U.S. patent application number 10/828624 was filed with the patent office on 2004-10-14 for optimal windows for obtaining optical data for characterization of tissue samples.
This patent application is currently assigned to MediSpectra, Inc.. Invention is credited to Flewelling, Ross, Kaufman, Howard, Schomacker, Kevin T., Zelenchuk, Alex.
Application Number | 20040204648 10/828624 |
Document ID | / |
Family ID | 30117977 |
Filed Date | 2004-10-14 |
United States Patent
Application |
20040204648 |
Kind Code |
A1 |
Schomacker, Kevin T. ; et
al. |
October 14, 2004 |
Optimal windows for obtaining optical data for characterization of
tissue samples
Abstract
The invention provides methods for determining a characteristic
of a tissue sample, such as a state of health, using spectral data
and/or images obtained within an optimal period of time following
the application of a chemical agent to the tissue sample. The
invention provides methods of determining such optimal windows of
time. Similarly, the invention provides methods of determining
other criteria for triggering the acquisition of an optical signal
for classifying the state of health of a region of a tissue
sample.
Inventors: |
Schomacker, Kevin T.;
(Maynard, MA) ; Zelenchuk, Alex; (Stoughton,
MA) ; Flewelling, Ross; (Chelmsford, MA) ;
Kaufman, Howard; (Newton, MA) |
Correspondence
Address: |
TESTA, HURWITZ & THIBEAULT, LLP
HIGH STREET TOWER
125 HIGH STREET
BOSTON
MA
02110
US
|
Assignee: |
MediSpectra, Inc.
Lexington
MA
|
Family ID: |
30117977 |
Appl. No.: |
10/828624 |
Filed: |
April 21, 2004 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
10828624 |
Apr 21, 2004 |
|
|
|
10295794 |
Nov 15, 2002 |
|
|
|
60394696 |
Jul 9, 2002 |
|
|
|
Current U.S.
Class: |
600/431 ;
424/9.6; 436/164 |
Current CPC
Class: |
A61B 5/7264 20130101;
A61B 5/0059 20130101; G01N 21/6428 20130101; Y10S 435/808 20130101;
G01N 21/31 20130101; A61B 5/7267 20130101; G01N 21/6408 20130101;
Y10T 436/25 20150115 |
Class at
Publication: |
600/431 ;
436/164; 424/009.6 |
International
Class: |
A61B 006/00 |
Claims
What is claimed is:
1-15 (cancelled)
16. A method of identifying a characteristic of a region of a
tissue sample, the method comprising the steps of: (a) applying a
contrast agent to a region of a tissue sample; (b) obtaining at
least one reflectance signal from the region of the tissue sample
within a window of time, wherein the window of time begins at about
30 seconds following application of the contrast agent and ends at
about 130 seconds following application of the contrast agent; (c)
obtaining a fluorescence signal from the region of the tissue
sample within the window of time; and (d) identifying a
characteristic of the region based at least in part on the
fluorescence signal and at least one of the at least one
reflectance signals.
17 The method of claim 16, the method further comprising the step
of obtaining a video signal from the region of the tissue sample
within the window of time.
18 The method of claim 17, wherein step (d) comprises identifying a
characteristic of the region based at least in part on the
fluorescence signal, at least one of the at least one reflectance
signals, and the video signal.
19 The method of claim 16, wherein step (b) comprises obtaining two
reflectance signals from the region of the tissue sample within the
window of time.
20-41 (cancelled)
42 The method of claim 16, wherein the characteristic is a state of
health.
43 The method of claim 42, wherein the state of health comprises at
least one of the group consisting of normal squamous tissue, normal
columnar tissue, metaplasia, immature metaplasia, mature
metaplasia, CIN1, CIN2, CIN3, CIS, and cancer.
44 The method of claim 16, wherein the identifying step (d)
comprises determining whether the region of the tissue sample is
CIN 2+ tissue.
45 The method of claim 16, wherein the contrast agent comprises
acetic acid.
46 The method of claim 16, wherein the contrast agent is selected
from a group consisting of formic acid, propionic acid, butyric
acid, Lugol's iodine, Shiller's iodine, methylene blue, toluidine
blue, indigo carmine, indocyanine green, and fluorescein.
47 The method of claim 16, wherein the tissue sample comprises
cervical tissue.
48 The method of claim 16, wherein the tissue sample comprises at
least one of a group consisting of colorectal tissue,
gastroesophageal tissue, urinary bladder tissue, lung tissue, and
skin tissue.
49 The method of claim 16, wherein the tissue sample comprises
epithelial cells.
50 The method of claim 16, wherein step (b) comprises obtaining the
at least one reflectance signal from the region within a period of
time that begins at about 60 seconds following application of the
contrast agent and ends at about 80 seconds following application
of the contrast agent.
51 The method of claim 16, wherein step (b) comprises obtaining the
at least one reflectance signal from the region within a period of
time that begins at about 70 seconds following application of the
contrast agent and ends at about 130 seconds following application
of the contrast agent.
52 The method of claim 16, wherein step (d) comprises identifying
the characteristic of the region with an accuracy of at least about
70%.
53 The method of claim 16, wherein step (b) comprises obtaining a
reflectance intensity from the region at each of a plurality of
wavelengths within the window of time.
54 The method of claim 16, wherein step (c) comprises obtaining a
fluorescence intensity from the region at each of a plurality of
wavelengths within the window of time.
55 The method of claim 54, wherein step (d) further comprises
obtaining a video signal from the region within the window of
time.
56 The method of claim 16, wherein step (a) comprises applying a
contrast agent to a plurality of regions of the tissue sample; step
(b) comprises obtaining at least one reflectance signal from each
of the plurality of regions within the window of time; step (c)
comprises obtaining a fluorescence signal from each of the
plurality of regions within the window of time; and step (d)
comprises identifying a characteristic of each of the plurality of
regions.
57 The method of claim 16, wherein step (d) comprises identifying a
characteristic of the region based substantially on at least one
optical signal, obtained within the window of time.
58 The method of claim 16, wherein the window of time begins at 20
seconds following application of the contrast agent.
Description
PRIOR APPLICATIONS
[0001] The present application claims the benefit of U.S.
Provisional Patent Application Serial No. 60/394,696, filed Jul. 9,
2002, which is hereby incorporated by reference.
FIELD OF THE INVENTION
[0002] The invention relates generally to spectroscopic methods.
More particularly, the invention relates to the diagnosis of
disease in tissue using spectral analysis and/or image
analysis.
BACKGROUND OF THE INVENTION
[0003] Spectral analysis is used to diagnose disease in tissue. For
example, data from spectral scans performed on the tissue of a
patient are used to screen tissue for disease. Some diagnostic
procedures include the application of a chemical contrast agent to
the tissue in order to enhance the image and/or spectral response
of the tissue for diagnosis. In an acetowhitening procedure, acetic
acid is used as the contrast agent. Use of a contrast agent
enhances the difference between data obtained from normal tissue
and data obtained from abnormal or diseased tissue.
[0004] Current techniques do not suggest an optimal time period
following application of a contrast agent within which to obtain
spectral and/or image data for the diagnosis of disease, nor do
current techniques suggest how such an optimal time period could be
determined.
SUMMARY OF THE INVENTION
[0005] The invention provides optimal criteria for selecting
spectral and/or image data from tissue that has been treated with a
contrast agent for disease screening. In particular, it has been
discovered that the sensitivity and specificity. of optical
diagnostic screening is improved by obtaining optical data at
optimal time points after application of a contrast agent.
[0006] Accordingly, methods of the invention provide optimal
windows in time for obtaining spectral data from tissue that has
been treated with a contrast agent in order to improve the results
of disease screening. The invention further provides methods for
identifying such windows in the context of any optical diagnostic
screen. Additionally, the invention provides methods for disease
screening using kinetic data obtained across multiple diagnostic
windows. Methods of the invention allow an optical diagnostic test
to focus on data that will produce the highest diagnostic
sensitivity and specificity with respect to the tissue being
examined. Thus, the invention allows the identification of specific
points in time after treatment of a tissue when spectral and/or
image data most accurately reflects the health of the tissue being
measured.
[0007] Time windows for observing selected spectral data may be
determined empirically or from a database of known tissue responses
to optical stimulation. For example, in one aspect the invention
comprises building and using classification models to characterize
the state of health of an unknown tissue sample from which optical
signals are obtained. As used herein, an optical signal may
comprise a discrete or continuous electromagnetic signal or any
portion thereof, or the data representing such a signal.
Essentially, optical diagnostic windows are based upon the points
at which classification models perform best. In practice, optimal
diagnostic windows of the invention may be predetermined segments
of time following application of a contrast agent to a tissue.
Optimal diagnostic windows may also be points in time at which an
optical measurement meets a predetermined threshold or falls within
a predetermined range, where the optical measurement represents the
change of an optical signal received from the tissue following
application of a contrast agent. For example, a window may be
selected to include points in time at which the change in optical
signal intensity from an initial condition is maximized. Finally,
the optical measurement upon which a window is based may also
reflect the rate of change in a spectral property obtained from the
tissue.
[0008] In a preferred embodiment, optimal windows are determined by
obtaining optical signals from reference tissue samples with known
states of health at various times following application of a
contrast agent. For example, one embodiment comprises obtaining a
first set of optical signals from tissue samples having a known
disease state, such as CIN 2/3 (grades 2 and/or 3 cervical
intraepithelial neoplasia); obtaining a second set of optical
signals from tissue samples having a different state of health,
such as non-diseased; and categorizing each optical signal into
"bins" according to the time it was obtained in relation to the
time of application of contrast agent. The optical signal may
comprise, for example, a reflectance spectrum, a fluorescence
spectrum, a video image intensity signal, or any combination of
these.
[0009] A measure of the difference between the optical signals
associated with the two types of tissue is then obtained, for
example, by determining a mean signal as a function of wavelength
for each of the two types of tissue samples for each time bin, and
using a discrimination function to determine a weighted measure of
difference between the two mean optical signals obtained within a
given time bin. This provides a measure of the difference between
the mean optical signals of the two categories of tissue
samples--diseased and healthy--weighted by the variance between
optical signals of samples within each of the two categories.
[0010] In one embodiment, the invention further comprises
developing a classification model for each time bin. After
determining a measure of difference between the tissue types in
each bin, an optimal window of time for differentiating between
tissue types is determined by identifying at least one bin in which
the measure of difference between the two tissue types is
substantially maximized. For example, an optimal window of time may
be chosen to include every time bin in which the respective
classification model provides an accuracy of 70% or greater. Here,
the optimal window describes a period of time following application
of a contrast agent in which an optical signal can be obtained for
purposes of classifying the state of health of the tissue sample
with an accuracy of at least 70%.
[0011] An analogous embodiment comprises determining an optimal
threshold or range of a measure of change of an optical signal to
use in obtaining (or triggering the acquisition of) the same or a
different signal for predicting the state of health of the sample.
Instead of determining a specific, fixed window of time, this
embodiment includes determining an optimal threshold of change in a
signal, such as a video image whiteness intensity signal, after
which an optical signal, such as a diffuse reflectance spectrum
and/or a fluorescence spectrum, can be obtained to accurately
characterize the state of health or other characteristic of the
sample. An embodiment includes monitoring reflectance and/or
fluorescence at a single or multiple wavelength(s), and upon
reaching a threshold change from the initial condition, obtaining a
full reflectance and/or fluorescence spectrum for use in diagnosing
the region of tissue. This method allows for reduced data retrieval
and monitoring since, in an embodiment, it involves continuous
tracking of a single, partial-spectrum or discrete-wavelength
"trigger" signal (instead of multiple, full-spectrum scans),
followed by the acquisition of one or more spectral scans for use
in diagnosis. Alternatively, the trigger may include more than one
discrete-wavelength or partial-spectrum signal. The diagnostic data
obtained will generally be more extensive than the trigger signal,
and may include one or more complete sets of spectral data. The
measure of change used to trigger obtaining one or more optical
signals for tissue classification may be a weighted measure, and/or
it may be a combination of measures of change of more than one
signal. The signal(s) used for tissue classification/diagnosis may
comprise one or more reflectance, fluorescence, and/or video
signals. In one embodiment, two reflectance signals are obtained
from the same region in order to provide a redundant signal for use
when one reflectance signal is adversely affected by an artifact
such as glare or shadow. Use of multiple types of classification
signals may provide improved diagnostic accuracy over the use of a
single type of signal. In one embodiment, a reflectance,
fluorescence, and a video signal from a region of a tissue sample
are all used in the classification of the region.
[0012] In a further embodiment, instead of determining an optimal
threshold or range of a measure of change of an optical signal, an
optimal threshold or range of a measure of the rate of change of an
optical signal is determined. For example, the rate of change of
reflectance and/or fluorescence is monitored at a single or
multiple wavelength(s), and upon reaching a threshold rate of
change, a full reflectance spectrum and/or fluorescence spectrum is
acquired for use in diagnosing the region of tissue. The measure of
rate of change used to trigger obtaining one or more optical
signals for tissue classification may be a weighted measure, and/or
it may be combination of measures of change of more than one
signal. For example, the measured rate of change may be weighted by
an initial signal intensity.
[0013] The invention also provides methods of disease screening
using kinetic data from optical signals obtained at various times
following application of a contrast agent. These methods comprise
techniques for using specific features of fluorescence and diffuse
reflectance spectra from reference cervical tissue samples of known
states of health in order to diagnose a region of a tissue sample.
These techniques allow monitoring of a particular optical signal
from a test sample during a specified period of time following
application of contrast agent to obtain pertinent kinetic data for
characterizing the sample. For example, two or more time-separated
measures of video intensity, fluorescence, and/or reflectance are
obtained for a test sample at times between which it is known that
an increase or decrease indicative of a given state of health
occurs. It is therefore possible to determine whether this increase
or decrease has occurred for the test sample, thereby indicating
the sample may have a given state of health. Alternatively or
additionally, a video, reflectance, and/or fluorescence signal from
a test sample may be monitored over time to determine a time at
which the signal reaches a maximum or minimum value. The time
following application of contrast agent at which this minimum or
maximum is reached can then be used to determine indication of a
disease state in the test sample.
[0014] In one embodiment, data used as a baseline in determining an
increase, decrease, maximum, or minimum as discussed above is not
obtained before, but is obtained immediately following application
of contrast agent to the tissue. In one case, the time period
immediately following application of contrast agent is about ten
seconds, and in another case, it is about five seconds, although
other time periods are possible. This may be done to avoid error
caused by movement of tissue or movement of the optical signal
detection device upon application of contrast agent, particularly
where such movement is not otherwise compensated for. Movement of
tissue may cause error where a change from an initial condition is
being monitored and the region of the tissue corresponding to the
location at which the initial signal was obtained shifts following
application of contrast agent.
BRIEF DESCRIPTION OF THE DRAWINGS
[0015] The objects and features of the invention can be better
understood with reference to the drawings described below, and the
claims. The drawings are not necessarily to scale, emphasis instead
generally being placed upon illustrating the principles of the
invention. In the drawings, like numerals are used to indicate like
parts throughout the various views.
[0016] FIG. 1A shows a graph depicting mean fluorescence spectra
before application of acetic acid and at various times following
the application of acetic acid for NED tissue (no evidence of
disease, confirmed by pathology).
[0017] FIG. 1B shows a graph depicting mean reflectance spectra
before application of acetic acid and at various times following
the application of acetic acid for NED tissue (no evidence of
disease, confirmed by pathology).
[0018] FIG. 2A shows a graph depicting mean fluorescence spectra
before application of acetic acid and at various times following
the application of acetic acid for CIN 2/3 tissue (grades 2 and/or
3 cervical intraepithelial neoplasia).
[0019] FIG. 2B shows a graph depicting mean reflectance spectra
before application of acetic acid and at various times following
the application of acetic acid for CIN 2/3 tissue (grades 2 and/or
3 cervical intraepithelial neoplasia).
[0020] FIG. 3A shows a graph depicting fluorescence intensity at
three different wavelengths relative to pre-AA (fluorescence before
application of acetic acid) as a function of time following
application of acetic acid for NED tissue.
[0021] FIG. 3B shows a graph depicting reflectance at three
different wavelengths relative to pre-AA (reflectance before
application of acetic acid) as a function of time following
application of acetic acid for NED tissue.
[0022] FIG. 3C shows a graph depicting fluorescence intensity at
three different wavelengths relative to pre-AA (fluorescence before
application of acetic acid) as a function of time following
application of acetic acid for CIN 2/3 tissue.
[0023] FIG. 3D shows a graph depicting reflectance at three
different wavelengths relative to pre-AA (reflectance before
application of acetic acid) as a function of time following
application of acetic acid for CIN 2/3 tissue.
[0024] FIG. 4A shows a graph depicting reflectance relative to
pre-AA at 425 nm as a function of time following application of
acetic acid for various tissue types.
[0025] FIG. 4B shows a graph depicting fluorescence relative to
pre-AA at 460 nm as a function of time following application of
acetic acid for various tissue types.
[0026] FIG. 5 shows a series of graphs depicting mean reflectance
spectra for CIN 2/3 and non-CIN 2/3 (NED and CIN 1) tissues at a
time prior to application of acetic acid, at a time corresponding
to maximum whitening, and at a time corresponding to the latest
time at which data was obtained.
[0027] FIG. 6 shows a graph depicting the reflectance
discrimination function spectra useful for differentiating between
CIN 2/3 and non-CIN 2/3 (NED and CIN 1) tissues.
[0028] FIG. 7 shows a graph depicting the performance of two LDA
(linear discriminant analysis) models as applied to reflectance
data obtained at various times following application of acetic
acid; one of the models is based on data obtained between 60 and 80
seconds following application of acetic acid, and the other model
is based on data obtained between 160 and 180 seconds following
application of acetic acid.
[0029] FIG. 8 shows a series of graphs depicting mean fluorescence
spectra for CIN 2/3 and non-CIN 2/3 (NED and CIN 1) tissues at a
time prior to application of acetic acid, at a time corresponding
to maximum whitening, and at a time corresponding to the latest
time at which data was obtained.
[0030] FIG. 9 shows a graph depicting the fluorescence
discrimination function spectra useful for differentiating between
CIN 2/3 and non-CIN 2/3 (NED and CIN 1) tissues.
[0031] FIG. 10 shows a graph depicting the performance of two LDA
(linear discriminant analysis) models as applied to fluorescence
data obtained at various times following application of acetic
acid; one of the models is based on data obtained between 60 and 80
seconds following application of acetic acid, and the other model
is based on data obtained between 160 and 180 seconds following
application of acetic acid.
[0032] FIG. 11 shows a graph depicting the performance of three LDA
models as applied to data obtained at various times following
application of acetic acid.
[0033] FIG. 12A shows a graph depicting the determination of an
optimal time window for obtaining diagnostic optical data using an
optical amplitude trigger.
[0034] FIG. 12B shows a graph depicting the determination of an
optimal time window for obtaining diagnostic data using a rate of
change of mean reflectance signal trigger.
DESCRIPTION OF THE ILLUSTRATIVE EMBODIMENT
[0035] The invention relates to methods for determining a
characteristic of a tissue sample using spectral data and/or images
obtained within an optimal window of time following the application
of a chemical agent to the tissue sample. The invention provides
methods of determining optimal windows of time. Similarly, the
invention provides methods of determining criteria, based on a
spectral amplitude or rate of amplitude change, for triggering the
acquisition of an optical signal for classifying tissue. Finally,
the invention comprises methods of diagnosing a tissue sample using
spectral data and/or images obtained within an optimal window.
[0036] Application of the invention allows the diagnosis of regions
of a tissue sample using various features of the time response of
fluorescence and/or reflectance spectra following the application
of a contrast agent such as acetic acid. For example, it is
possible to diagnose a region of a tissue sample by determining a
time at which a minimum value of fluorescence spectral intensity is
reached following application of a contrast agent.
[0037] Methods of the invention are also used to analyze tissue
samples, including cervical tissue, colorectal tissue,
gastroesophageal tissue, urinary bladder tissue, lung tissue, or
other tissue containing epithelial cells. The tissue may be
analyzed in vivo or ex vivo, for example. Tissue samples are
generally divided into regions, each having its own characteristic.
This characteristic may be a state of health, such as
intraepithelial neoplasia, mature and immature metaplasia, normal
columnar epithelia, normal squamous epithelia, and cancer. Chemical
contrast agents which are used in practice of the invention include
acetic acid, formic acid, propionic acid, butyric acid, Lugol's
iodine, Shiller's iodine, methylene blue, toluidine blue, indigo
carmine, indocyanine green, fluorescein, and combinations
comprising these agents. In embodiments where acetic acid is used,
concentrations between about 3 volume percent and about 6 volume
percent acetic acid are typical, although in some embodiments,
concentrations outside this range may be used. In one embodiment, a
5 volume percent solution of acetic acid is used as contrast
agent.
[0038] Optical signals used in practice of the invention comprise,
for example, fluorescence, reflectance, Raman, infrared, and video
signals. Video signals comprise images from standard
black-and-white or color CCD cameras, as well as hyperspectral
imaging signals based on fluorescence, reflectance, Raman,
infrared, and other spectroscopic techniques. For example, an
embodiment comprises analyzing an intensity component indicative of
the "whiteness" of a pixel in an image during an acetowhitening
test.
[0039] A preferred embodiment uses optical signals obtained from
tissue samples within optimal windows of time. Obtaining an optical
signal may comprise actually acquiring a signal within an optimal
window of time, or, of course, simply triggering the acquisition of
an optical signal within an optimal window of time. The optimal
window of time may account for a delay between the triggering of
the acquisition of a signal, and its actual acquisition. An
embodiment of the invention may comprise determining an optimal
window of time in which to trigger the acquisition of an optical
signal, as well as determining an optimal window of time in which
to actually acquire an optical signal.
[0040] One embodiment comprises determining an optimum time window
in which to obtain spectra from cervical tissue such that sites
indicative of grades 2 and 3 cervical intraepithelial neoplasia
(CIN 2/3) can be separated from non-CIN 2/3 sites. Non-CIN 2/3
sites include sites with grade 1 cervical intraepithelial neoplasia
(CIN 1), as well as NED sites (which include mature and immature
metaplasia, and normal columnar and normal squamous epithelia).
Alternately, sites indicative of high grade disease, CIN 2+, which
includes CIN 2/3 categories, carcinoma in situ (CIS), and cancer,
may be separated from non-high-grade-disease sites. In general, for
any embodiment in which CIN 2/3 is used as a category for
classification or characterization of tissue, the more expansive
category CIN 2+ may be used alternatively. One embodiment comprises
differentiating amongst three or more classification categories.
Exemplary embodiments are described below and comprise analysis of
the time response of diffuse reflectance and/or 337-nm fluorescence
spectra of a set of reference tissue samples with regions having
known states of health, as listed in the Appendix Table, to
determine temporal characteristics indicative of the respective
states of health. These characteristics are then used in building a
model to determine a state of health of an unknown tissue sample.
Other embodiments comprise analysis of fluorescence spectra using
other excitation wavelengths, such as 380 nm and 460 nm, for
example.
[0041] While the invention is particularly shown and described
herein with reference to specific examples and specific
embodiments, it should be understood by those skilled in the art
that various changes in form and detail may be made therein without
departing from the spirit and scope of the invention.
EXAMPLE 1
Analysis of the Temporal Evolution of Spectral Data from Reference
Samples with Known States of Health.
[0042] Diffuse reflectance and/or 337-nm fluorescence emission
spectra are taken from cervical tissue samples that are categorized
as CIN 2/3 (having grades 2 and/or 3 cervical intraepithelial
neoplasia), CIN 1 and NED (no evidence of disease, confirmed by
pathology, including normal squamous tissue, normal columnar
tissue, immature metaplasia tissue, and mature metaplasia tissue).
All spectra are filtered then placed in the time bins indicated in
Table 1. Data affected by artifacts such as glare, shadow, or
obstructions may be removed and/or compensated for by using the
technique disclosed in the co-owned U.S. patent application
entitled, "Method and Apparatus for Identifying Spectral
Artifacts," filed on Sep. 13, 2002, and identified by attorney
docket number MDS-033, the contents of which are hereby
incorporated by reference. Means spectra and standard deviations
are calculated for the spectra in each time bin. Although not shown
in this example, some embodiments use spectral and/or image data
obtained at times greater than 180 s following application of
contrast agent.
1TABLE 1 Time bins for which means spectra are calculated in an
exemplary embodiment Bin Time after application of Acetic Acid (s)
1 t .ltoreq. 0 2 0 < t .ltoreq. 40 3 40 < t .ltoreq. 60 4 60
< t .ltoreq. 80 5 80 < t .ltoreq. 100 6 100 < t .ltoreq.
120 7 120 < t .ltoreq. 140 8 140 < t .ltoreq. 160 9 160 <
t .ltoreq. 180 10 t > 180
[0043] FIGS. 1A, 1B, 2A, and 2B show mean fluorescence and
reflectance spectra for exemplary healthy tissue (NED tissue--no
evidence of disease, confirmed by pathology) and CIN 2/3 (grades 2
and/or 3 cervical intraepithelial neoplasia) tissue samples. These
figures demonstrate the temporal effect of acetic acid on the
spectral data. In the application of one embodiment, one or more
characteristics of the time responses shown in FIGS. 1A, 1B, 2A,
and 2B are determined. Subsequently, the time response of a sample
of unknown type is obtained, and the sample is then diagnosed
according to one or more features of the response, compared against
those of the known sample set.
[0044] FIG. 1A shows a graph 102 depicting mean fluorescence
spectra for each of the 10 time bins 108 of Table 1 for NED tissue
(no evidence of disease, confirmed by pathology). Mean fluorescence
intensity (relative counts/.mu.J) 104 is plotted as a function of
wavelength (nm) 106 for each time bin shown in the legend 108. The
curve corresponding to the first time bin 110 is a graph of the
mean fluorescence intensity as a function of wavelength for data
collected prior to acetic acid application, and the curve
corresponding to the last time bin 128 is a graph of the mean
fluorescence intensity as a function of wavelength for data
collected at times greater than 180 seconds (with an average of
about 210 seconds). Each of the curves in between (112, 114, 116,
118, 120, 122, 124, 126) is a graph of the mean fluorescence
intensity as a function of wavelength for data collected in the
respective time bin shown in the legend 108. The value of N shown
in the legend 108 beside each curve denotes the number of spectra
that are in the respective time bin for this particular
embodiment.
[0045] FIG. 1B shows a graph 150 depicting mean reflectance spectra
for each of the 10 time bins 108 of Table 1 for NED tissue (no
evidence of disease, confirmed by pathology). Mean reflectance 152
is plotted as a function of wavelength (nm) 106 for each time bin
shown in the legend 108. The curve corresponding to the first time
bin 154 is a graph of the mean reflectance as a function of
wavelength for data collected prior to acetic acid application, and
the curve corresponding to the last time bin 172 is a graph of the
mean reflectance as a function of wavelength for data collected at
times greater than 180 seconds (with an average of about 210
seconds). Each of the curves in between (156, 158, 160, 162, 164,
166, 168, 170) is a graph of the mean reflectance as a function of
wavelength for data collected in the respective time bin shown in
the legend 108. The value of N shown in the legend 108 beside each
curve denotes the number of spectra that are in the respective time
bin for this particular embodiment.
[0046] FIG. 2A shows a graph 202 depicting mean fluorescence
spectra for each of the 10 time bins 204 of Table 1 for CIN 2/3
tissue (grades 2 and/or 3 cervical intraepithelial neoplasia). Mean
fluorescence intensity (relative counts/.mu.J) 104 is plotted as a
function of wavelength (nm) 106 for each time bin shown in the
legend 204. The curve corresponding to the first time bin 206 is a
graph of the mean fluorescence intensity as a function of
wavelength for data collected prior to acetic acid application, and
the curve corresponding to the last time bin 224 is a graph of the
mean fluorescence intensity as a function of wavelength for data
collected at times greater than 180 seconds (with an average of
about 210 seconds). Each of the curves in between (208, 210, 212,
214, 216, 218, 220, 220) is a graph of the mean fluorescence
intensity as a function of wavelength for data collected in the
respective time bin shown in the legend 204. The value of N shown
in the legend 204 beside each curve denotes the number of spectra
that are in the respective time bin for this particular
embodiment.
[0047] FIG. 2B shows a graph 250 depicting mean reflectance spectra
for each of the 10 time bins 204 of Table 1 for CIN 2/3 tissue
(grades 2 and/or 3 cervical intraepithelial neoplasia). Mean
reflectance 152 is plotted as a function of wavelength (mn) 106 for
each time bin shown in the legend 204. The curve corresponding to
the first time bin 254 is a graph of the mean reflectance as a
function of wavelength for data collected prior to acetic acid
application, and the curve corresponding to the last time bin 272
is a graph of the mean reflectance as a function of wavelength for
data collected at times greater than 180 seconds (with an average
of about 210 seconds). Each of the curves in between (256, 258,
260, 262, 264, 266, 268, 270) is a graph of the mean reflectance as
a function of wavelength for data collected in the respective time
bin shown in the legend 204. The value of N shown in the legend 204
beside each curve denotes the number of spectra that are in the
respective time bin for this particular embodiment.
EXAMPLE 2
Analysis of Optical Kinetic Data from Reference Samples with Known
States of Health
[0048] Data from FIGS. 1A, 1B, 2A, and 2B are further analyzed as
shown in FIGS. 3A, 3B, 3C, and 3D. FIG. 3A shows a graph 302
depicting the time response of fluorescence intensity relative to
pre-AA (fluorescence prior to application of acetic acid) 304 of
NED tissue at 390, 460 and 600 nm wavelengths following application
of acetic acid. FIG. 3B shows a graph 320 depicting the time
response of reflectance relative to pre-AA 322 for NED tissue at
425, 500, and 630 nm wavelengths following application of acetic
acid. FIG. 3C shows a graph 350 depicting the time response of
fluorescence intensity relative to pre-AA 304 of CIN 2/3 tissue at
390, 460, and 600 nm wavelengths following application of acetic
acid. FIG. 3D shows a graph 370 depicting the time response of
reflectance relative to pre-AA 322 for CIN 2/3 tissue at 425, 500,
and 630 nm wavelengths following application of acetic acid.
[0049] The fluorescence intensity in the NED group continues to
drop over the time period studied while some recovery is seen in
the fluorescence intensity of the CIN 2/3 group. FIG. 3A reveals a
continuous drop in fluorescence for the NED group over the
measurement period at the three wavelengths. In contrast, FIG. 3C
shows partial recovery at all three wavelengths for CIN 2/3 tissue.
Each of the curves representing CIN 2/3 tissue labeled 352, 354,
and 356 in FIG. 3C demonstrates a generalized local minimum at a
time from about 70 to about 130 seconds following application of
acetic acid, whereas each of the curves representing NED tissue
labeled 310, 312, and 314 in FIG. 3A does not show such a local
minimum.
[0050] The fluorescence and reflectance kinetics are similar for
the CIN 2/3 group but differ for the NED group. Partial recovery
(return toward initial condition) is noted in both the reflectance
and the fluorescence curves at all 3 wavelengths for CIN 2/3
tissue, as shown in the curves labeled 352, 354, 356, 372, 374, and
376 in FIG. 3C and FIG. 3D. However, partial recovery is noted only
in the reflectance curves for NED tissue (curves 326, 328, and 330
of FIG. 3B), while the NED fluorescence intensities continue to
drop (curves 310, 312, and 314 of FIG. 3A).
[0051] The magnitude of change in the time response of reflectance
and fluorescence data following application of acetic acid is
different between the CIN 2/3 group and the NED group. The relative
maximum change in reflectivity at about 425 nm is about twice as
large for CIN 2/3 (i.e. line segment 274 in FIG. 2B) compared to
non-CIN (i.e. line segment 174 in FIG. 1B), while the maximum
change for fluorescence is approximately equivalent for CIN 2/3 and
non-CIN samples. Here, the magnitude of change in the reflectance
signal depends on tissue type while the magnitude of change in the
fluorescence signal does not depend on tissue type.
[0052] The time to reach the maximum change in fluorescence is
delayed for NED spectra. This is shown by comparing curves 310,
312, and 314 of FIG. 3A with curves 352, 354, and 356 of FIG. 3C.
It is therefore possible, for example, to use the time required to
reach a minimum value of fluorescence spectral intensity to
distinguish CIN 2/3 from NED samples.
[0053] The fluorescence line-shape changes with time post acetic
acid, particularly at later times where a valley at about 420 nm
and a band at about 510 nm become more distinct. The valley at
about 420 nm is shown in FIG. 1A at reference number 130 and in
FIG. 2A at reference number 226, while the band at about 510 nm can
be seen in FIG. 1A at reference number 132 and in FIG. 2A at
reference number 228. One explanation for this change is that
collagen and NADH decrease tissue fluorescence and FAD increases
tissue fluorescence. Upon introduction of a change in pH from 7 to
3.5, the fluorescence intensity of NADH decreases by a factor of
two while FAD increases six-fold. Increased scattering in the
epithelial layer would decrease the contribution of collagen
fluorescence from the submucosal layer. Characterization of such
changes in spectral curve shape is useful, for example, in
distinguishing tissue types.
[0054] In one embodiment, an optimal window for obtaining spectral
and/or image data is a period of time in which there is a peak
"whitening" as seen in image and/or reflectance data. In another
embodiment, an optimal window is a period of time in which there is
a peak "darkening" of fluorescence of the tissue. Still another
embodiment uses a subset of the union of the two optimal windows
above. FIGS. 1A, 1B, 2A, 2B, 3A, 3B, 3C, and 3D demonstrate
"whitening" of reflectance and "darkening" of fluorescence as a
function of wavelength and time following application of acetic
acid. The maximum change observed in the CIN 2/3 group is
determined from the data shown in FIGS. 2A, 2B, 3C, and 3D. Here,
the peak "darkening" of the fluorescence data lags peak "whitening"
of the reflectance data. From the reflectance data, the window for
peak whitening lies between about 30 s and about 110 s following
the application of acetic acid with a maximum at about 70 s. In one
embodiment, the peak whitening window lies between about 30 s and
about 130 s; and in another embodiment from about 20 s to about 180
s. For fluorescence, the peak "darkening" window lies between about
50 s and about 150 s with a minimum at about 80 s. In one
embodiment, the peak darkening window lies between about 60 s and
about 220 s. Peak "whitening" for the non-CIN reflectance spectra
is less intense but similar in shape to that found for CIN 2/3.
Peak darkening in non-CIN fluorescence appears later than in CIN
2/3 fluorescence.
[0055] FIGS. 4A and 4B depict the influence of acetic acid on
reflectance and fluorescence intensities at about 425 nm and about
460 nm, respectively, for various reference tissue classes. These
classes include CIN 2/3 (curves 406 and 454), CIN 1 (curves 408 and
456), metaplasia TT016 and TT017 (curves 410 and 458), normal
columnar TT022 (curves 412 and 460) and normal squamous TT025
(curves 414 and 462) tissues, as shown in FIGS. 4A and 4B. In
general, the reflectance curves of FIG. 4A show some distinct
differences with tissue type, with CGN 2/3 tissue (curve 406)
having the largest change. Columnar epithelial tissue (curve 412)
shows rapid relatively intense whitening followed by rapid recovery
while squamous epithelial tissue (curve 414) has a weak, slow
response with very little recovery. Metaplastic tissues (curve 410)
and tissue with CIN 1 (curve 408) have similar behavior with a
relatively fast increase and decay. The acetowhitening response of
all tissue groups ride on top of a slowing, increasing background,
thereby suggesting a secondary response to acetic acid. This
secondary response is most distinct in the CIN 1 group and appears
to be the predominant response in the normal squamous group.
[0056] The magnitude of the acetodarkening effect for fluorescence
is similar independent of tissue type, as shown in FIG. 4B. The
time to reach a minimum fluorescence is different for different
tissue classes, with normal squamous tissue (curve 462) having the
slowest response and normal columnar tissue (curve 460) having the
fastest response. The response for CIN 2/3 (curve 454), CIN 1
(curve 456), and metaplastic tissues (curve 458) are very similar.
There is partial recovery from the acetic acid effect in the CIN
2/3 group (curve 454).
EXAMPLE 3
Using a Discrimination Function to Determine Optimal Windows for
Obtaining Diagnostic Optical Data.
[0057] An embodiment of the invention comprises determining an
optimum window for obtaining diagnostic spectral data using
fluorescence and/or reflectance time-response data as shown in the
above figures, and as discussed above. In one embodiment, an
optimum window is determined by tracking the difference between
spectral data of various tissue types using a discrimination
function.
[0058] In one embodiment, the discrimination function shown below
in Equation (1) is used to extract differences between tissue
types: 1 D ( ) = ( test ( ) ) NEDPATH1 - ( test ( ) ) CIN23ALL 2 (
test ( ) ) NEDPATH1 + 2 ( test ( ) ) CIN23ALL ( 1 )
[0059] The quantity .mu. corresponds to the mean optical signal and
.sigma. corresponds to the standard deviation. In one embodiment,
the optical signal includes diffuse reflectance. In another
embodiment, the optical signal includes 337-nm fluorescence
emission spectra. Other embodiments use fluorescence emission
spectra at another excitation wavelength such as 380 nm and 460 nm.
In still other embodiments, the optical signal is a video signal,
Raman signal, or infrared signal. Some embodiments comprise using
difference spectra calculated between different phases of
acetowhitening, using various normalization schema, and/or using
various combinations of spectral data and/or image data as
discussed above.
[0060] One embodiment comprises developing linear discriminant
analysis models using spectra from each time bin as shown in Table
1. Alternatively, nonlinear discriminant analysis models may be
developed. Generally, models are trained using reflectance and
fluorescence data separately, although some embodiments comprise
use of both data types to train a model. In exemplary embodiments
discussed below, reflectance and fluorescence intensities are
down-sampled to one value every 10 nm between 360 and 720 nm. A
model is trained by adding and removing intensities in a forward
manner, continuously repeating the process until the model
converges such that additional intensities do not appreciably
improve tissue classification. Testing is performed by a
leave-one-spectrum-out jack-knife process.
[0061] FIG. 5 shows the difference between the mean reflectance
spectra for non-CIN 2/3 tissues (including CIN 1 and NED tissues)
and CIN 2/3 tissues at three times--at a time prior to the
application of acetic acid (graph 502), at a time corresponding to
maximum whitening (graph 520, about 60-80 seconds post-AA), and at
a time corresponding to the latest time period in which data was
obtained (graph 550, about 160-180 seconds post-AA). Here, the time
corresponding to maximum whitening was determined from reflectance
data, and occurs between about 60 seconds and 80 seconds following
application of acetic acid. In the absence of acetic acid, the
reflectance spectra for CIN 2/3 (curve 510 of graph 502 in FIG. 5)
are on average lower than non-CIN 2/3 tissue (curve 508 of graph
502 in FIG. 5). Following the application of acetic acid, a
reversal is noted with CIN 2/3 tissues having higher reflectance
than the other tissues. The reflectance of CIN 2/3 and non-CIN 2/3
tissues increase with acetic acid, with CIN 2/3 showing a larger
relative percent change (compare curves 522 and 524 of graph 520 in
FIG. 5). From about 160 s to about 180 s following acetic acid, the
reflectance of CIN 2/3 tissue begins to return to the pre-acetic
acid state, while the reflectance of the non-CIN 2/3 group
continues to increase (compare curves 552 and 554 of graph 550 in
FIG. 5)
[0062] In one embodiment, discrimination function `spectra` are
calculated from the reflectance spectra of CIN 2/3 and non-CIN 2/3
tissues shown in FIG. 5. In one example, discrimination function
spectra comprise values of the discrimination function in Equation
(1) determined as a function of wavelength for sets of spectral
data obtained at various times. FIG. 6 shows a graph 602 depicting
the discrimination function spectra evaluated using the diffuse
reflectance data of FIG. 5 obtained prior to application of acetic
acid, and at two times post-AA. Curve 608 corresponds to the
discrimination function 604 evaluated as a function of wavelength
606 using non-CIN 2/3 data and CIN 2/3 data obtained prior to
application of acetic acid. Curve 610 corresponds to the
discrimination function 604 evaluated as a function of wavelength
606 using non-CIN 2/3 data and CIN 2/3 data obtained between about
60 and about 80 seconds after application of acetic acid; and curve
612 corresponds to the discrimination function 604 evaluated as a
function of wavelength 606 using non-CIN 2/3 data and CIN 2/3 data
obtained between about 160 and about 180 seconds after application
of acetic acid. Distinguishing between CIN 2/3 and non-CIN 2/3
tissues using reflectance data is improved with the application of
acetic acid. Here, the largest differences (for example, the
largest absolute values of discrimination function) are found from
data measured from about 60 s to about 80 s post-acetic acid (curve
610), and these agree with the differences seen in the mean
reflectance spectra of FIG. 5 (curves 522 and 524 of graph 520 in
FIG. 5).
[0063] Performing multivariate linear regression analysis addresses
wavelength interdependencies in the development of a classification
model. An application of one embodiment comprises classifying data
represented in the CIN 2/3, CIN 1, and NED categories in the
Appendix Table into CIN 2/3 and non-CIN 2/3 categories by using
classification models developed from the reflectance data shown in
FIG. 5. Here, reflectance intensities are down-sampled to one about
every 10 nm between about 360 nm and about 720 nm. The model is
trained by adding intensities in a forward-stepped manner. Testing
is performed with a leave-one-spectrum-out jack-knife process. The
result of this analysis shows which wavelengths best separate CIN
2/3 from non-CIN 2/3, as shown in Table 2 for an exemplary
embodiment.
2TABLE 2 Forwarded selected best reflectance wavelengths for
classifying CIN 2/3 from non-CIN 2/3 spectra obtained at different
times pre and post-AA. Time from AA (s) LDA Model Input Wavelengths
(nm) Accuracy -30 370 400 420 440 530 570 590 610 66 30 420 430 450
600 74 50 360 400 420 430 580 600 74 70 360 370 420 430 560 580 600
77 90 360 420 430 540 590 73 110 360 440 530 540 590 71 130 360 420
430 540 590 71 150 370 400 430 440 540 620 660 690 720 72 170 490
530 570 630 650 75
[0064] The two best models for separating CIN 2/3 and non-CIN 2/3
for this embodiment include the model using reflectance data
obtained at peak CIN 2/3 whitening (from about 60 s to about 80 s)
and the model using reflectance data from the latest time measured
(from about 160 s to about 180 s post acetic acid). The first model
uses input wavelengths between about 360 and about 600 nm, while
the second model uses more red-shifted wavelengths between about
490 and about 650 nm. This is consistent with the behavior of the
discrimination function spectra shown in FIG. 6.
[0065] FIG. 7 demonstrates one method of determining an optimal
window for obtaining reflectance spectral data in the diagnosis of
the state of health of a region of a sample as CIN 2/3 or non-CIN
2/3. FIG. 7 shows a graph 702 depicting the performance of the two
LDA models described in Table 2 above as applied to reflectance
spectral data obtained at various times following application of
acetic acid 706. Curve 710 in FIG. 7 is a plot of the diagnostic
accuracy of the LDA model based on reflectance spectral data
obtained between about 60 and about 80 seconds ("peak whitening
model") as applied to reflectance spectra from the bins of Table 1,
and curve 712 in FIG. 7 is a plot of the diagnostic accuracy of the
LDA model based on reflectance spectral data obtained between about
160 and about 180 seconds, as applied to reflectance spectra from
the bins of Table 1. For the peak-whitening model, the highest
accuracy was obtained at about 70 s, while accuracies greater than
70% were obtained with spectra collected in a window between about
30 s and about 130 s. The 160-180 s model had a narrower window
around 70 s, but performs better at longer times.
[0066] FIG. 8 shows the difference between the mean 337-nm
fluorescence spectra for non-CIN 2/3 tissues (including CIN 1 and
NED tissues) and CIN 2/3 tissues at three times--at a time prior to
application of acetic acid (graph 802), at a time corresponding to
maximum whitening (graph 820, about 60 to about 80 seconds
post-AA), and at a time corresponding to the latest time period in
which data was obtained (graph 850, about 160 to about 180 seconds
post-AA). The time corresponding to maximum whitening was
determined from reflectance data, and occurs between about 60
seconds and 80 seconds following application of acetic acid. In the
absence of acetic acid, the fluorescence spectra for CIN 2/3 tissue
(curve 812 of graph 802 in FIG. 8) and for non-CIN 2/3 tissue
(curve 810 of graph 802 in FIG. 8) are essentially equivalent with
a slightly lower fluorescence noted around 390 nm for CIN 2/3
sites. Following the application of acetic acid, the fluorescence
of CIN 2/3 and non-CIN 2/3 tissues decrease, with CIN 2/3 showing a
larger relative percent change (compare curves 824 and 822 of graph
820 in FIG. 8). From about 160 s to about 180 s following acetic
acid application, the fluorescence of CIN 2/3 tissue shows signs of
returning to the pre-acetic acid state while the fluorescence of
the non-CIN 2/3 group continues to decrease (compare curves 854 and
852 of graph 850 in FIG. 8).
[0067] In one embodiment, discrimination function `spectra` are
calculated from the fluorescence spectra of CIN 2/3 and non-CIN 2/3
tissues shown in FIG. 8. In one example, discrimination function
spectra comprise values of the discrimination function in Equation
(1) determined as a function of wavelength for sets of spectral
data obtained at various times. FIG. 9 shows a graph 902 depicting
the discrimination function spectra evaluated using the
fluorescence data of FIG. 8 obtained prior to application of acetic
acid, and at two times post-AA. Curve 908 corresponds to the
discrimination function 904 evaluated as a function of wavelength
906 using non-CIN 2/3 data and CIN 2/3 data obtained prior to
application of acetic acid. Curve 910 corresponds to the
discrimination function 904 evaluated as a function of wavelength
906 using non-CIN 2/3 data and CIN 2/3 data obtained between about
60 and about 80 seconds after application of acetic acid; and curve
912 corresponds to the discrimination function 904 evaluated as a
function of wavelength 906 using non-CIN 2/3 data and CIN 2/3 data
obtained between about 160 and about 180 seconds after application
of acetic acid. Distinguishing between CIN 2/3 and non-CIN 2/3
tissues using fluorescence data is improved with the application of
acetic acid. Here, the largest absolute values are found from data
measured within the range of about 160-180 s post-acetic acid
(curve 912), and these agree with the differences seen in the mean
fluorescence spectra of FIG. 8 (curves 852 and 854 of graph 850 in
FIG. 8).
[0068] Performing multivariate linear regression analysis addresses
wavelength interdependencies in the development of a classification
model. An application of one embodiment comprises classifying data
represented in the CIN 2/3, CIN 1, and NED categories in the
Appendix Table into CIN 2/3 and non-CIN 2/3 categories by using
classification models developed from the fluorescence data shown in
FIG. 8. Fluorescence intensities are down-sampled to one about
every 10 nm between about 360 and about 720 nm. The model is
trained by adding intensities in a forward manner. Testing is
performed by a leave-one-spectrum-out jack-knife process. The
result of this analysis shows which wavelengths best separate CIN
2/3 from non-CIN 2/3, as shown in Table 3 for an exemplary
embodiment.
3TABLE 3 Forwarded selected best 337-nm fluorescence wavelengths
for classifying CIN 2/3 from non-CIN 2/3 spectra obtained at
different times pre and post-AA. Time from AA (s) LDA Model Input
Wavelengths (nm) Accuracy -30 380, 430, 440, 610, 660, 700, 710 61
30 370, 380, 390, 640 61 50 410 54 70 360, 390, 490, 580, 590, 670
63 90 370, 380, 420, 460, 500, 560, 660 64 110 360, 390, 400, 710
51 130 370 53 150 370, 380, 440, 620, 640, 700 65 170 370, 480,
510, 570, 600, 700, 720 76
[0069] The two best models for separating CIN 2/3 and non-CIN 2/3
for this embodiment include the models using data obtained at peak
CIN 2/3 whitening (60-80 s) and the model using data at the latest
time measured (160-180 s post acetic acid). The first model uses
input wavelengths between about 360 and about 670 nm, while the
second model uses wavelengths between about 370 and about 720 nm.
This is consistent with the discrimination function spectra shown
in FIG. 9.
[0070] FIG. 10 demonstrates one method of determining an optimal
window for obtaining fluorescence spectral data in the diagnosis of
the state of health of a region of a sample as CIN 2/3 or non-CIN
2/3. FIG. 10 shows a graph 1002 depicting the performance of the
two LDA models described in Table 3 above as applied to
fluorescence spectral data obtained at various times following
application of acetic acid 1006. Curve 1010 in FIG. 10 is a plot of
the diagnostic accuracy of the LDA model based on fluorescence
spectral data obtained between about 60 and about 80 seconds ("peak
whitening model") as applied to fluorescence spectra from the bins
of Table 1, and curve 1012 in FIG. 10 is a plot of the diagnostic
accuracy of the LDA model based on fluorescence spectral data
obtained between about 160 and about 180 seconds, as applied to
fluorescence spectra from the bins of Table 1. The accuracies of
these models vary depending on when the fluorescence spectra are
recorded relative to the application of acetic acid, as shown in
FIG. 10. The predictive ability of the fluorescence models in FIG.
10 tend to be less than that of the reflectance models in FIG. 7.
Accuracies greater than 70% are obtained with spectra collected
after about 160 seconds post-AA.
[0071] Another embodiment comprises classifying data represented in
the CIN 2/3, CIN 1, and NED categories in the Appendix Table into
CIN 2/3 and non-CIN 2/3 categories by using fluorescence divided by
diffuse reflectance spectra. Models are developed based on time
post acetic acid. Ratios of fluorescence to reflectance are
down-sampled to one every 10 nm between about 360 and about 720 nm.
The model is trained by adding intensities in a forward manner.
Testing is performed by a leave-one-spectrum-out jack-knife
process. For this analysis, the model is based on intensities at
about 360, 400, 420, 430, 560, 610, and 630 nm. In general, the
results are slightly better than a model based on fluorescence
alone. Improved performance is noted from spectra acquired at about
160 s post acetic acid.
[0072] FIG. 11 shows a graph 1102 depicting the accuracy of three
LDA models as applied to spectral data obtained at various times
following application of acetic acid. Curve 1110 in FIG. 11 is a
plot of the diagnostic accuracy of the LDA model based on
reflectance spectral data obtained between about 60 and about 80
seconds ("peak whitening model"), also shown as curve 710 in FIG.
7. Curve 1112 in FIG. 11 is a plot of the diagnostic accuracy of
the LDA model based on fluorescence spectral data obtained between
about 60 and about 80 seconds ("peak whitening model"), also shown
as curve 1010 in FIG. 10. Curve 1114 in FIG. 11 is a plot of the
diagnostic accuracy of the LDA model based on fluorescence
intensity divided by reflectance, as described in the immediately
preceding paragraph.
[0073] The exemplary embodiments discussed above demonstrate that
the ability to distinguish between non-CIN 2/3 and CIN 2/3
fluorescence and reflectance spectra is improved with the
application of acetic acid or other contrast agent. For the
peak-whitening LDA model using reflectance data, the highest
accuracy for the exemplary applications of the embodiments
discussed herein is obtained at about 70 s following introduction
of acetic acid, while accuracies greater than about 70% are
obtained with spectra collected in a window between about 30 s and
about 130 s. The predictive ability of the fluorescence models in
the examples above tend to be less than that of the reflectance
models for the examples discussed above. Accuracies greater than
70% are obtained with fluorescence at times greater than about 160
s post acetic acid. The intensity of fluorescence continuously drop
over the measurement period in the non-CIN groups while partial
recovery occurs at all 3 emission wavelengths in the CIN 2/3 group,
suggesting that fluorescence spectral data obtained at times
greater than about 180 s is useful in diagnosing CIN 2/3.
EXAMPLE 4
Other Kinetics-based Approaches for Obtaining Diagnostic Optical
Data within an Optimal Window
[0074] As an alternative to the techniques discussed above, other
kinetics-based approaches may be used to determine classification
models and, hence, corresponding optimum windows for classification
of tissue samples. The time response of fluorescence intensity or
the time response of reflectance following application of contrast
agent, as shown in FIG. 3 and FIG. 4, may be curve-fitted to
determine one or more parameters sensitive to a curve feature of
interest. For example, a parameter sensitive to a local minimum may
be determined for a given set of fluorescence response data. In one
embodiment, a parameter is determined by curve-fitting fluorescence
time response data to a sigmoidal function. Values of the parameter
and/or goodness-of-fit data are then used to develop a statistical
model for classifying a sample in terms of a characteristic of the
sample, such as its state of health. The model is built using
reference data with known states of health. Then, the time response
of spectral intensity of a test sample with unknown state of health
following application of a contrast agent is obtained. By
curve-fitting this response data, values of the indicated
parameter(s) may be obtained, and the model may be used to either
directly determine the characteristic of the test sample, or to
indicate an optimal window in which spectral data should be
obtained and used to accurately classify the tissue. In one
embodiment, the parameter determined by curve-fitting spectral time
response curves is not used directly to classify the tissue, but is
used to determine an optimal window. The parameter indicates a
window of time in which one or more complete sets of spectral
and/or video data should be obtained for accurate diagnosis of the
tissue.
EXAMPLE 5
Using a Relative Change or Rate-of-change Trigger to Obtain
Diagnostic Optical Data
[0075] An embodiment of the invention comprises determining and
using a relative amplitude change and/or rate of amplitude change
as a trigger for obtaining diagnostic optical data from a sample.
The trigger can also be used to determine an optical window of time
for obtaining such diagnostic optical data. By using statistical
and/or heuristic methods such as those discussed herein, it is
possible to relate more easily-monitored relative changes or
rates-of-change of one or more optical signals from a tissue sample
to corresponding full spectrum signals that can be used in
characterizing the state of health of a given sample. For example,
by performing a discrimination function analysis, it may be found
for a given tissue type that when the relative change in
reflectance at a particular wavelength exceeds a threshold value,
the corresponding full-spectrum reflectance can be obtained and
then used to accurately classify the state of health of the tissue.
In addition, the triggers determined above may be converted into
optimal time windows for obtaining diagnostic optical data from a
sample.
[0076] FIG. 12A shows how an optical amplitude trigger can be used
to determine an optimal time window for obtaining diagnostic
optical data. The graph 1200 in FIG. 12A plots the normalized
relative change of mean reflectance signal 1202 from tissue samples
with a given state of health as a function of time following
application of acetic acid 1204. The mean reflectance signal
determined from CIN 1, CIN 2, and Metaplasia samples are depicted
in FIG. 12A by curves 1210, 1208, and 1212, respectively. Here, it
was determined that when the normalized relative change of mean
reflectance reaches or exceeds 0.75, the image intensity data
and/or the full reflectance and/or fluorescence spectrum for a
given sample is most indicative of a given state of health of a
sample. Thus, for CIN 2 samples, for example, this corresponds to a
time period between t.sub.1 and t.sub.2, as shown in the graph 1200
of FIG. 12A. Therefore, spectral and/or image data obtained from a
tissue sample between t.sub.1 and t.sub.2 seconds following
application of acetic acid can be used in accurately determining
whether or not CIN 2 is indicated for that sample. In one
embodiment, the relative change of reflectance of a tissue sample
at one or more given wavelengths is monitored, and when that
relative change is greater than or equal to the 0.75 threshold,
more comprehensive spectral and/or image data is obtained from the
sample for purposes of characterizing whether or not the sample is
indicative of CIN 2. FIG. 12A demonstrates the use of a threshold
value of relative optical signal change. In another embodiment, a
predetermined range of values of the relative optical signal change
is used such that when the relative signal change falls within the
predetermined range of values, additional spectral and/or image
data is captured in order to characterize the sample.
[0077] FIG. 12B shows how a rate-of-change of optical amplitude
trigger can be used to determine an optimal time window for
obtaining diagnostic optical data. The graph 1250 of FIG. 12B plots
the slope of mean reflectance signal 1252 from tissue samples with
a given state of health as a function of time following application
of acetic acid 1204. The slope of mean reflectance is a measure of
the rate of change of the mean reflectance signal. The rate of
change of mean reflectance determined from CIN 1, CIN 2, and
Metaplasia samples are depicted in FIG. 12B by curves 1258, 1256,
and 1260, respectively. Here, it was determined that when the
absolute value of the slope has an absolute value less than or
equal to 0.1, for example, in the vicinity of maximum reflectance,
the image intensity data and/or the full reflectance and/or
fluorescence spectrum for a given sample is most indicative of a
given state of health of a sample. Thus, for CIN 2 samples, for
example, this corresponds to a time period between t.sub.1 and
t.sub.2 as shown in the graph 1250 of FIG. 12B. Therefore, spectral
and/or image data obtained from a tissue sample between t.sub.1 and
t.sub.2 seconds following application of acetic acid can be used in
accurately determining whether or not CIN 2 is indicated for that
sample. In one embodiment, the rate of change of reflectance of a
tissue sample is monitored at one or more given wavelengths, and
when that rate of change has an absolute value less than or equal
to 0.1, more comprehensive spectral and/or image data is obtained
from the sample for purposes of characterizing whether or not the
sample is indicative of CIN 2. FIG. 12B demonstrates use of a range
of values of rate of optical signal change. Other embodiments use a
single threshold value.
EXAMPLE 6
Using Fluorescence, Reflectance and/or Image Time Response Data to
Diagnose Regions of Tissue
[0078] The figures discussed herein include time-response
fluorescence and reflectance data obtained following application of
a contrast agent to tissue. In addition to an acetowhitening effect
observed in the reflectance data, an "acetodarkening" effect is
observed in the fluorescence data. For example, the fluorescence
intensity of diseased regions decreases to a minimum at about 70 s
to about 130 s following application of acetic acid. Thus, the
presence of a minimum fluorescence intensity within this window of
time, as well as the accompanying increase in fluorescence from
this minimum, may be used to indicate disease. An embodiment of the
invention comprises a method of identifying a characteristic of a
region of a tissue sample including applying a contrast agent to a
region of a tissue sample, obtaining at least two values of
fluorescence spectral intensity corresponding to the region,
determining whether the fluorescence spectral intensity
corresponding to the region increases after a predetermined time
following the applying step, and identifying a characteristic of
the region based at least in part on the determining step. In an
embodiment, the obtaining step comprises obtaining a fluorescence
spectral intensity signal corresponding to the region as a function
of time following the applying step. In an embodiment, the method
further comprises determining whether the fluorescence spectral
intensity corresponding to the region decreases following the
applying step, then increases after the predetermined time. In an
embodiment, the predetermined time is about 80 seconds.
[0079] An embodiment comprises a method of identifying a
characteristic of a region of a tissue sample comprising applying a
contrast agent to a region of a tissue sample, obtaining a
fluorescence spectral intensity signal from the region of the
tissue sample, determining an elapsed time following the applying
step at which the fluorescence spectral intensity signal has a
minimum value, and identifying a characteristic of the region based
at least in part on the elapsed time.
[0080] An embodiment comprises a method of identifying a
characteristic of a region of a tissue sample comprising applying a
contrast agent to a region of a tissue sample, obtaining a
reflectance signal from the region of the tissue sample,
determining a change in reflectance spectral intensity
corresponding to the region of the tissue sample following the
applying step, and identifying a characteristic of the region based
at least in part on the change in reflectance spectral intensity.
In an embodiment, the change in reflectance spectral intensity
corresponding to the region comprises a change relative to an
initial condition of the region.
[0081] An embodiment comprises a method of identifying a
characteristic of a region of a tissue sample comprising applying a
contrast agent to a region of a tissue sample, obtaining an optical
signal from the region of the tissue sample, determining a rate of
change of the optical signal corresponding to the region of the
tissue sample, and identifying a characteristic of the region based
at least in part on the rate of change. In an embodiment, the
optical signal comprises fluorescence spectral intensity at a given
wavelength. In an embodiment, the optical signal comprises
reflectance spectral intensity at a given wavelength.
[0082] An embodiment comprises a method of identifying a
characteristic of a region of a tissue sample comprising applying a
contrast agent to a region of a tissue sample, obtaining a
fluorescence signal from the region of the tissue sample, obtaining
a reflectance signal from the region of the tissue sample, and
identifying a characteristic of the region based at least in part
on the fluorescence signal and the reflectance signal.
[0083] An embodiment comprises obtaining an optical signal from 499
regions, each region having a diameter of approximately 1 mm,
covering an area of tissue about 25 mm in diameter. An embodiment
may also comprise obtaining a video image of about 480 by about 560
pixels covering the same 25-mm diameter area of tissue.
4APPENDIX TABLE Number of spectra (number of subjects) for each
tissue class in each time bin for exemplary embodiments discussed
herein. Time CIN 2/3 CIN 1 Metaplasia TT_022.sup.1 TT_025.sup.1
NEDPath1.sup.1 t .ltoreq. 0 451 (62) 202 (46) 329 (77) 202 (56) 294
(70) 816 (186) 0 < t .ltoreq. 40 118 (21) 72 (14) 147 (33) 51
(14) 113 (22) 307 (64) 40 < t .ltoreq. 60 300 (47) 135 (31) 255
(58) 116 (32) 230 (51) 597 (133) 60 < t .ltoreq. 80 375 (54) 162
(39) 300 (68) 179 (42) 262 (61) 731 (157) 80 < t .ltoreq. 100
455 (60) 195 (42) 308 (70) 190 (49) 263 (64) 752 (167) 100 < t
.ltoreq. 120 446 (60) 209 (45) 328 (76) 208 (52) 284 (68) 811 (178)
120 < t .ltoreq. 140 303 (44) 135 (30) 200 (48) 165 (43) 185
(51) 545 (129) 140 < t .ltoreq. 160 130 (18) 82 (17) 75 (19) 96
(23) 66 (21) 232 (53) 160 < t .ltoreq. 180 53 (9) 50 (9) 34 (9)
38 (12) 19 (6) 91 (24) t > 180 14 (3) 26 (3) 33 (6) 23 (6) 30
(5) 86 (15) .sup.1TT 022 = Normal columnar tissue; TT 025 = Normal
squamous tissue; NEDPath1 = NED = Metaplasia, TT_022, and
TT_025.
Equivalents
[0084] While the invention has been particularly shown and
described with reference to specific preferred embodiments, it
should be understood by those skilled in the art that various
changes in form and detail may be made therein without departing
from the spirit and scope of the invention as defined by the
appended claims.
* * * * *