raman and ir spectroscopy - university of illinois at urbana-champaign

6/8/2009

1

Beckman InstituteBeckman InstituteBeckman Instituteat The University of Illinois

Vibrational Spectroscopy Vibrational Spectroscopy and Imaging

Rohit Bhargava

Spectroscopy generalization

Known radiation goes in Collect exiting radiation

Compare the light that went in with the light that comes out

A general definition of spectroscopy is the study of the interaction between radiation and an analyte as a function of wavelength.

Interacts with sample

Spectroscopy (Radiation)

Radio

1010-108 cm-1: Re-arrangement of elementary particles (gamma ray)108-106 cm-1: Transitions of inner electrons (X-ray)106-104 cm-1: Transitions of valence electrons (UV-vis spectroscopy)104-102 cm-1: Transitions between vibrational levels (IR and Raman)102-100 cm-1: Transitions between rotational levels (Microwave, IR, Raman, THz)100-10-2 cm-1: Transitions between electron spin levels (magnetic – ESR)10-2-10-4 cm-1: Transitions between electron nuclear spin levels (magnetic – NMR)

Cosmic Gamma X

UV IR Micro UHF

Short

Medium

Long

Ultra violet Infrared

Near Mid Far

1 400 750 2,500 16,000 1,000,000 nm

Vis

6/8/2009

2

Spectroscopy of liquid water

200 250 300 350 4000

0.01

0.02

0.03

0.04

0.05

0.06

0.07

400 500 600 7000

0.01

0.02

0.03

0.04

0.05

0.06

0.07

1000 1500 2000 2500

20

40

60

80

100

120

800 1000 1200

0

0.5

1

1.5

2

2.5

4000 6000 8000 10000 12000 14000 160000

2000

4000

6000

8000

10000

12000

14000UV Spectrum

Wavelength (nm)

Abs

orpt

ion

(1/c

m)

Visible Spectrum NIR Spectrum IR Spectrum


Near Mid Far

1 400 750 2,500 16,000 1,000,000 nm

Vis

Wavelength (nm)

Why is water blue?

Gas Assignment liquid

3651 v1 3400 3755 v3

5332 v2+v3 51507251 v1+v3 69008807 v1+v2+v3 840010613 2v1+v3 10300

•Gas and liquid have different positions

•Combinations and overtones are weaker

• Significant example of absorption determining color10613 2v1+v3 10300

13831 3v1+v3 13160 (760 nm)

14319 v1+3v3 13510 (740 nm)

15348 3v1+v2+v3 15150 (660 nm)

15832 v1+v2+3v3 15150 (660 nm)

16822 3v3+2v2+v1 weak

determining color

•What is the effect of cold water? Ice?

So, what are these vibrational modes

Symmetricalstretching

Antisymmetricalstretching

Scissoring Rocking Wagging Twisting

Simplest Model of a Diatomic Molecule: beads and spring harmonic oscillatorSimplest Model of a Diatomic Molecule: beads and spring harmonic oscillator

- Hooke’s Law Restoring force is proportional to displacement (K~105 dynes- Newton’s law: force if mass x acceleration

012

~ 1012-1014 Hz

6/8/2009

3

Spectroscopy : Molecular Basis

: 0 2 0

: 0 2 0

: 0 2

: 00

0 …….

0 0 2 012 0 0

0 0 2 0 2 0

Raman Scattering

:

Note: - change in polarizability, excitation frequency and shift by vibrational frequency- Two types of shifts to lower (Stokes) and higher (anti-Stokes): which population is favored?

Raman Spectroscopy: Basic Concept

785

Detector

First observed by Sir C. Venkata Raman in 1928 using sunlight and photographic filters (won Nobel price in physics in 1930). Nobelprize.org


Near Mid Far

1 400 750 2,500 16,000 1,000,000 nm

Vis

785nm Sample


Electronic States

UV-VISAbsorption

http://en.wikipedia.org/wiki/File:Raman_energy_levels.jpg

Chance of occurring is Wavelength DependantResonance Raman Scattering?

6/8/2009

4


Electronic States

UV-VISAbsorption


If light is not absorbed… the majority of the photons pass through the sample by means of Rayleigh Scattering


Electronic States

UV-VISAbsorption


Some of the photons (~1/10,000,000) that are not absorbed will pass through the sample and undergo Stokes Raman Scattering


Electronic States

UV-VISAbsorption


Even fewer photons (temperature dependant) that are not absorbed will pass through the sample and undergo Anti-Stokes Raman Scattering

6/8/2009

5


Molecules in the sample

785 nm

Photon 786nm787nm788nm789nm790nm791nm792nm

780nm781nm782nm783nm784nm

… …


Near Mid Far

1 400 750 2,500 16,000 1,000,000 nm

Vis

785nm Sample

Detector

Detector Array

793nm794nm795nm796nm797nm798nm799nm800nm801nm802nm803nm804nm805nm806nm… …


Molecule in the sample

786nm787nm788nm789nm790nm791nm792nm

93


… …


Near Mid Far

1 400 750 2,500 16,000 1,000,000 nm

Vis

785nm Sample

Detector

Detector Array



93



785 nm LightGets rejected


… …

Rayleigh Scatter

Detector Array



Near Mid Far

1 400 750 2,500 16,000 1,000,000 nm

Vis

785nm Sample

Detector

6/8/2009

6



785 nm



… …


Near Mid Far

1 400 750 2,500 16,000 1,000,000 nm

Vis

785nm Sample

Detector

Detector Array





93


… …


Near Mid Far

1 400 750 2,500 16,000 1,000,000 nm

Vis

785nm Sample

Detector

Detector Array



93



781 nm LightPassed through to the detector


… …

Anti-stokes Raman

Detector Array



Near Mid Far

1 400 750 2,500 16,000 1,000,000 nm

Vis

785nm Sample

Detector

6/8/2009

7



785 nm



… …


Near Mid Far

1 400 750 2,500 16,000 1,000,000 nm

Vis

785nm Sample

Detector

Detector Array





93


… …


Near Mid Far

1 400 750 2,500 16,000 1,000,000 nm

Vis

785nm Sample

Detector

Detector Array



93



790 nm LightPassed through to the detector


… …

Stokes Raman

Detector Array



Near Mid Far

1 400 750 2,500 16,000 1,000,000 nm

Vis

785nm Sample

Detector

6/8/2009

8

786nm787nm788nm789nm790nm791nm792nm793nm794nm795nm796nm797nm798nm799nm800nm801nm

Billions of 785nm photonsMolecules in a soybean



Near Mid Far

1 400 750 2,500 16,000 1,000,000 nm

Vis

785nm

Detector

Detector Array

801nm802nm803nm804nm805nm806nm… …

Soybean

5

6

7

8

9

nten

sity

(A.U

.)

786nm787nm788nm789nm790nm791nm792nm793nm794nm795nm796nm797nm798nm799nm800nm801


Detector A

rray

786nm787nm788nm789nm790nm791nm792nm793nm794nm795nm796nm797nm798nm799nm800nm801nm802nm803nm804nm805nm806nm

…

…

400 600 800 1000 1200 1400 1600 18000

1

2

3

4

Raman Shift (cm-1)

Ram

an I

n

810nm 915nm

Detector Array

801nm802nm803nm804nm805nm806nm… …

5

6

7

8

9

Raman Spectroscopy: Raman Shift

785nm = 0.0000785cm

1

0.0000785cm= 12738.9 cm-1

12738.9 cm-1 – 400 cm-1 = 12338.9 cm-1

Excitation Frequency

Wavelength of 400 cm-1

Raman shift in nm

Soybean

400 600 800 1000 1200 1400 1600 18000

1

2

3

4

Raman Shift (cm-1)810nm 915nm

1

12338.9 cm-1= 0.0000810cm

810nm

***The Raman shift is independent of excitation frequency… the spectral band positions will be the same whether you excite with a 785nm laser or a 532 nm laser

6/8/2009

9

Raman Spectroscopy: Raman Shift

Important advantage of Raman… wavelength select ability to avoid light absorption and fluorescence… however

Wavelength Dependence of Raman Cross-Section

the higher the wavelength the lower the Raman efficiency (Raman Cross Section).

3400 cm-1 O-H stretch of liquid water

Appl. Opt. 36, 2686-2688 (1997)

Raman Spectroscopy: Selection Rules

Raman Bands result from the Vibrations of specific chemical Bonds.

Soybean band assignments:

800-900 C-C Backbone

1003 Phe Ring breathing

Wave number Molecule Assignment

1010-1200 C-C Backbone

1265 RC=ONH2 Amide III

1450 CH2 Stretch

1660 RCH=CHR Cis

1670 RCH=CHR Trans

1750 RC=OOR Ester

NIR Spectra

Raman Spectroscopy: Examples

4

5

6

7

8

9

n In

tens

ity

(A.U

.)

Raman Spectrum

400 600 800 1000 1200 1400 1600 18000

1

2

3

4

Raman Shift (cm-1)

Ram

an

810nm 915nm

IR Spectrum

2500nm 11,000nm

6/8/2009

10

Raman Spectroscopy: Instrumentation

Las

CCD

Kaiser Optical

ser


grap

h

-This is a axial dispersive spectrograph

-Reflective spectrographs are also common

-Interferometers are used for FT-Raman

Kaiser Optical

Spec

tro


ctor

-Charged-coupled devices (CCDs) are the most common detectors(256x1024 pixel array)

-FT Raman is used for frequencies where CCDs have low quantum efficiency

Det

ec

Spectral Dimension

Hei

ght

6/8/2009

11

Raman Spectroscopy: Advantages and Disadvantages

Advantages:

-Little sample preparation (Polishing and fixing to a slide is common)

-Not sensitive to water (Good for biological samples)

-High chemical specificity (Narrow spectral bands)

-Qualitative and Quantitative information

-Non-destructive (A measurement does not chemically or physically change the sample)

-Can take measurements on solids, liquids, or gases

-Measurements are taken without touching the sample (Remote sensing)

-Easily coupled with fiber-optics

Raman Spectroscopy: Advantages and Disadvantages

Disadvantages

-Acquisition times tend to be longer than other techniques (real time measurements have been demonstrated… but something like video rate imaging is not yet a reality)

-Raman signal tends to be weak

-Raman signal is often mixed with a fluorescent background signal, which can make signal processing difficult.

-High laser powers and burn delicate samples

Raman Microscopy

Spectrograph & CCD

785 nm Laser

Dichroic mirror

Microscope

1.2

)

0

0.2

0.4

0.6

0.8

1

307

342

376

411

446

480

515

549

584

619

653

688

722

757

792

826

861

895

930

965

999

1034

1068

1103

1138

1172

1207

Wavelengh (nm)

Tran

smitt

ance

(%)

Wavelength (nm)

6/8/2009

12

Confocal Raman Spectroscopy

Confocal Raman Microscopy

Tabaskblat et al., Appl. Spec 1992

Hyperspectral Raman Mapping

Spectrograph & CCD

785 nm Laser

Fram

e G

rabb

er

Microscope

PhAT Probe

~~~~Fiber Launcher

Illumination Collection

~~~~ 1

50

Collection fibers are transposed into a line when to enter the spectrograph

White Light Illuminated Region

Bone

PMMA50μm

Hyperspectral Raman Mapping

1

2

3

4

5

6

78

9

10

11 1213

14

15

16 17

18

1920

21

22

23

24

2526

27

2829

30

31

32

3334

35

36

3738

39

40

4142

43

44

45

4647

48

49

50

Raman map

Raman Shift (cm-1)

Ram

an In

tens

ity (A

.U.)

Raman Shift (cm-1)

Ram

an In

tens

ity (A

.U.)

6/8/2009

13

Spatially Offset Raman Spectroscopy

1

2

3

4

5

6

7 8

9

10

11

12

13

14

15

16

17

18

19

20

2122

2324

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

4041

42

43

44

45

46 47

48

49

50

ID

OD

Illumination ring

Spacing

Spectrograph& CCD

785 nm Laser

Fiber

Probe

Converging

Collection Fibers

Collection Fibers

g

launcher

Sample/Specimen

Fiber collim

ator

Axicon

Converging lens

Diverging

lens(es)

Dichroicmirror

Converging lens

Illumination

Collection

Spatially Offset Raman Spectroscopy

ensi

ty (A

.U.) Transcutaneous

Exposed boneBone factor

Anesthesia

Wat

er c

oolin

g

Schulmerich et. al., Appl. Spectrosc., 63(3) 286-295

Ram

an In

te

Raman Shift (cm-1)

Other Raman Techniques

-Surfaced enhanced Raman spectroscopy (SERS)

-Tip enhanced Raman spectroscopy (TERS)

-Resonance Raman spectroscopy

-Raman tomography

-Raman imaging

-Coherent anti-stokes Raman spectroscopy (CARS)

-Stimulated Raman spectroscopy

-Others I can’t think of at the moment and much more to come!

6/8/2009

14

Infrared Spectroscopy

Theory of Infrared – Chemical bonds

• Infrared spectroscopy : Principle: chemical bonds rotate or vibrate at specific frequencies (Group theory basis of IR Spectroscopy)

• Practical AspectsThe frequency at which resonance occurs depends on the properties of the chemical bondFactors include shape of molecular bond, energy levels and mass of atoms.Chemical bonds can be divided into those which are IR active and those which are IR inactive

Molecular Basis of IR Absorption

• Direct absorption of light

• Beer’s law relates absorption to concentration0 log 0

Δ

C‐Hstretching C‐H

bending

Electronic States

UV-VISAbsorption

• Selection rule– Dipole moment (symmetrical mode: diatom)

Increasing wavenumber (energy, frequency)

Increasing absorption of IR radiation

Increasing wavelength

C‐Cbending

6/8/2009

15

• To be an IR active mode, the motion must have a change in the electric dipole moment of the bond

• ImplicationsThe intensity of absorbance peaks is related to size of dipole moment.Bonds with higher dipole moments tend to be covalent bonds with highly different electronegativities e.g C=O

Theory of Infrared – Chemical bonds

highly different electronegativities e.g C O Symmetrical bonds are typically IR inactive e.g. C-C, N-NThe same chemical bond can also have multiple modes of vibration e.g. phosphate has a symmetrical and antisymmetrical modes

Infrared Spectra• Vibrational frequencies are directly resonant with optical frequencies – vibrational transition is energy matching

Infrared Spectroscopy

• IR was discovered in 1801 by William Herschel who split the EM spectrum using a prism – noted increase in temperature beyond the red part of visible spectrum

• 1930’s began to be exploited for spectroscopy studies

• Two mains types of spectrometer – Dispersive and Fourier TransformDi i h i f li h d h f• Dispersive uses monochromatic source of light and changes frequency over time by moving grating, mirror or detector. Originally used prisms and later grating.

• Largely replaced by Fourier Transform – Allows for all frequencies to be measured at once: Fellgetts advantage – No slits: higher throughput

6/8/2009

16

IR Spectroscopy: Theory

1 13

2 1

0.5 1 2∞

∞

2 and 2’

3

∞

0 0.5 1 2∞

∞

or

Detector

Sample1 2

3

4Focusing mirror

4000 3200 2400 1600 800 0

1.2

1.4

1.6

1.8

2.0

2.2

Retardation (cm)

I0

0.0

0.5

1.0

1.5

2.0

2.5

BD,0

-0.05 0.05 0.1 0.15 0.2 0.25

1

Cb

W

Computation (FT)

0.5 1 2∞

4or

,0 0.5 2∞

∞

, 0.5 2∞

∞

, 2 , 2∞

0

Moving mirror

Beamsplitter

Fixed mirror

BroadbandSource δ= 0

δ= δmax

2’

Wavenumber (cm-1)Cb:Centerburst, W:Wings

4000 3500 3000 2500 2000 1500 100010

100

1000

10000

Abs

orpt

ion

Coe

ff(c

m-1

)

Wavenumber (cm-1)

WaterTissue

A popular commercial instrument

FTIR Microscope

combining the FTIR spectrometer with a microscope

Point spectra or Mapping/Imaging?

A) Single Point Detector(mapping)

B) Linear Array Detector(imaging)

C) Focal Plane Array(imaging)

Example improvements in speed1 16 16,384

Single point 16 detector linear array 128x128 Focal Plane Array

In fact, Imaging is even quicker and allows for diffraction limit measurements

6/8/2009

17

FT-IR spectroscopy Imaging

0.4

1.6

100 μm

0.6

a.u.

)

y

• Typical characteristics– Wavelengths (2048 elements over 2.5 – 12.5 μm), x, y typically ~1024

• Computation is essential to recover data– Manual examination is prohibitive

• Trade-offs: spatial coverage vs. resolution, spectral resolution vs. signal-to-noise ratio, time vs. data quality/size vs. information

3200 2800 2400 2000 16000.0

0.2

0.4

Abs

orba

nce

(a

Wavenumber (cm-1)

x

z

FTIR Chemical Imaging

Sample

Aperture

Aperture

CCD Visible Detector

Mirror

Stage

10 -310 -2

10 -1

11010 2

10 3

10 4

10 5

MAPPING 3

1

2

Tim

e (

hrs

)

A

4

5 7a

6

8910

11

7b

8 cm-1 , SNR of 1000:1, 1 Mpix, 6 x6 micron res., 1 cm x 1 cmBlue – hardware, red – interferometry/software, green – emerging technology

Microscope

Mirror

Visible LightSource

Rapid-ScanInterferometer

1. E.N. Lewis et al. Anal. Chem. 67, 3377-3386 (1995) ; 2. Snively et al. Appl. Spectrosc. (1998); 3. Bhargava et al. Appl. Spectrosc. 53, 1313-1322 (1999); 4. Snively et al. Opt. Lett., 24; 1841-1843 (1999); 5. Bhargava et al. Appl. Spectrosc. 54, 486-495 (2000); 6. Perkin-Elmer Inc.; 7a. Varian Inc. ; 8. NIH-FBI camera; 9. Reddy et al (To be submitted); 10 & 11 – under development.

1992 1996 2000 2004 2008

10 -4

Year

Settings and Parameters

• Sample preparation is critical!• Signal to Noise – Number of co-additions

A number of important considerations must be taken into account prior to experiments;

• Mirror speed• Spectral resolution• Spatial resolution• Point spectra vs. Point Mapping vs. Imaging

6/8/2009

18

Sample preparation• Transmission or reflection

• IR transparent substrates – BaF2, CaF2• IR reflective substrates – Gold, MirrIR Slides• Reflection is low-cost, however reduction in spectral quality and

increase in scattering artifacts

Signal to noise• It is important to get high quality noise-free data –

This can involve multiple strategies;– Better sample preparation (more or less sample)– More scans and averaging– Reduce Michelson mirror speed

• Spectral resolution is important this is the number of• Spectral resolution is important, this is the number of points within a spectra, the fewer the faster the scans but potential for loss of information

• Spatial resolution is also important, the higher the spatial resolution, the lower the signal.Take home message; Many factors must be adapted depending on what

you are interested in

Application 1: Forensic Sciences

0.005

Absorbance at 2920 cm-1

Body oils

1 mm

0.000

Absorbance at 1568 cm-1

Skin Flakes

6/8/2009

19

An example of imaging…..

• Polymer crystallization….

900 1000 1100 1200 1300 1400 1500

-0.50

-0.25

0.00

0.25

0.50

0.75

1.00

1.25

Abso

rban

ce (a

.u.)

Wavenumber (cm-1)

Amorphous

Crystalline

Difference

9 s 21 s 24 s 27 s

0.75

0.45

0.60

0.90

Wavenumber (cm )

30 s 36 s 45 s 75 s

0.15

0.30

10 100

0.0

0.2

0.4

0.6

0.8

1.0

Abs

orba

nce

(a.u

.)

Time (s)

A B C D Average

A

B

C

D

Maximum Rate Maximum Rate Time (s)

0.075

0.015

0.045

0.030

0.060

0.090

60

20

40

30

50

70

Motivation

• Cancer pathology– Prostate cancer as a paradigm – 1

in 6 men– Biopsies: >1 million annually with

disease ~20%– Diagnoses: >200, 000 annually

with lethal ~20%G di i bj ti i bl

Screening

Biopsy

Is it suspicious?

Is it cancer?

D I A G N O S T I C

– Grading is subjective, variable, leads to conflicting therapy routes

– Prognosis tools are not perfect –97% undergo therapy

– “Holy Grail” of oncologic pathology– Primary evaluative standard for

research

• Manual recognition in stained tissue

Therapy

Follow-up

Diagnosis

What is the grade? Stage?Will it metastasize?

Is adjuvant therapy needed?

Has cancer recurred?

R E S E A R C H

Cancer and Diagnostic Process

Biopsy

• No stains, no manual decisions

Epithelium

Stroma

BenignGrade 3Grade 4

American Cancer Society, est. 2008

6/8/2009

20

From Data to Knowledge

Information

x=2048, y=2048, z=2048, t ~ ms to days

￭ Model based design of experiments￭ Hypothesis driven analysis – supervised data analysis￭ Biologically inspired statistical pattern recognition of spectra￭ Approach – Model, Train algorithm, Classify, Validate

x

y

z

Analytical Approaches

The types of analysis available to spectroscopists can be divided into two groups;

Un-SupervisedAll data has the same weighting and no-prior information about what the

t d t i kUn Supervised

Supervised

spectra corresponds to is known-Useful for interest of discovery

Known information about data e.g. classes is used for data analysis

-Useful for classification (more amenable to clinical setting?)

PLSPCA

Analytical Approaches

Which MVA approach?

Bayesian

SIMCA

ANNKNNLDA

FUZZYHCA

Important to determine which are best for analysis of point spectra and which for images

6/8/2009

21

Model

Potential Metric Set

Acquired Data

Epith

elium

Fib. Strom

a

Pale Strom

a

Smooth M

uscle

Stone

Lymphocytes

Endothelium

Nerve

Ganglion

Blood

Prostate Histology

Integrative classification process

Gold Standard

Algorithm(Modified Bayesian)

Optimized PredictionAlgorithm

Optimization

Optimal Metric Set

Calibration/ValidationStatistics

Sensitivity Analysis

Visualizing Classification Origin and Variance

met

rics

patients

3

6

9

12

1

4

7

10

2

5

8

11

15

18

13

16

14

17

96 3 104 5 6 1171 8 12 15 181316 14 17 19 20 22 262321 24 25 2927 30 343128 32 33 35 36 37 38

Determining outliers – quality controlVariance and classification

Spectral Diversity and Clustering Control and significant Variables

Examples of FTIR for Biomedical applications

Fundamental biological processes

ApoptosisNecrosisStem cell differentiationToxicologyDisease initiation and progression

Clinical Applications

p g

Disease diagnosisDisease prognosis

6/8/2009

22

FTIR-Biomedical Applications

Biochemical-cell fingerprint region

Integrative Process and Key Technologies

Hi-fi, fast FTIR Imaging

Samples

Custom Tissue Microarrays

* Kononen et al. Nat. Med. 4, 844-847 (1998).Levin and Bhargava Annu. Rev. Phys. Chem. 56, 429-474 (2005).Kong and Bhargava, Submitted (2009): Living tissue arrays

* Lewis et al. Anal. Chem. 67, 3377-3381 (1995).

Biopsy

Pattern Recognition Algorithms

Imaging Data

Diagnosis

Statistical Validation

Llora , Priya, Bhargava Nat. Computing In Press (2009).Bhargava et al. Biochim Biophys Acta. 1758, 830-845 (2006).Reddy and Bhargava (2009): Statistical Model

Fernandez et al. Nat. Biotechnol. 23, 469-474 (2005).Bhargava et al. Nat. Biotechnol. 25, 31-33 (2007).Bhargava Anal. Bioanal. Chem. 389, 1155-1169 (2007).

Lewis et al. Anal. Chem. 67, 3377 3381 (1995).Bhargava and Levin Anal. Chem. 73, 5157-5167 (2001)Bhargava and Reddy Anal. Chem., Submitted (2008): Noise rejection

Prognosis

6/8/2009

23

Application 1: Prostate HistologyValidation of Prostate Histology: 970 samples, 10 million spectra

EPITHELIUM

FIBROUS STROMA

MIXED STROMA

SMOOTH MUSCLE

NERVE

GANGLION CELLS

BLOOD

EPITHELIUM

FIBROUS STROMA

MIXED STROMA

SMOOTH MUSCLE

NERVE

GANGLION CELLS

BLOOD

EPITHELIUM

FIBROUS STROMA

MIXED STROMA

SMOOTH MUSCLE

NERVE

GANGLION CELLS

BLOODBLOOD

LYMPHOCYTES

CORPORA AMYLACEA

ENDOTHELIUM

BLOOD

LYMPHOCYTES

CORPORA AMYLACEA

ENDOTHELIUM

BLOOD

LYMPHOCYTES

CORPORA AMYLACEA

ENDOTHELIUM

1.3 mm

Fernandez, Bhargava, Hewitt and Levin Nat. Biotechnol., 23, 469-474 (2005)

Prostate Cancer Diagnosis

• Overall pixel accuracy ~ 88.5% ; Heterogeneity in samples?• 1 cancer sample classified as benign (71)• 1 benign sample classified as cancerous (69)• Sensitivity and specificity exceeding human capabilities• Large validation studies underway

Array – 80 Patients Array – Histology Pathology Design Pathology Result

Whats in store?

• Label-free methods– Talks on Monday focus on applications

• Lab toursTheory• More information Theory

Instrumentation

SamplingData Analysis

Applications

raman and ir spectroscopy - university of illinois at urbana-champaign

Documents