identify
and quantitate a specific analyte or characteristic, which is, in
turn, used to extrapolate a clinical condition.
[0008] This background information is provided for the purpose of
making known information believed by the applicant to be of
possible relevance to the present invention. No admission is
necessarily intended, nor should be construed, that any of the
preceding information constitutes prior art against the present
invention.
SUMMARY OF THE INVENTION
[0009] An object of the present invention is to provide a
correlation technique for analysis of bodily fluid or tissue of an
individual. In accordance with an aspect of the present invention,
there is provided a method for investigating the clinical condition
of an individual, which method comprises the steps of: measuring
spectral properties of bodily fluid or tissue of said individual
using at least two optical techniques; and correlating the spectral
properties to a corresponding clinical condition, wherein said
spectral properties are measured in visible wavelengths,
near-infrared wavelengths or both. In accordance with a specific
embodiment of this aspect of the present invention, the method is
used to investigate an individual's disease state.
[0010] In accordance with another aspect of the present invention,
there is provided a method of monitoring changes in an individual's
clinical condition comprising the steps of: measuring spectral
changes in bodily fluid or tissue of said individual using at least
two optical techniques; and correlating the measured changes to a
corresponding change in clinical condition, wherein said spectral
changes are measured in visible wavelengths, near-infrared
wavelengths or both. In accordance with a specific embodiment of
this aspect of the present invention, the method is used to monitor
disease progression, onset, regulation or treatment in an
individual.
[0011] In accordance with another aspect of the present invention,
there is provided A method for deriving an index or indices for
correlation to an observed clinical condition of a subject
comprising the steps of: obtaining a body of raw spectral data by
measuring spectral properties of bodily fluid or tissue of said
subject using at least two optical techniques; and comparing the
raw spectral data with the clinical condition of said subject.
[0012] In accordance with another aspect of the present invention,
there is provided a method for overcoming confounding or
interfering influences, such as oxygen saturation, hematocrit,
hemoglobin, heparin, pH or environmental factors (e.g.,
temperature, humidity, etc.) on measured optical spectra by
obtaining a body of raw spectral data from measured spectral
properties of a bodily fluid or tissue of a subject using at least
two optical techniques.
BRIEF DESCRIPTION OF THE FIGURES
[0013] FIG. 1: Pre- and post-dialysis whole-blood spectra in linear
normalized units for (a) transmission and (b) diffuse reflection
modes. Blood spectra from individual patients are in gray, while
average pre- and post-dialysis values (8 patients) are indicated by
dashed and solid black lines, respectively. Shaded regions of the
abscissa indicate wavelength ranges with signal-to-noise ratio
<3 dB. Agreement between the published absorption spectrum of
oxyhemoglobin (S. Prahl, "Optical absorption of hemoglobin,"
http://omlc.ogi.edu/spectra/hemoglobin/.) and the spectrum in (b)
converted to logarithmic absorbance units is shown in (c).
[0014] FIG. 2: Whole blood difference spectra
(post-dialysis-pre-dialysis) for 8 patients (each line is for a
single patient) in (a) transmission and (b) diffuse reflection
modes. A horizontal line at zero-change is included for reference.
The discontinuities at 1015 nm (transmission) and 980 nm (diffuse
reflection) are due to concatenation of spectra obtained from two
different spectrometers, while shaded regions of the abscissa
indicate <3 dB signal-to-noise ratio.
[0015] FIG. 3: Partial correlation spectra for (a) WRR with the
effect of KRR removed, (b) URR with the effect of KRR removed and
(c) KRR with the effect of WRR removed. The spectra represent
partial correlation with the measured transmission or diffuse
reflection difference spectra from FIG. 2. Horizontal lines
indicate r.sub.crit for significance at the P=0.05 level, while
shaded regions indicated on the abscissa have been excluded from
the analysis due to low signal-to-noise ratio.
[0016] FIG. 4: Correlation spectra obtained from a hemodialysis
patient; A Transmission spectra and B Diffuse Reflection
spectra.
[0017] FIG. 5: A plot of principal components (eigenvectors) for 13
patients pre- and post-hemodialysis.
[0018] FIG. 6A depicts transmission spectra of typically low
oxygenated samples, mean subtracted and FIG. 6B depicts
transmission spectra of typically high oxygenated samples, mean
subtracted.
[0019] FIG. 7 depicts known oxyhemoglobin and deoxyhemoglobin
spectra.
[0020] FIG. 8 is a PCA score plot using two PC's from spectra
obtained from 13 pre- and post-hemodialysis patients.
[0021] FIG. 9 is a PCA score plot obtained by projecting
transmission spectra obtained from three hemodialysis patients on
the basis eigenvector space defined by results from the 13
hemodialysis patients (FIG. 8).
[0022] FIG. 10 depicts is a PCA score plot obtained by projecting
transmission spectra obtained from three hemodialysis patients over
time on the basis eigenvector space defined by results from the 13
hemodialysis patients (FIG. 8).
DETAILED DESCRIPTION OF THE INVENTION
[0023] The present application provides a method for measuring
spectral properties of bodily fluid and/or tissue of a subject and
using the spectral information to assess the state of health of the
subject. The spectral properties are correlated to the patient's
condition. In a specific example of the present invention, the
subject's condition is a disease state, for example, from a
well-understood clinical or medical condition.
[0024] The term "bodily fluid" as used herein, refers to any fluid
of the body, including, but not limited to, sputum, saliva, whole
blood, vitreous fluid, plasma, peritoneal fluid, cerebrospinal
fluid and so forth.
[0025] The term "clinical condition" as used herein, refers to any
condition of a human or animal that is effected by one or more an
internal or external effector. Examples of internal effectors
include, but are not limited to, genetic traits, congenital
abnormalities and so forth. Examples of external effectors include,
but are not limited to, pharmaceutical ingestion, food intake,
exercise, stress, infection and so forth.
[0026] Since the method of the present invention involves the use
of optical methods for detecting properties of the patient's bodily
fluid and/or tissue, the method is amenable to non-invasive or
non-contact uses. In addition, the method of the present invention
allows for real-time, continuous measurements; in contrast to
standard assay-based methods of blood analysis.
Spectroscopic Measurements
[0027] The measurements obtained using the method of the present
invention are relative and are not chemical-specific. Therefore,
difficulties associated with absolute calibration of absorption and
scatter, corrections for subject variability, and developing a
spectral library associated with known chemical entities are
largely circumvented. A further advantage of this approach is its
suitability for the detection of signatures from complex
treatment-initiated biochemical events in vivo, which involve
either too many molecules to isolate and quantify (even for
invasive techniques), involve unknown molecules and interactions,
or are rendered undetectable outside of a living body.
[0028] In general, the method of the present invention comprises
the steps of illuminating bodily fluid and/or tissue of a patient
(in vivo or ex vivo) with light and obtaining at least two optical
measurements based on two distinct light-fluid or light-tissue
interaction phenomena. For example, the method of the present
invention can make use of a combination of any two of the following
phenomena: diffuse reflection, inelastic (Raman) scattering,
absorption/transmission, or fluorescence.
[0029] In accordance with a specific embodiment of the present
invention, the illumination light used in the method of the present
invention is in the visible and/or near infrared spectrum of
radiation (i.e., in the range of 400-2500 nm). Alternatively, the
illumination light can include light in the UV and mid-IR regions,
such that the full illumination spectrum range can span 300 nm to
30 microns A specific example of an optical measurement of
light-tissue interaction is the measurement of tissue
autofluorescence, for example, as described by Meerwaldt, et al.
(2005) J. Amer. Soc. Nephrology 16:3687-3693. The measured optical
properties are then correlated to a known clinical condition or a
change in clinical condition by comparison to optical properties
known to be associated with certain conditions or by comparison to
optical properties measured over time, respectively.
[0030] By making use of two distinct light-fluid or light-tissue
interaction phenomena, the ability to correlate bodily fluid or
tissue optical changes to a clinical condition or a change in
clinical condition is improved in comparison to the use of a single
light-fluid or light-tissue interaction phenomena. In particular,
there is an increased robustness and sensitivity associated with
the use of at least two light-fluid or light-tissue interaction
phenomena in comparison to the use of one such phenomena.
[0031] Spectral information from bodily fluid and tissue can be
obtained using various spectrometric techniques and, as would be
appreciated by a worker skilled in the art, such techniques can be
used in the method of the present invention. The goal is to obtain
wavelength dependent measures of the interaction of light with the
bodily fluid or tissue of interest. For example, a broad spectrum
source (all wavelengths in the source) can be used to illuminate
the bodily fluid or tissue of interest. A dispersive element, (for
example a diffraction grating or a prism) can be used to separate
the different wavelengths of light returning from the sample and
collect the intensity at each wavelength on a detector, digital or
analog. Alternatively, sources of different wavelengths can be used
to illuminate the bodily fluid or tissue of interest, followed by
measurement of the intensity of light returning after the
light/tissue interaction has occurred. Possible detector types
could be but are not limited to, Photomultiplier tubes, diode
arrays, Charge Coupled Devices, cmos detectors photographic film
and any type of photodiode including avalanche photodiodes. Example
of light sources include, but are not limited to, arc, incandescent
or fluorescent lamps, optionally in combination with one or more
optical filters, any of the many kinds of light emitting diodes,
any of the many kinds of lasers, and natural light.
[0032] In vivo work requires that the possibility of light damage
to the tissue during measurements be minimized or avoided
altogether. Thus, in practicing present invention in vivo the light
intensities employed should be below the threshold for tissue
damage as per published standards. It is then possible that the
intensity of light returning from the tissue will be low. To
overcome these problems certain embodiments of the method of the
present invention include an additional step or steps to improve
signal to noise ratio. Suitable techniques for use in the method of
the present invention include, but are not limited to, lock-in
detection strategies as well as, analog and digital filtering
techniques. In lock-in detection strategies, the illumination
source intensity is modulated using a well controlled pattern. The
detection system is simultaneously synchronized to the modulation
pattern. The result is that only signal modulated with the same
pattern as the source is optimally detected, other signals (i.e.,
the noise) are not efficiently detected and, thus, the noise signal
is reduced, improving the signal quality. Filtering techniques,
whether analog or digital, can be used to selectively remove
signals that are assumed (based on expected responses of the
system) to contribute mostly to noise. An analog implementation
would filter the signal using an electronic circuit and a digital
technique using post acquisition algorithms.
[0033] In accordance with certain embodiments of the present
invention the method includes an enhancement technique to maximize
the signal-to-noise ratio. For example, positive correlation
filters at the light source output can be used to provide optimum
illumination of only those wavelengths contributing significantly
to the spectral variables of interest, thereby allowing increases
in illumination power at these wavelengths, further increasing
sensitivity (U.S. Pat. No. 5,747,806). The demonstration of
chemical-specific detection through the retinal blood (U.S. Patent
Application No. US2002/0072658)--a more demanding
approach--provides further evidence that the method of the present
invention has sufficient sensitivity to detect small changes in the
presence of confounders in the eye and in the blood.
[0034] A broad range of probe wavelengths as separate or combined
probes for both absorbed and scattered light are available for use
in the method of the present invention. This permits very sensitive
detection of evidence of effects in the bodily fluid and/or tissue
using the method of the present invention. Since the method of this
invention is not substance-specific, many of the specificity issues
encountered with previous methodologies are minimized or
avoided.
[0035] The poor reproducibility in the measurements obtained using
previous methods can be readily minimized or avoided using the
method of the present invention. Spectral fluid and tissue
fluctuations can be large and random even during baseline
measurements or in measuring a control group. In such instances,
however, lower wavelength resolution can be implemented through
binning procedures, or a single measurement of absorption or
scatter alone may be used to reduce variability. Inter-treatment
and inter-patient variability is not a concern when using the
method of the present invention to monitor disease progression,
regulation or treatment, due to the relative nature of the
measurement approach. Furthermore, patient-specific measurements
and algorithms can be employed in the present method, wherein
certain wavelength regions of prognostic or diagnostic value may
only apply to a specific individual or small group of
individuals.
[0036] In accordance with one embodiment of the invention the
measured spectral properties are compared to the spectral
properties of a known population as an indicator or measure of a
similarity or difference in clinical condition of the subject under
study in comparison to the average clinical condition of the known
population. This technique is useful for diagnostic applications as
well as for research applications.
[0037] In accordance with one embodiment of the invention changes
in the spectral properties are monitored over time, relative to
baseline measurements, as an indicator of a change in clinical
condition over time and, for example, with the addition, removal or
change in one or more internal or external effectors. With
appropriate correction of baseline shifts and drift, spectral
changes over time can be used, for example, as indicators of
treatment and disease regulation, as opposed to the hitherto
isolative approach of chemical-specific detection. This technique
is useful for diagnostic applications as well as for research
applications. In research, the emphasis is elucidation of blood
properties and/or changes related to treatment-physiology
interactions in a well-known clinical situation, rather than
diagnostic capability.
[0038] In accordance with this embodiment of the present invention,
the steps of illumination and optical property measurement are
repeated at discrete time intervals in order to monitor changes in
clinical condition over time, for example, when monitoring changes
in disease state in response to therapy. The selection of the time
intervals for testing is well within the abilities of a worker
skilled in the art and is made based on the specific application of
the method, taking into consideration, for example, the disease or
condition afflicting the patient, the type of treatment, the length
of treatment or the uptake and/or metabolism of a pharmaceutical
used in the treatment.
[0039] In an alternative of this embodiment of the present
invention, changes in optical properties are measured by continuous
real-time monitoring of the optical properties of a subject's
bodily fluid and/or tissue.
[0040] Irrespective of whether the spectral changes over time are
obtained using discrete time samples or continuous real-time
monitoring, the spectral changes are subsequently correlated to a
change in the clinical condition of the subject, for example, a
change in disease state.
Data Processing
[0041] After the raw optical spectra are obtained, they are
processed in order to separate and emphasize features within the
spectra correlated to a clinical condition, while minimizing those
features arising from instrumental artifacts and undesired physical
effects (sources of noise). As indicated above, multiple techniques
(i.e., absorption, scatter, fluorescence) can be used according to
the present invention to obtain the initial spectral information.
The optical parameters that correlate to a particular clinical
condition or change in clinical condition can be selected from the
entire data set of spectral information from the multiple
techniques (following the application of techniques such as
multivariate analysis), using well established or new methodologies
specific to the application.
[0042] In accordance with one embodiment of the present invention,
the analysis used to process the raw optical absorption and scatter
spectra combines the principles of spectral nephelometry (Mignani,
A. (2003) "Spectral Nephelometry For Making Extravirgin Olive Oil
Fingerprints" Sensors and Actuators 90: 157-162) with the methods
of chemometric analysis used in NIR absorption spectroscopy
(Caspers, P. (2002) "Verification of the identity of pharmaceutical
substances with near-infrared spectroscopy" Bilthoven, The
Netherlands, National Institute of Public Health and the
Environment). This approach comprises the following features:
[0043] normalization of spectra relative to baseline sample (bodily
fluid/tissue or pre-treatment measurement); [0044] normalization
relative to total integrated power (where spectral shape is desired
instead of absolute intensity information); [0045] combination of
chosen spectral bands and possible aggregation of absorption and
scatter spectra; [0046] application of spectral pretreatments:
baseline correction, standard normal variate transformation,
multiple scatter correction, wavelength selection, smoothing,
derivatives, etc. (Caspers, P. (2002) "Verification of the identity
of pharmaceutical substances with near-infrared spectroscopy"
Bilthoven, The Netherlands, National Institute of Public Health and
the Environment); and [0047] application of a chemometric
algorithm.
[0048] Advantageously, based upon the applicants' experience and
good results obtained in the literature, the chemometric technique
of principal component analysis (PCA) (Cowe, I. (1985) "The Use of
Principal Components in the Analysis of near-infrared spectra,"
Applied Spectroscopy 39(2): 257-266) can be used as the chemometric
algorithm. PCA is a widely used multivariate analysis technique
that enables the expression and visualization of complex spectra in
terms of the independent elements responsible for variation within
the spectra. Many other techniques such as, but not limited to,
Linear Discriminant Analysis or non-linear models of spectral
compositions can be used to summarize the data into a small number
of clinically relevant indices/parameters.
[0049] In accordance with one embodiment of the present invention,
the method is used in a clinical setting. In one example of such a
clinical implementation, the values of the indices/parameters
extracted from the spectral analysis can be compared to benchmark
values for these indices. The benchmark values can be from an
earlier time point for the same patient (to quantify changes in the
patient's health) or based on data obtained from a human population
(to identify patient(s) that could have a particular clinical
condition, for example, a disease state).
Applications of the Method
[0050] The method of the present invention has broad application to
any situation in which it is desirable to identify, or monitor
changes in, the clinical condition of a subject. In general, the
method takes advantage of the fact that the optical properties of
bodily fluid and/or tissue changes over time or due to the
presence, absence or change of an external or internal effector
(e.g., disease, drug ingestion, infection, change in health status,
etc.). For example, measurement of the optical spectrum of bodily
fluid and/or tissue and comparing to spectra of bodily fluid and/or
tissue of individuals known to be either affected or unaffected by
the external factor allows correlation to a particular clinical
condition affected by the presence or absence of the external
factor. As a further example, measurement of the optical spectrum
of bodily fluid and/or tissue of a subject at various points in
time allows the determination of changes in spectral response from
the bodily fluid and/or tissue that are highly correlated to a
change in clinical condition. The spectra obtained from the subject
under study can be compared to standard spectra obtained using a
standard reference method, or can be analyzed based on medical
opinion or subjective assessment in situations in which no standard
is available. In the latter case, the observed change in the
optical properties of the bodily fluid and/or tissue can be
correlated with a change in the observed change in, for example,
symptoms of the subject.
[0051] The external factor may or may not be an identified chemical
species. Where no species is known or where numerous species may be
involved in a particular clinical condition, the correlation
technique of the present invention is particularly useful as it
correlates bodily fluid and/or tissue changes directly with the
external or internal factor, bypassing the isolation/identification
of candidate species.
[0052] In accordance with certain embodiments of the present
invention, the method is used either as an alternative or a
supplement to standard clinical and laboratory analyses.
[0053] In accordance with particular embodiments of the present
invention, the method is used as part of a routine health
assessment of an individual. For example, the present method can be
used to verify the absence/presence of a clinical condition or
monitor changes in a clinical condition, such as a disease, during
the course of treatment (physical, pharmaceutical or other). In
embodiments, the present method can be used to monitor the course
of treatment of clinical conditions or diseases including but not
limited to: diabetes, cancer, heart disease, and end-stage renal
disease. The method of the present invention is also amenable to
medical testing such as may be employed during surgery (such as
online monitoring during cardiopulmonary bypass), as part of
at-home monitoring, during therapeutic treatment (such as online
monitoring during renal dialysis, physiotherapy, chemotherapy or
radiation therapy), or hospital bedside monitoring or other
point-of-care monitoring.
[0054] It should be readily understood that the method of the
present invention can be used to correlate optical properties with
a clinical condition, such as disease, as the endpoint of the
method, and, to correlate optical properties to clinical outcomes,
prediction of outcome, or prediction of response to treatment (in
the case of therapeutic applications). Alternatively, the method of
the present invention can be used to investigate spectral changes
to eventually isolate bodily fluid or tissue factors to develop new
biomarkers and/or interventions.
[0055] It should be readily appreciated that the method of the
present invention is not limited to medical applications. Rather,
the method can be used to monitor or identify any perturbation of
an individual's condition, such as in response to a particular
stimulus. By way of example, the method can be used for monitoring
athletic conditioning, dieting, response to stress, etc.
[0056] The method of the present invention can also be used to
monitor individuals at risk, or at high risk, for developing
certain clinical conditions such as diabetes, cancer, and heart
disease. By monitoring changes in the optical properties of an
individual's bodily fluid and/or tissue it can be possible to
facilitate early detection of the onset of disease, which, in turn,
will permit early treatment or prevention. Similarly, since the
method of the present invention is sensitive to molecular and
biochemical changes in an individual, it can be used as a research
or diagnostic tool to identify changes in the optical properties of
bodily fluid and/or tissue that can then be used as an early step
in the search for the root cause of observed changes in the
clinical condition of the individual.
[0057] In accordance with a specific embodiment of the present
invention, the method is used to monitor known bodily fluid (e.g.,
blood) components, or total bodily fluid and/or tissue changes in
response to a stimulus (irrespective of whether or not there is any
knowledge of the relevant components).
[0058] To gain a better understanding of the invention described
herein, the following examples are set forth. It should be
understood that these examples are for illustrative purposes only.
Therefore, they should not limit the scope of this invention in any
way.
Examples
Example 1
Hemodialysis Monitoring In Whole Blood Using Transmission and
Diffuse Reflection Spectroscopy
1. Introduction
[0059] Hemodialysis is a medical treatment that involves diffusive
and convective removal of solutes and water from the blood of
patients with end-stage renal disease (ESRD), whose kidneys can no
longer perform this task. Current standard measures of treatment
adequacy and dose of hemodialysis are based upon the clearance of
the low molecular weight compound urea from the blood, determined
from pre- and post-dialysis blood samples analyzed in a clinical
laboratory. While measurement of urea clearance is the most widely
used method to assess dialysis adequacy in ESRD, urea is only one
of many metabolites that accumulate in ESRD and it may represent a
surrogate marker rather than a principal toxin.sup.1-3. While the
hemodialysis procedure filters the blood of low-molecular weight
water-soluble molecules, a host of potentially toxic middle- and
high-molecular weight molecules remain unfiltered. These include
protein-bound molecules that may contribute to the development of
complications in ESRD patients, such as the uremic syndrome and
vascular disease.sup.3. Investigation of uremic toxins is clearly
of critical importance in developing treatment strategies that
improve patient quality-of-life and longevity.
[0060] One means of achieving this goal is to identify, classify
and characterize the clinical importance of as many candidate toxic
molecules as possible, as is the mandate of the European Uremic
Toxin Work Group (EUTox) initiated in 2000.sup.4. As a consequence,
however, of the complexity and the limited understanding of the
chemistry of kidney disease and its treatment, it has proven
difficult to find individual analytes (isolated from the rest of
the blood chemistry) that can accurately and reliably describe the
disease state or report the efficacy of treatment. An alternative
approach is to monitor whole blood as a complex structure and
correlate any changes in this structure to the hemodialysis
treatment and the patient's clinical status or condition. The
emergence of consistent patterns of change or absence of change in
blood properties could lead to new candidate toxicity indicators
and to the subsequent investigation of factors underlying the
observed patterns. An advantage of this latter approach is the
simultaneous inclusion of numerous molecules including their
interactions and indirect effects as they contribute to the
observed blood properties. In this regard, optical spectroscopy can
be an effective tool to probe the complex response of blood to
treatment in the clinic.
[0061] Several approaches using optical spectroscopy to monitor
hemodialysis treatment have been reported. These studies utilize
light as a non-contact tool for reagentless determination of urea
and other solute concentrations in spent dialysate fluid.sup.2,5-7.
Although such approaches are useful for online monitoring of
filtered analytes, the direct optical detection of analytes
retained in blood (including unfiltered compounds) has not been
reported. While whole blood optical monitoring during hemodialysis
has been achieved, the reported methods have used a small number of
discrete wavelengths to monitor specific blood parameters such as
hematocrit, blood volume, oxygen saturation and hemoglobin
levels.sup.8-10. These parameters, however, are not indicative of
potential toxins within the blood nor do they provide a means to
assess the efficacy of treatments.
[0062] The present study demonstrates that features in the optical
spectrum of undialyzed versus dialyzed whole blood showed a
significant difference as a result of the hemodialysis treatment.
Additionally, the detected changes in the spectrum of whole blood
were found to be consistent with accepted clinical outcomes, as
determined by comparing the spectroscopic results to
clinically-measured analyte changes (as a gold standard) following
dialysis treatment. While the optical monitoring techniques used
are readily adaptable for online monitoring, the whole-blood
approach can enable be used to identify surrogate markers for
toxicity or for patient prognosis through established disease
pattern recognition techniques.sup.11-14.
2. Materials and Methods
2.1. Clinical Design
[0063] A sample population of eight ESRD patients undergoing
regular hemodialysis treatment (four-hour sessions, three times a
week) was recruited on a volunteer basis for the pilot study.
[0064] Volunteers represented a broad cross-section of ESRD
patients in terms of age and gender (6 male, 2 female; mean age
61.5 years; age range 39-75 yrs), time since initiation of dialysis
(18 mos. to 284 mos.) and the presence of other systemic conditions
(hypertension--3 patients, Type-II diabetes--3 patients). Blood
extraction from subjects occurred immediately before and after a
single hemodialysis treatment (<1 min). The day of blood
extraction coincided with the monthly laboratory blood testing day
for each volunteer, thereby allowing subsequent correlation of
spectral data to clinical laboratory results.
2.2. Blood Sample Preparation
[0065] Samples of whole blood for the study were obtained at the
same time standard clinical blood samples were obtained. Blood was
drawn into standard 3 ml purple-top collection tubes containing 5.4
mg K.sub.2-EDTA as an anticoagulant. The collection tubes were
manually agitated to provide a homogeneous suspension and 1 ml from
each tube was transferred to an optical cell and sealed. The delay
between sample collection and the start of optical measurements
averaged one hour, with a maximum delay of two hours. To test the
influence of the delay, optical spectra from a single blood sample
were taken hourly over a four hour period at room temperature, with
no significant change observed in the spectra (data not shown).
[0066] Serum urea and potassium levels were quantified using an
automated Beckman Coulter LX20 analyzer. Manufacturer-supplied
reagents were used and an indirect ion selective electrode method
was used to quantify potassium while a coupled enzymatic rate
method was used for urea. The intra-assay variability of these
techniques is nominally accepted to be approximately 2%
(coefficient of variation).
2.3. Measurement of Transmission and Diffuse Reflection Spectra
[0067] Optical cells with a 2 mm optical path length were used,
made from optical glass with >80% transmission over the
wavelength range 365 nm-2500 nm (Varsal Inc.). For measurements,
the optical cells were placed in a custom sample holder ensuring
both repeatable cell placement and minimal optical-mechanical
interference to avoid stray light and spurious reflections. A
focused spot (4 mm diameter) from a current-stabilized 20 W
tungsten-halogen light source (model ASB-W-020, Spectral Products,
Inc.) was used to illuminate the optical cell. Light transmitted
through the cell was focused into a collection optical fiber (400
.mu.m core diameter low-OH fiber, Ocean Optics Inc.) connected to a
spectrometer. Light backscattered in a cone over a
10.degree.-30.degree. angle relative to the incident beam was
captured by a wide-aperture achromatic lens and focused to a second
collection optical fiber (600 .mu.m core diameter low-OH fiber,
Ocean Optics Inc.) connected to a spectrometer. The off-axis
geometry used for backscatter collection minimized the interference
of both specular reflection and edge effects from the optical
cell.
[0068] Optical spectra were acquired over the 400 nm-1700 nm region
using two spectrometers spanning wavelength ranges of 400 nm-1000
nm (model SD2000, Ocean Optics Inc., 2048-element silicon
photodiode array; spectral resolution 0.33 nm) and 900 nm-1700 nm
(model InGaAs512, StellarNet Inc., 512-element InGaAs photodiode
array; spectral resolution 2.25 nm). Spectra were acquired through
computer control with acquisition times of 8 ms and 800 ms for
transmission and diffuse reflection, respectively, for the near
infrared spectrometer, and 10 ms for both modes using the
visible/near infrared spectrometer.
[0069] Blood samples were maintained at room temperature and sample
heating was minimized by using a mechanical shutter kept open only
during the spectral acquisition period. Three data sets were
obtained for each sample, where for each set the cell was removed,
agitated and replaced. The three spectra were subsequently
area-normalized and then averaged to account for sources of
variation resulting from optical cell placement and variations in
light intensity level. Prior to averaging, the maximum coefficient
of variation among any set of three normalized spectra was 2% and
9% for wavelengths below and above 1000 nm, respectively. All
subsequent analyses reported here were performed using normalized,
averaged spectra. Influences of the spectral properties of the
light source, the optical cell, and optical elements in the light
delivery and detection paths were removed by dividing the
normalized transmission and diffuse reflection spectra by a
reference measurement taken with an empty optical cell
(transmission path) and a broadband mirror placed behind an empty
optical cell (diffuse reflection path).
[0070] About 1 W of focused optical power was delivered to the
blood sample. Typically about 15% of the incident light was
transmitted through a sample while about 5% was diffusely reflected
in the direction of the detection cone. Although the transmitted
and diffusely reflected light levels were high, signal-to-noise
ratio was reduced due to manual attenuation of the delivered and/or
detected light streams which was necessary to accommodate both a
limited photodetector dynamic range and a requirement for the
simultaneous measurement of transmitted and diffusely reflected
paths with differing light levels. Wavelength regions where optical
signal-to-noise levels failed to exceed an imposed 3 dB minimum
threshold were identified and excluded from subsequent analyses. Of
particular note, strong water absorption in the 1400-1550 nm band
resulted in a high dynamic range detection requirement and
therefore a reduced signal-to-noise ratio.
3. Data Analysis and Interpretation
3.1. Whole Blood Spectra
[0071] Thirty-two separate spectra were obtained; these
corresponded to pre- and post-dialysis blood for 8 patients in both
transmission and diffuse reflection modes, FIGS. 1(a) and (b). The
spectra differed depending upon the light interaction mode, which
is consistent with results reported elsewhere.sup.15. In FIG. 1(c)
the data in FIG. 1(b) was plotted as an absorbance along with
published absorption data for pure oxyhemoglobin.sup.16. Good
agreement was observed between the measured and published results,
and in particular the characteristic double-peaked absorption of
oxyhemoglobin was visible, with peaks at 542 nm and 575 nm.
Hemoglobin (principally its oxygenated and deoxygenated forms)
dominates whole blood absorption of wavelengths shorter than 1000
nm, whereas water generally dominates absorption above 1000 nm. The
well-known broad absorption peaks of water centered at 970 nm and
1440 nm and a minor peak at 1200 nm were also visible in the
measured spectra in FIG. 1(c).sup.17.
[0072] Other than area-normalization, averaging, and referencing,
the spectra were not filtered, smoothed, derivatized or pre-treated
prior to analysis. In this manner the detailed fluctuations
characteristic of turbid media were preserved and permitted the
interpretation of raw spectral changes in whole blood independent
of data processing methods.
3.2. Intra-Group Comparison
[0073] To assess the significance of spectral changes observed as a
result of hemodialysis, quantitative analysis of full-spectrum
change was performed using the principal component analysis (PCA)
method.sup.18. Briefly, in the PCA method a group of input spectra
were mathematically decomposed into a small set of uncorrelated,
orthonormal variables (the principal components) which account for
the major sources of variation across the group of spectra.
Moreover, for each individual spectrum it was possible to derive a
set of principal component `scores` or `weights` representing the
contribution of each principal component to the linear
decomposition of each spectrum in terms of the principal
components. By limiting the analysis to only the most significant
sources of variation among the spectra, an entire optical spectrum
can be represented by one or a few variables.
[0074] PCA was performed separately for transmission and diffuse
reflection using the two groups of 16 whole blood spectra shown in
FIG. 1. Eigenvalues corresponding to the first principal component
had a value greater than 1, and therefore only the first principal
component was retained as an indicator of the most significant
source of variation among the 16 optical spectra. The first
principal component accounted for 94% and 63% of the variation in
the transmission and diffuse reflection spectra, respectively.
Scores for the first principal component for the spectrum of each
patient are given in Table 1. Mean values for each group (pre- or
post-dialysis) are shown along with the corresponding results of a
paired t-test for the null hypothesis (no spectral difference due
to treatment). For both light interaction modes the null hypothesis
could be rejected, indicating significant whole-blood spectral
changes in transmission (P<0.003) and diffuse reflection
(P<0.001) due to hemodialysis treatment.
TABLE-US-00001 TABLE 1 Principal component scores used to test the
null hypothesis of no difference between pre- and post-dialysis
blood based on full-spectrum analysis. First Principal Component
Scores Transmission Diffuse reflection Patient Pre Post Pre Post 1
-13.19 -13.82 9.54 10.96 2 -10.68 -11.78 7.72 9.68 3 -14.51 -16.03
9.84 11.16 4 -11.04 -13.20 9.23 11.08 5 -11.35 -11.73 7.60 8.36 6
-12.48 -14.84 9.74 11.77 7 -14.54 -15.05 10.35 10.75 8 -14.21
-15.41 10.16 11.30 Mean -12.75 -13.98 9.27 10.63 paired t 4.69
-6.59 P 0.0022 0.0003
[0075] An optical difference spectrum
(post-treatment-pre-treatment) was derived from the data shown in
FIG. 1 for each patient and measurement mode and is given in FIG.
2. In the difference spectra coinciding local extrema were observed
as well as isobestic wavelengths of near-zero change for all
patients, which differ in location depending on the light
interaction mode. Interestingly (and not obvious from the
monochrome plot), was the difference observed between patients in
both the magnitude of change across the spectrum and among the two
modes. For example, for first three local extrema in transmission
(FIG. 2a), Patient 6 exhibited the largest change while Patient 4
exhibited the largest change for the following two extrema. In
diffuse reflection (FIG. 2b) however, Patient 2 showed the largest
change at the first minimum while Patient 3 showed the largest
change at the following peak and Patient 4 showed the largest
change at the next minimum.
[0076] Possible origins of the complex spectral changes observed
include an alteration in the concentration, binding or other
molecular properties of hemoglobin or changes in the environment
surrounding hemoglobin, as well as possible changes in water
constitution including electrolyte levels, acidity, and
intra/extra-cellular fluid balance and hydration.sup.19,20.
[0077] Periodic fluctuations observed in the diffuse reflection
change at longer wavelengths are due to multiple reflections caused
by the blood, optical cell, and air interfaces. Also, transmission
spectra exhibit greater signal-to-noise ratio than the diffuse
reflection spectra due to the higher absolute optical power levels
detected in transmission. As described earlier, only a portion of
the diffusely reflected light (20.degree. solid angle) was captured
in the present setup resulting in a lower absolute intensity level.
It has been reported that increasing the solid angle of collection
of diffusely reflected light using an integrating sphere
significantly improves the quality of whole blood
spectra.sup.21.
3.3. Correlation with Clinical Measures
[0078] To confirm that the observed whole blood spectral changes
were consistent with clinical parameters used in routine patient
care, changes in clinically-based measures of hemodialysis were
correlated with the difference spectra. For each patient, clinical
charts were used to compile key measures of hemodialysis
performance to allow a direct comparison of optical measurements
and clinical outcomes. The analysis presented here focuses on a few
key clinical measures of hemodialysis treatment: the urea reduction
ratio (URR=1-post/pre blood urea concentration); a derived measure,
the potassium reduction ratio (KRR=1-post/pre blood potassium ion
concentration); a derived measure relating to fluid removal by
filtration, the weight retention ratio (WRR=post/pre body weight);
and Kt/V (dialysis dose, where K is urea clearance rate of the
dialyzer in L/min, t is treatment time in min and V is the urea
distribution volume for the patient in L).
[0079] The clinical values for the chosen measures along with the
range of values among the patient group are shown in Table 2. The
value of Pearson's correlation coefficient r among the clinical
parameters has also been given in Table 2. A nearly perfect
positive correlation was found between URR and Kt/V which was
expected given that Kt/V values are derived from the URR of a
patient.sup.22. URR thus suffices to describe the behavior of Kt/V
in the following analysis. Furthermore, while absolute potassium
reduction is used as an accepted clinical measure, KRR was derived
for the purposes of this study to provide a relative, unit-less
parameter for consistency in the analysis. Absolute potassium
reduction and the KRR were highly correlated (r>0.97) indicating
nearly identical behavior.
TABLE-US-00002 TABLE 2 Hemodialysis parameter values for the
patient group obtained from clinical blood laboratory results and
patient charts. Minimum and maximum parameter values within the
group are also given. Correlations among the parameter values are
also shown. Hemodialysis Measure Patient WRR URR KRR Kt/V 1 0.975
0.755 0.261 1.80 2 0.945 0.768 0.318 1.88 3 0.970 0.765 0.220 1.88
4 0.958 0.855 0.453 2.24 5 0.960 0.775 0.396 1.88 6 0.946 0.830
0.395 2.12 7 0.979 0.791 0.217 1.96 8 0.953 0.728 0.415 1.72 Weight
Urea Potassium Min 62 kg 2.7 mM 2.6 mM value Max 100 kg 27.2 mM 5.3
mM value Parameter Correlation WRR URR KRR Kt/V WRR -- -0.171
-0.704 -0.190 URR -- 0.369 0.997 KRR -- 0.370 Kt/V --
[0080] Additionally in Table 2 correlations were observed between
WRR and KRR as well as URR and KRR. To remove the effect of
intervening parameters, the partial correlation coefficient
r.sub.xy,z has been used to represent the correlation of x (the set
of optical difference values at a given wavelength data point in
FIG. 2) with y (the set of clinical parameter values), while
removing the influence of z (a correlated clinical parameter). The
result is a partial correlation spectrum for each clinical
parameter in both transmission and diffuse reflection modes with
the influence of the clinical parameter with the highest
correlation with the chosen parameter partialed out. Using the
two-tailed t-test for significance of the partial correlation
statistic with N=8 patients yields critical values of |r|=0.754 and
0.874 to be significant at the P=0.05 and 0.01 levels,
respectively. The partial correlation spectra for the WRR, URR, and
KRR are given in FIG. 3. Correlation spectra from a single optical
interaction mode (transmission or diffuse reflection) have been
shown for each parameter as the other mode did not exhibit regions
of correlation exceeding the P=0.05 critical value. Analysis of the
correlation spectra has been restricted to wavelength regions with
adequate signal-to-noise ratio and where the correlation
coefficient exceeded the critical value over a broad wavelength
range (>5 nm). Sharp, noise-like spikes in the correlation
coefficient have thereby been excluded from the analysis.
Wavelength regions of significant correlation with clinical
parameters are summarized in Table 3.
TABLE-US-00003 TABLE 3 Wavelength regions of clinical parameter
correlation r exceeding the critical value r.sub.crit for
significance at the P = 0.05 level over a wavelength band >5 nm
as determined from FIG. 3. Within each region the peak correlation
coefficient value r.sub.max is given along with the corresponding
two-tailed significance level (P-value) for |r| > |r.sub.max|.
Wavelength Significance Clinical range (nm) Peak correlation
(P-value) Parameter |r| > r.sub.crit coefficient r.sub.max |r|
> |r.sub.max| WRR 595-605 0.869 0.011 702-710 -0.824 0.023
716-735 -0.931 0.002 1085-1094 0.972 0.0002 URR 593-599 -0.851
0.015 740-748 0.854 0.014 KRR 588-602 0.933 0.002 702-764 -0.984
<0.0001
[0081] Several wavelength regions of diffuse reflection change
(centered at 600 nm, 720 nm, and 1090 nm) were highly correlated to
the WRR. Correlations below 1000 nm are likely to represent a
modification in the scattering environment of the hemoglobin
molecule or red blood cells (RBCs) caused by either the direct
removal of fluid (source of the weight change) or the removal of
analytes along with the fluid that may influence hemoglobin. The
strongest observed correlation was above 1000 nm where water
absorption dominates the spectrum, possibly indicating the relation
between fluid removal through hemodialysis and a change in water
content of the blood.
[0082] Regions of significant URR correlation with optical
transmission change were observed centered at 595 nm and 745 nm. It
has been suggested that the correlation with optical transmission
in the 595 nm region is related to urea-mediated blood cell volume
changes.sup.19. In addition, second- and third-order overtones of
the N-H symmetric and asymmetric stretching vibrations in the urea
molecule are broad bands centered at 960-1000 nm and 720-750 nm,
respectively.sup.6. First order overtones of these stretches occur
in the 1450-1500 nm region, but these are buried under the strong
water absorption band in this region.sup.6. The observed
correlation around 745 nm corresponds well to the third-order N-H
overtone while high correlations in the second-order overtone
region were also observed (as seen in FIG. 3) but did not exceed
the critical value for significance over a 5 nm window and have
therefore been excluded from Table 3.
[0083] Significant correlation for the KRR was observed in two
broad regions. Accordingly, potassium ions in whole blood have been
shown to exhibit a broad correlation in the 500-1000 nm wavelength
region.sup.20. The observed regions of high correlation may
correspond to the ion concentration itself or to the indirect
effects of its removal. The latter effect is plausible as the
potassium ion balance between intracellular and extra-cellular
fluid can influence blood cell volume thereby affecting the
scattering of light.
4. Discussion
[0084] To our knowledge, the present study documents for the first
time the change in the optical transmission and diffuse reflection
spectrum of whole blood resulting from a hemodialysis treatment
session. Using visible and near infrared wavelengths, a
statistically significant whole-spectrum change due to treatment
was found for both light interaction modes. The clear distinction
of pre- and post-dialysis blood in this manner can serve as the
basis for a useful approach to on-line hemodialysis monitoring.
When calibrated with a larger spectral database from multiple
treatments, an online measure such as the principal component score
could be used to monitor the progress of a treatment session. As
the score reflects multiple blood parameters including unfiltered
analytes, it can provide a means to determine the adequacy of
treatment and can be a candidate for a comprehensive indicator of
longer-term clinical patient outcomes.
[0085] Although the patients recruited in this study represented a
cross-section of age, gender, and clinical condition, the observed
spectral shape changes due to treatment were consistent among
patients, though the magnitude of change differed
substantially.
[0086] In this study a few key clinical measures of uremic toxicity
and hemodialysis adequacy were chosen and shown to correlate with
whole blood optical transmission and diffuse reflection change in
certain spectral bands. The spectral bands of high correlation had
in most instances a plausible explanation due to molecular or
clinical factors, a result which serves to validate the
spectroscopy methods employed. Upon comparison of FIGS. 2 and 3,
however, it is evident that broad regions of optical change due to
hemodialysis do not necessarily correspond to similar regions of
high correlation for the chosen clinical measures. Instead, direct
changes in urea and potassium may only have minor effects on the
optical spectrum compared to their indirect effects in altering the
optical properties of blood. The bulk of the optical property
changes seen were likely due to changes in the molecular and
cellular environment of major absorbing and scattering components
in whole blood, namely hemoglobin, oxygen, water, and red blood
cells (RBCs). Besides the chosen clinical measures, a host of other
factors and processes could potentially modify this environment. To
exemplify this point, clinical observations indicate that prior to
hemodialysis treatment patients usually exhibit a mild metabolic
acidosis, while the effect of bicarbonate in standard dialysate
solution results in a mild alkalosis post-treatment.sup.23. Using
near infrared spectroscopy of whole blood it has been reported that
pH-induced changes in the hemoglobin molecule correlate with RBC
size and oxygen saturation changes.sup.24,25. Such changes would
directly modify the optical absorption and scatter properties of
whole blood. An additional related factor is sodium, which plays an
important role in fluid balance regulation and directly affects the
RBC volume, strongly influencing optical absorption and scatter in
the near infrared region.sup.10. In addition, urea removal has also
been proposed as a contributor to RBC volume change.sup.19.
Moreover, in addition to the amplitude changes observed in FIG. 2,
directional changes of extrema are also evident, where minima for
some patients correspond to maxima for others. This indicates
possible competing processes and patient-specific responses,
further illustrating the complex nature of hemodialysis-induced
changes in whole blood. The broad spectral effects observed are not
easily accounted for by measuring the concentration of a few
analytes. While certain correlations between optical properties in
blood and clinical parameter levels exist, the relation is unlikely
a simple causal one. In this respect, full optical spectrum
measures can be more useful in assessing broader factors such as
disease status, treatment efficacy, or patient outcomes.
[0087] Correlation between measured spectra and the clinical
indicators chosen may also be affected by the use of serum-based
analyte levels in the clinic. While analyte levels in hemolyzed
blood may better correlate with whole-blood spectral changes,
routine laboratory analysis procedures were followed in the study
as these represented the clinical standard upon which patients were
assessed and treatment was routinely delivered.
[0088] A potential confounding factor in this study was the chance
that oxygen saturation (O.sub.2-sat) changes in blood could
significantly influence optical properties primarily below 1100 nm.
In the present study mixed arterio-venous blood was drawn following
standard clinical protocol. Blood drawn in this manner typically
has a high partial pressure of oxygen and a corresponding high
O.sub.2-sat, as confirmed by the similarity and consistency of
measured pre- and post-dialysis spectra with that of pure
oxyhemoglobin as seen in FIG. 1c. The maintenance of O.sub.2-sat
levels during hemodialysis has also been noted by others.sup.10.
Spectral features characteristic of a change in the high
O.sub.2-sat level are also absent from the measured difference
spectra. In particular, spectral shape changes in the 540-580 nm
double-peak region and changes in the 760 nm region due to the
presence of deoxyhemoglobin.sup.16 are absent in the difference
spectra shown in FIG. 2. These factors indicate that O.sub.2-sat
changes were minimal and therefore had a negligible confounding
effect upon the analysis.
[0089] Another potential confounder in the analysis was the change
in hematocrit level due to treatment. As hemodialysis removes fluid
while the blood cells remain, a hemoconcentrating effect is
expected leading to an increased fraction of optically absorbing
and scattering species within the blood. Because the standard
clinical protocol used in this study excluded post-dialysis
hematocrit determination, the influence of hematocrit was
investigated using a separate set of blood samples from ten
hemodialysis patients. Hematocrit change due to treatment was
determined by volumetric hematocrit determination after
centrifuging, followed by comparison of pre- to post-dialysis
levels. Change in hematocrit was subsequently compared to the
transmission and diffuse reflection spectrum changes measured in
the samples. Hematocrit change due to treatment was found to vary
from -17% to +20%, reflecting both an increase due to
hemoconcentration as well as a counteracting effect due to rapid
plasma refilling in some patients. Hematocrit changes were compared
with transmission and diffuse reflection changes in the samples and
no significant correlation was found in the 500-1700 nm wavelength
range (data not shown). The effect of hematocrit change on the
observed spectral changes is therefore expected to be minimal.
[0090] Although in this study no attempt was made to smooth,
filter, or pre-process the spectral data, such schemes can be
useful in extracting meaningful spectral information for molecular
identification.sup.27,21 or disease pattern recognition.sup.11.
With a larger patient group and a broader set of clinical
parameters, multivariate methods such as PCA, linear discriminant
analysis, and the partial least squares method can be used with the
full optical spectrum to investigate the underlying mechanisms
resulting in the observed blood changes and to potentially predict
treatment outcomes.
[0091] Finally, although measurements in the present study were
limited to wavelengths shorter than 1700 nm, it is beneficial to
use the entire near infrared wavelength range (up to 2500 nm).
Fundamental overtones of molecular vibrations present at longer
wavelengths yield more distinct spectral features and thereby a
more robust characterization of whole blood properties. While the
absorption due to water increases at longer wavelengths, useful
features in whole blood spectra throughout the near infrared region
have been reported in the literature despite this
interference.sup.26,27.
5. Conclusion
[0092] In the present study, changes in whole blood resulting from
hemodialysis treatment for ESRD were investigated using
transmission and diffuse reflection spectroscopy in the 500-1700 nm
wavelength region. Using the PCA method, the full optical spectrum
of blood from 8 patients was analyzed and it was found that a
significant difference could be detected between dialyzed versus
undialyzed blood in the patient group at a level of P<0.01 in
both transmission and diffuse reflection modes. Consistent changes
in transmission and diffuse reflection difference spectra were also
observed among the diverse patient group as a result of treatment.
The difference spectra were shown in certain wavelength regions to
have significant correlation with clinical measures of hemodialysis
including fluid removal, urea, and potassium (P.ltoreq.0.01 for all
measures). While the spectroscopic techniques presented may provide
a limited usefulness in monitoring specific molecular parameters,
the complexity of hemodialysis-induced changes in whole blood
indicate a full-spectrum monitoring approach can be better suited
to the investigation of macroscopic clinical questions relating to
hemodialysis adequacy, disease progression, and overall toxicity.
Wide-spectrum monitoring combined with a database of spectral
patterns can enable complex relations among numerous parameters to
be recognized as a pattern differing from an ideal or baseline. In
this manner an ensemble of physiologic and molecular changes in the
blood are monitored together with an implicit weighting. The
resulting patterns may correlate better with treatment-related
complications, disease progression, and quality-of-life factors
than current clinical parameters. With rapid, non-contact and
non-destructive monitoring, both time- and wavelength-resolved
information could be significantly more useful, and it is apparent
that extending the present technique to an online system would be
valuable to investigating the provision of optimum hemodialysis for
the ESRD patient population.
Example 2
Correlation of Spectral Data to Clinically Relevant Parameters
[0093] Whole-blood samples from 3 hemodialysis (HD) patients were
extracted before a typical HD session, every hour during the
treatment, and after the session. In total, each patient was
sampled five times with one hour between samples (4-hour HD
session). Each hour, two blood samples were taken from each
patient; one was sent to a laboratory for standard blood analysis
and the other was subjected to analysis with a spectrophotometer.
The transmission and diffuse reflection spectra of each whole blood
sample was taken using the methodology and instrumentation as
described in Example 1, although one skilled in the art could
obtain such spectra by a number of methods and using various
instrumentation.
[0094] The blood laboratory quantified levels of the following
clinical parameters in the blood samples: hemoglobin, hematocrit,
potassium, carbon dioxide, urea, creatinine, phosphate, using
standard clinical chemistry techniques and analyzers. Additionally,
oxygen saturation (O.sub.2-sat) was quantified from the samples
analyzed in the spectrophotometer.
[0095] The O.sub.2-sat value for each blood sample was estimated
using the whole-blood spectrum, similar to the method employed in
clinical oximeters, Each mean blood spectrum was fitted to a linear
combination of published absorption spectra of pure oxyhemoglobin
and deoxyhemoglobin in the 600-950 nut spectral region. Each fit
was optimized by varying both the oxyhemoglobin to total hemoglobin
(oxyhemoglobin+deoxyhemoglobin) ratio (O.sub.2-sat) and a linear
offset value in order to maximize the coefficient of determination
(R.sup.2) between the measured spectrum and the fit. The fit was
performed using the built-in Excel solver (Excel 2002; Microsoft,
Inc., Redmond, Wash.) and yielded optimum O.sub.2-sat estimates for
each sample.
[0096] Transmission and diffuse reflection from the five blood
samples from a typical patient were measured in the wavelength
region spanning 600 nm to 1600 nm. The spectra were
area-normalized, mean-centered, and plotted in absorbance units (as
described in Example 1). At each wavelength data point, the five
absorbance values were correlated with values of each
laboratory-measured blood parameter, using Pearson's r. The
resulting correlation spectra are shown in FIG. 4.
[0097] In addition, the Pearson correlation was repeated amongst
the clinical variables themselves, to yield the following
result:
TABLE-US-00004 Hct K CO.sub.2 urea Creatinine PO.sub.4 O.sub.2Sat
Hgb 0.962 -0.745 0.790 -0.833 -0.816 -0.689 0.554 Hct -0.541 0.604
-0.654 -0.630 -0.472 0.307 K -0.940 0.978 0.982 0.979 -0.952
CO.sub.2 -0.947 -0.949 -0.907 0.896 Urea 1.000 0.974 -0.920 creat-
0.980 -0.931 inine PO.sub.4 -0.977
[0098] For these results, the critical level for two-tailed
significance of r is r.sub.crit=0.878 at the 95% confidence level.
Highlighted values in the above table indicate significant
correlation among clinical parameters.
[0099] There are two groups of correlated clinical variables, and
they are mutually exclusive (i.e., uncorrelated with each
other):
[0100] Group 1: hemoglobin, hematocrit;
[0101] Group 2: potassium, urea, carbon dioxide, creatinine,
phosphate, oxygen saturation
[0102] Looking at the correlation plots in FIG. 4, significant
correlation of Groups 1 and/or 2 were noted in various spectral
regions. In transmission, hemoglobin/hematocrit correlates highly
with most of the spectrum below 1400 nm, while in diffuse
reflection, this correlation was weaker. Hemoglobin/hematocrit is
expected to correlate with transmission as the optical absorption
of hemoglobin dominates whole blood by at least two orders of
magnitude.
[0103] If Group 1 is considered as a source of interference, then
spectral regions in both transmission and diffuse reflection were
identified where Group 2 has high correlation and Group 1 has low
correlation (i.e., transmission: 800-820 nm, 940-960 nm; diffuse
reflection: 1130-1320 nm). These are potential regions for
monitoring patient toxicity over time, in particular for the
patient studied herein. Each patient may have a slightly different
spectral region for monitoring toxicity.
[0104] The present example serves to illustrate one method for
using two different light-tissue interaction techniques to extract
potentially clinically-relevant information from the optical
spectral data.
Example 3
Correlation Method Using an Aggregate Spectrum
[0105] Another approach to incorporate more than one technique to
extract clinically-relevant information from optical spectra is to
concatenate the spectra from two techniques together to create an
`aggregate spectrum`. This aggregate spectrum represents two
distinct types of information (i.e., cell properties and properties
of the extracellular space, or electronic and vibrational molecular
states of molecules in the tissue, etc.). Information contained
within multiple aggregate spectra can be used with established data
reduction techniques, such as singular value decomposition, partial
least squares or principal component analysis, to extract indices
of high correlation with an observed clinical condition, its
progression or its treatment.
[0106] Blood samples were taken from 10 HD patients and 10 healthy
control subjects as described below.
[0107] Blood samples were collected from end-stage renal disease
(ESRD) patients and from healthy subjects. All subjects gave
voluntary consent and the study protocol and use of human subjects
was approved by the Ottawa Hospital Research Ethics Board. ESRD
patients were chosen from the patient population undergoing
3.times. weekly chronic hemodialysis treatments at the Ottawa
Hospital (General Campus) Dialysis Unit. Patients with active
infections were excluded from the study: no other exclusion
criteria were used. Vascular access was via an arterio-venous
fistula or tunneled central venous catheter. Prior to the
initiation of hemodialysis, patients received a standard dose of
heparin to minimize the risk of coagulation during treatment.
[0108] In addition to the measurement of the diffuse reflection
spectrum from these whole blood samples, the transmission spectrum
was also taken (as described in Example 1).
[0109] A principal component analysis was performed using the
diffuse reflection spectra alone, the transmission spectra alone,
and the combined diffuse reflection and transmission spectra. The
table below indicates the results of the analysis:
TABLE-US-00005 Transmission only Diffuse reflection only Combined
Principal % explained % explained % explained Component Eigenvalue
variance Eigenvalue variance Eigenvalue variance 6 0.0262 0.64
0.0135 0.52 0.0889 0.71 5 0.0619 1.51 0.0232 0.9 0.1048 0.83 4
0.0726 1.77 0.0365 1.41 0.3693 2.93 3 0.1983 4.84 0.2925 11.3
0.5546 4.41 2 0.7777 18.98 0.603 23.3 1.9178 15.24 1 2.9179 71.23
1.6001 61.84 9.3855 74.57 Total % 98.97 99.27 98.69
[0110] The eigenvalues from the three analyses are different as are
the distributions of level of explained variance among the
eigenvalues, indicating differences in the information content of
the analysis when both techniques are used together. The principal
component scores for each blood sample with respect to the first
six principal components are also different for each analysis. The
combined analysis, taking into account a larger data set, produces
a set of indices (the principal component scores) that are
fundamentally different from indices created using a single data
set (technique) alone. Indices thus derived from the combined
analysis can correlate better with an observed clinical condition,
progression of a condition or its treatment, than would indices
derived from a single analysis alone.
Example 4
Use of Principal Component Plots
[0111] A set of whole blood samples was obtained from 13 HD
patients pre and post-dialysis in a similar manner as the samples
were obtained in previous examples. Transmission spectra were
obtained as described in Example 1. Additionally, the global mean
of each spectrum was removed from that spectrum (offset removal),
and a principal component analysis (PCA) was performed on this
mean-shifted data. Note that the first step in the PCA algorithm is
to subtract from each data point the mean of all absorbance values
at that wavelength (mean subtraction).
[0112] From the PCA, the proportion (in percent)explained variance
from each principal component is given below for the first 6
principal components (PCs), which together accounted for 99. 1% of
the variance in the data set.
TABLE-US-00006 Principal Explained Component variance (%) 6 0.48 5
2.6 4 4.55 3 9.52 2 21.32 1 60.63
[0113] The principal components themselves (eigenvectors) are
plotted in FIG. 5.
[0114] Oxygen saturation (a potential source of interference) was
assessed in each of the 26 blood samples by the technique described
in Example 2, which uses a different light-tissue interaction
method (diffuse reflection). Two of the 26 spectra had a lower
oxygen saturation (<97%). These two spectra (in mean-subtracted
form) are shown in FIG. 6A, while the rest are shown in FIG.
6B.
[0115] For reference, the published oxyhemoglobin and
deoxyhemoglobin spectra are given in FIG. 7 (taken from S. Prahl,
"Optical absorption of hemoglobin,"
http://omlc.ogi.edu/spectra/hemoglobin/.).
[0116] Comparing this to the PC plot, PCs 2,3,4,6 all have the 740
nm deoxyhemoglobin peak. PC3 in particular is very similar to the
deoxyhemoglobin spectrum in the 600-950 nm region. PCs 1 and 5 do
not seem to be affected by oxygenation.
[0117] A pearson correlation was performed with the estimated
oxygen saturation level, the measured hematocrit level in the
samples and the first 6 PCs:
TABLE-US-00007 PC6 5 4 3 2 1 O.sub.2Sat -0.088 0.054 0.185 -0.735
0.634 -0.183 Hct -0.258 0.424 -0.437 -0.252 -0.203 -0.361
[0118] PCs 2 and 3 are significantly correlated with oxygenation
(based on a two-tailed significance threshold r.sub.crit=0.561 for
significance at the 99% level). The only PCs that do not resemble
the deoxyhemoglobin spectrum are PC1 and PC5.
[0119] PC1 and PC5 could therefore be considered `indices` for the
hemodialysis treatment, substantially free from interference due to
oxygen saturation in the blood sample.
[0120] This is useful, because when all 26 original spectra are
plotted in terms of these two PCs (score plot), we get the
distribution shown in FIG. 8. The centroids (.+-.SD) are given as
well as the average direction of shift with treatment (decreasing
PC1, increasing PC5). Of the 13 patients, all 13 had decreasing PC1
after treatment, and 10 out of 13 had increasing PC5. For Patients
8, 10, and 13 PC5 decreased after treatment.
[0121] The PCs from the above analysis can be considered as the
basis vectors defining the independent sources of variation in
whole blood spectra across patients and treatments.
[0122] If the transmission spectra obtained from the 3 HD patients
(from Example 4, above) are projected onto the basis vector space
defined by the 13 HD patients, and the results are expressed in
terms of PC1 and PC5 scores alone, the score plot shown in FIG. 9
is obtained.
[0123] The scores for each patient are co-located and are separate
from each other. The net direction of change with treatment is
given by the arrows, and for all 3 patients this follows the
general rule of decrease in PC1 and increase in PC5 with
treatment.
[0124] It is interesting to note, however, the evolution of
treatment at 1-hour intervals. There is certainly not a simple
relationship to predict the state of the blood at intermediate
points during treatment. While for all hourly samples Patient J
maintained the same direction, this was not the case for the
others. Patients D and P changed direction in a complex way, but
the net final direction was similar.
[0125] The use of PC1 and PC5 gives us a means to investigate
changes in patients as a result of their treatment. The technique
may also be used to monitor patient blood or tissue over multiple
treatments and in the short and long-term. Anomalies and trends in
the spectral indices observed within patients and across a patient
group may lead to the development of indicators of presence,
progression, treatment, and outcome of a clinical condition. The
use of two different light-tissue interaction techniques in the
development of spectrally-derived indices (in this case, diffuse
reflection to assess the level of a potential source of
interference and transmission to represent the clinical condition)
is beneficial.
Example 5
Monitoring Disease Progression and Treatment
[0126] To illustrate the concept of monitoring patients over
multiple treatments to assess the progression of a disease or
longer-term impact of treatment, four HD patients were recruited.
Blood samples pre- and post-dialysis were taken from each patient
on one Thursday treatment, and this was repeated for the following
three Thursdays. The four patients were then followed up six months
later, with a pre- and post-dialysis blood sample again taken from
a regular Thursday HD treatment session. The transmission spectra
from these blood samples were then projected onto the PCA basis
created from the 13 HD patients as described above, and the
principal component scores for PC1 and PC5 for these patients were
plotted. The result is given in FIG. 10, where the arrows connect
pre-to-post for a single treatment, and the circles indicate the
6-month follow-up scores.
[0127] Besides each patient having pre- and post-hemodialysis data
grouped together, 15 out of 18 treatments had decreasing PC1, and
11 out of 18 treatments had increasing PC5. This was still the
predominant treatment direction.
[0128] More variety is apparent in the direction of treatment,
although where increasing PC1 occurred (3 cases) the magnitude of
the increase was small. When PC5 decreased (7 cases), the magnitude
was sometimes large. One direction seems to be forbidden in all
patients thus far: increasing PC1 and decreasing PC5.
[0129] Patient 1 had anomalies: increasing PC1 and decreasing PC5
in separate weeks.
[0130] Patient 4 had anomalous data every time: either increasing
PC1 or decreasing PC5.
[0131] From these spectral indices it is apparent that dialysis
treatments, even in consecutive weeks in the same patient, may
differ substantially in terms of their impact in altering the light
interaction properties of the blood. Also note that for Patients 1
and 4, 6-month pre-dialysis PC1 scores deviated substantially from
the 4-week baseline values.
[0132] The indices presented provide a means to monitor short and
long-term changes in patients that may be associated with their
state of health and may correlate to eventual clinical outcomes.
These indices were derived by means of combining diffuse reflection
data (to quantify the level of interference from oxygen saturation
and allow suitable parameters to be chosen to minimize this source
of interference) with transmission data (containing the spectral
information indicative of the clinical condition).
[0133] All publications, patents and patent applications mentioned
in this Specification are indicative of the level of skill of those
skilled in the art to which this invention pertains and are herein
incorporated by reference to the same extent as if each individual
publication, patent, or patent applications was specifically and
individually indicated to be incorporated by reference.
[0134] The invention being thus described, it will be obvious that
the same may be varied in many ways. Such variations are not to be
regarded as a departure from the spirit and scope of the invention,
and all such modifications as would be obvious to one skilled in
the art are intended to be included within the scope of the
following claims.
REFERENCES
[0135] 1. 2001 Report, Vol. 1: Dialysis and Renal Transplantation,
Canadian Organ Replacement Register, Canadian Institute for Health
Information, Ottawa, August 2001. [0136] 2. P. S. Jensen, J. Bak,
S. Ladefoged, and S. Andersson-Engels, "Determination of urea,
glucose, and phosphate in dialysate with Fourier transform infrared
spectroscopy," Spectrochim. Acta., Part A 60, 899-905 (2004).
[0137] 3. R. Vanholder, G. Glorieux, R. de Smet, and N. Lameire,
"New insights in uremic toxins," Kidney Int. 63, supplement 84,
S6-S10 (2003). [0138] 4. R. Vanholder, A. Argiles, U. Baurmeister
et al., "Uremic toxicity: present state of the art," Int. J. Artif.
Organs 24, 695-725 (2001). [0139] 5. P. S. Jensen, J. Bak, S.
Ladefoged, S. Andersson-Engels, and L. Friis-Hansen, "Online
monitoring of urea concentration in dialysate with dual-beam
Fourier transform near-infared spectroscopy," J. Biomed. Opt. 9(3),
553-557 (2004). [0140] 6. C. V. Eddy and M. A. Arnold,
"Near-infrared spectroscopy for measuring urea in hemodialysis
fluids," Clin. Chem. 47(7), 1279-1286 (2001). [0141] 7. J. T.
Olesberg, M. A. Arnold, and M. J. Flanigan, "Online measurement of
urea concentration in spent dialysate during hemodialysis," Clin.
Chem. 50(1), 175-181 (2004). [0142] 8. E. Mancini, A. Santoro, M.
Spongano, F. Paolini, M. Rossi, and P. Zucchelli, "Continuous
on-line optical absorbance recording of blood volume changes during
hemodialysis," Artif. Organs 17(8), 691-694 (1993). [0143] 9. J. P.
M. de Vries, C. G. Olthof, V. Visser, P. M. Kouw, A. van Es, A. J.
M. Donker, and P. M. J. M. de Vries, "Continuous measurement of
blood volume during hemodialysis by an optical method," ASAIO J.
38, M181-185 (1992). [0144] 10. R. R. Steuer, D. A. Bell, and L. L.
Barrett, "Optical measurement of hematocrit and other biological
constituents in renal therapy," Adv. Renal Repl. Ther. 6(3),
217-224 (1999). [0145] 11. W. Petrich, B. Dolenko, J. Fruh et al.,
"Disease pattern recognition in infrared spectra of human sera with
diabetes mellitus as an example," Appl. Opt. 39(19), 3372-3379
(2000). [0146] 12. Z. Ge, K. T. Schomacker, and N. S. Nishioka,
"Identification of colonic dysplasia and neoplasia by diffuse
reflectance spectroscopy and pattern recognition techniques," Appl.
Spectrosc. 52(6), 833-839 (1998). [0147] 13. B. K. Lavine, C. E.
Davidson, A. J. Moores, "Genetic algorithms for spectral pattern
recognition," Vibr. Spectrosc. 28, 83-95 (2002). [0148] 14. W. Lin,
X. Yuan, P. Yuen, W. I. Wei, J. Sham, P. Shi, and J. Qu,
"Classification of in vivo autofluorescence spectra using support
vector machines," J. Biomed. Opt. 9(1), 180-186 (2004). [0149] 15.
R. C. Schneider, K. A. Kovar, "Analysis of ecstasy tablets:
comparison of reflectance and transmittance near infrared
spectroscopy," Forensic Sci. Int. 134, 187-195 (2003). [0150] 16.
S. Prahl, "Optical absorption of hemoglobin,"
http://omlc.ogi.edu/spectra/hemoglobin/. [0151] 17. G. M. Hale and
M. R. Querry, "Optical constants of water in the 200 nm to 200
.mu.m wavelength region," Appl. Opt. 12(3), 555-563 (1973). [0152]
18. I. A. Cowe and J. W. McNicol, "The use of principal components
in the analysis of near-infrared spectra," Appl. Spectrosc. 39(2),
257-266 (1985). [0153] 19. G. A. Martinez and R. Bragos, "On-line
measurement of urea in blood using optical spectroscopy in the
visible range; validation of the cell shrinkage hypothesis," IEEE
Instrumentation and measurement technology conference ITMC 2004,
Como, Italy, 1966-1969 (2004). [0154] 20. B. R. Soller, J. Favreau,
and P. O. Idwasi, "Investigation of electrolyte measurement in
diluted whole blood using spectroscopic and chemometric methods,"
Appl. Spectrosc. 57(2), 146-151 (2003). [0155] 21. Y. J. Kim, S.
Kim, J. W. Kim, and G. Yoon, "Data preprocessing and partial least
squares regression analysis for reagentless determination of
hemoglobin concentrations using conventional and total transmission
spectroscopy," J. Biomed. Opt. 6(2), 177-182 (2001). [0156] 22. A.
S. Goldfarb-Rumyantzev, M. H. Schwenk, S. Liu, E. Wrone, and J. K.
Leypoldt, "New empiric expressions to calculate single pool Kt/V
and equilibrated Kt/V," ASAIO J. 48(5), 570-576 (2002). [0157] 23.
L. Gabutti, N. Ferrari, G. Giudici, G. Mombelli, and C. Marone,
"Unexpected haemodynamic instability associated with standard
bicarbonate haemodialysis," Nephrol. Dial. Transplant. 18(11),
2369-2376 (2003). [0158] 24. M. K. Alam, M. R. Rohrscheib, J. E.
Franke, T. M. Niemczyk, J. D. Maynard, and M. R. Robinson,
"Measurement of pH in whole blood by near-infrared spectroscopy,"
Appl. Spectrosc. 53(3), 316-324 (1999). [0159] 25. N. A. Rosen, W.
E. Charash, and E. F. Hirsch, "Near-infrared spectrometric
determination of blood pH," J. Surg. Res. 106(2), 282-286 (2002).
[0160] 26. I. Valyi-Nagy, K. J. Kaffka, J. M. Jako, E. Gonczol, and
G. Domjan, "Application of near infrared spectroscopy to the
determination of haemoglobin," Clin. Chim. Acta 264, 117-125
(1997). [0161] 27. J. T. Kuenstner, K. H. Norris, and W. F.
McCarthy, "Measurement of hemoglobin in unlysed blood by
near-infrared spectroscopy," Appl. Spectrosc. 48(4), 484-488
(1994).
* * * * *
References