protocol

Using Fourier transform IR spectroscopy to analyze

biological materials
Matthew J Baker1,13, Júlio Trevisan2,3, Paul Bassan4, Rohit Bhargava5, Holly J Butler2, Konrad M Dorling1,
Peter R Fielden6, Simon W Fogarty2,7, Nigel J Fullwood7, Kelly A Heys2, Caryn Hughes4, Peter Lasch8,
Pierre L Martin-Hirsch2, Blessing Obinaju2, Ganesh D Sockalingum9, Josep Sulé-Suso10, Rebecca J Strong2,
Michael J Walsh11, Bayden R Wood12, Peter Gardner4 & Francis L Martin2
1Centre for Materials Science, Division of Chemistry, University of Central Lancashire, Preston, UK. 2Centre for Biophotonics, Lancaster Environment Centre,
Lancaster University, Lancaster, UK. 3School of Computing and Communications, Lancaster University, Lancaster, UK. 4Manchester Institute of Biotechnology
(MIB), University of Manchester, Manchester, UK. 5Department of Bioengineering, University of Illinois at Urbana-Champaign, Urbana, Illinois, USA. 6Department
of Chemistry, Lancaster University, Lancaster, UK. 7Division of Biomedical and Life Sciences, School of Health and Medicine, Lancaster University, Lancaster, UK.
8Proteomics and Spectroscopy (ZBS 6), Robert-Koch-Institut, Berlin, Germany. 9Equipe MéDIAN-Biophotonique et Technologies pour la Santé, Université de Reims
Champagne-Ardenne, UnitéMEDyC, CNRS UMR7369, UFR Pharmacie, SFR CAP-Santé FED4231, Reims, France. 10Institute for Science and Technology in Medicine,
School of Medicine, Keele University, Stoke-on-Trent, UK. 11Department of Pathology, College of Medicine Research Building (COMRB), University of Illinois at
Chicago, Chicago, Illinois, USA. 12Centre for Biospectroscopy and School of Chemistry, Monash University, Clayton, Victoria, Australia. 13Present address: WestCHEM,
Department of Pure and Applied Chemistry, University of Strathclyde, Glasgow, UK. Correspondence should be addressed to F.L.M. (f.martin@lancaster.ac.uk).

Published online 3 July 2014; doi:10.1038/nprot.2014.110

© 2014 Nature America, Inc. All rights reserved.

IR spectroscopy is an excellent method for biological analyses. It enables the nonperturbative, label-free extraction of biochemical
information and images toward diagnosis and the assessment of cell functionality. Although not strictly microscopy in the
conventional sense, it allows the construction of images of tissue or cell architecture by the passing of spectral data through a variety
of computational algorithms. Because such images are constructed from fingerprint spectra, the notion is that they can be an objective
reflection of the underlying health status of the analyzed sample. One of the major difficulties in the field has been determining a
consensus on spectral pre-processing and data analysis. This manuscript brings together as coauthors some of the leaders in this field
to allow the standardization of methods and procedures for adapting a multistage approach to a methodology that can be applied to
a variety of cell biological questions or used within a clinical setting for disease screening or diagnosis. We describe a protocol for
collecting IR spectra and images from biological samples (e.g., fixed cytology and tissue sections, live cells or biofluids) that assesses
the instrumental options available, appropriate sample preparation, different sampling modes as well as important advances in
spectral data acquisition. After acquisition, data processing consists of a sequence of steps including quality control, spectral
pre-processing, feature extraction and classification of the supervised or unsupervised type. A typical experiment can be completed
and analyzed within hours. Example results are presented on the use of IR spectra combined with multivariate data processing.

INTRODUCTION
The use of Fourier transform IR (FTIR) spectroscopic techniques the examination of complex tissues and heterogeneous samples 5.
for the nondestructive analysis of biological specimens is a rapidly Detection by microscopy (see schematic of instrumentation in
expanding research area, with much focus on its utility in Fig. 2) may be accomplished by raster-scanning a point illumi-
cytological and histological diagnosis through the generation nated on the sample or by using wide-field illumination and focal
of spectral images1,2. Molecular bonds with an electric dipole plane array (FPA) or linear array detectors6. At present, wide-field
moment that can change by atomic displacement owing to natu- scanning of a sample is possible in seconds, providing tens of
ral vibrations are IR active. These vibrational modes are quan- thousands of spectra. A variety of choices are available for the IR
titatively measurable by IR spectroscopy3, providing a unique, source, including globar7, synchrotron8–12 and quantum-cascade
label-free tool for studying molecular composition and dynamics lasers (QCLs)13, as well as for the detector (2D FPA, linear array
without perturbing the sample. For interrogating biological mate- or single element)14. The three major IR-spectroscopic sampling
rials, the most important spectral regions measured are typically modes (Fig. 2b) are transmission, transflection and attenuated
the fingerprint region (600–1,450 cm−1) and the amide I and total reflection (ATR). Each mode offers convenience for some
amide II (amide I/II) region (1,500–1,700 cm−1). The higher- samples and challenges for others. In transflection mode, for
wavenumber region (2,550–3,500 cm−1) is associated with illustration, the sample is placed on an inexpensive IR-reflecting
stretching vibrations such as S-H, C-H, N-H and O-H, whereas surface (such as that found on low-emissivity (Low-E) slides)
the lower-wavenumber regions typically correspond to bend- and measurements are generated by a beam passing through the
ing and carbon skeleton fingerprint vibrations4. Together, these sample and reflecting back from the substrate (i.e., the reflective
regions comprise a biochemical fingerprint of the structure and surface) through the sample. As is clear from both theoretical
function of interrogated cellular specimens. A typical biological and experimental studies15,16, the recorded spectral intensities
IR spectrum with molecular assignments is shown in Figure 1. depend on both sample morphology and chemistry. Hence, care
should be taken on substrate choice17,18. Recently, topographi-
IR microspectroscopy cal features of the sample and its effects have been shown to be
Although the spectral domain allows chemical identification, minimized by inputting second derivative spectra in the clas-
the combination with microscopy (microspectroscopy) permits sification model; better segregation of normal versus various

nature protocols | VOL.9 NO.8 | 2014 | 1771

 protocol
Figure 1 | Typical biological spectrum showing biomolecular peak Lipid Protein Nucleic acid Carbohydrate

assignments from 3,000–800 cm − 1, where ν  =  stretching vibrations,

δ  =  bending vibrations, s  =  symmetric vibrations and as  =  asymmetric
vibrations. The spectrum is a transmission-type micro-spectrum from
a human breast carcinoma (ductal carcinoma in situ). The sample was
1.0
cryosectioned (8 µm thick) and mounted on BaF2 slides (1 mm thick)

Amide I
before IR microspectroscopy. Equipment: Bruker IR scope II, circular
diameter of aperture ~60 µm; a.u., arbitrary units. 0.8

Amide II
Absorbance (a.u.)

vas (CH2)
0.6
disease categories facilitates potential spectral histopathological
diagnosis19. Research by Cao et al.20 has demonstrated that if this

vas (CH3)

vas (PO2 )
–
vs (COO–)

vs (PO2 )
vs (CH2)
0.4

–
pre-processing data analysis approach is performed (e.g., after

�s (CH2)

vs (CO-O-C)
both transflection and transmission measurements on dried

vs (C=O)
cellular monolayers), the resulting classification is the same. 0.2

This example suggests that irrespective of sampling geometry,

mathematical tools can be applied to minimize confounding 0

effects and to interpret their influence. As such, spectral process- 2,800 2,400 2,000 1,600 1,200 800
ing may determine the diagnostic efficacy of spectral processing, Wavenumber (cm–1)
not only from a biological perspective but also from the ability
to control optical or distorting influences.
© 2014 Nature America, Inc. All rights reserved.

FTIR imaging provides spatially resolved information based on aligned. One excellent interpretation application of IR imaging
chemically specific IR spectra in the form of an information-rich data is to consider it as a metabolomic tool that allows the in situ,
image of the tissue or cell type being interrogated21–23. Further nondestructive analysis of biological specimens, (e.g., determin-
multivariate data analysis allows potential diagnostic markers to ing the glycogen levels in cervical cytology)25.
be elucidated, thus providing a fast and label-free technology to Data can be recorded from a variety of samples, ranging from
be used alongside conventional techniques such as histology2,22. live cells to formalin-fixed, paraffin-embedded (FFPE) archival
At present24, rapid imaging permits imaging in hours for a whole- tissue typical of a pathology specimen. IR spectra representing
organ cross-section, such as that from the prostate; this not only distinguishing fingerprints of specific cell types (e.g., stem cells
allows one to objectively visualize pathology in situ but the afore- versus transit-amplifying cells versus terminally differentiated
mentioned classification models could also allow one to grade cells) within a defined tissue architecture (e.g., crypts of the
disease on the basis of the cateogries into which spectra might be gastrointestinal tract and cornea)9,26 are now easily recorded.
Consequently, spectral analyses delineate cellular hierarchy on the
basis of protein, lipid and carbohydrate composition and/or DNA
a FPA detector
conformational changes27. For biomedical analyses, the major goal
today is to derive an image of tissue architecture expressing the
underlying biochemistry in a label-free fashion28, a development
Single-point CCD camera
that can considerably extend our diagnostic potential beyond
detector present capabilities. For example, to distinguish cells committed
Aperture
toward a pathological process (e.g., transformation) that conven-
tional methods (e.g., visual scoring) might identify as normal. The
Objective
screening of cervical cytology specimens to distinguish normal
IR source
versus low-grade versus high-grade cells4,29, to grade primary neo-
Beamsplitter
Microscope stage plasia30, or to determine whether tissue margins and ­potential
metastatic sites are tumor free31,32 are examples of this concept
across many types of tissues. It is this bridge from the technology
and potential of IR spectroscopy and imaging to biological, mainly
Fixed mirror clinical, applications that is the subject of this protocol (Fig. 3).
Moving mirror
IR spectroscopy in cancer classification and imaging
b Transmission Transflection
By using IR spectroscopy either as an imaging tool or by clas-
Infrared light
sifying spectral categories, it has been possible to distinguish

Figure 2 | The instrumentation underlying the main forms of IR

Substrate spectroscopic sampling. (a) Schematic of modern FTIR-imaging
ATR (e.g., Low-E slides)
Substrate (e.g., spectrometer. Reproduced with permission from ref. 6. (b) Schematic
calcium fluoride) representation of the three main sampling modes for FTIR spectroscopy.
Evanescent wave
Reprinted from Trends Biotechnol, 31, Dorling, K.M. and M.J. Baker,
ATR crystal
Highlighting attenuated total reflection Fourier transform infrared
(e.g., diamond) spectroscopy for rapid serum analysis, 327–328, Copyright 2013 with
permission from Elsevier (ref. 132).

1772 | VOL.9 NO.8 | 2014 | nature protocols

 protocol
Figure 3 | FTIR spectroscopy work flow for Sample preparation FTIR spectral acquisition
imaging and diagnosis. The three major steps
are sample preparation, FTIR spectral acquisition Sample format Sample substrate/ Light source Sampling mode
mount
and data analysis. Sample preparation may differ Globar Transmission
Fixed cell/tissue BaF2/CaF2 slides
depending on the sample format, requiring
different materials and procedures. At FTIR Synchrotron ATR
Live cells IR-reflective slides
spectral acquisition, several options have to be Other
Biofluids Microfluidic devices
considered for light source and sampling mode.
Data analysis presents different paths depending
on the analysis goal (i.e., imaging or diagnosis).
The framework for diagnosis is somewhat more
Data analysis Data analysis (diagnosis)
complex, involving training of classification Spectral cube (imaging)
systems and validation of these systems using y Training Test
data set data set
test data sets. Although not illustrated, x
the data sets used for testing are also obtained
Quality control Quality control
through sample preparation followed by FTIR
spectral acquisition. Pre-processing Pre-processing
Pre-processing Pre-processing

Trained
Clustering Feature extraction classification Feature extraction
Feature extraction (unsupervised system
between benign and malignant tumors in classification)
Supervised
tissue samples of breast32–35, colon22,23,36, classifier training
Classification
Chemical Indexed
© 2014 Nature America, Inc. All rights reserved.

lung37 and prostate8,30,38,39 along with imaging color

cervical cytology or biopsies4,28,40. IR
Predicted
spectroscopic analysis is also an ideal tool tissue
structure Predicted class
for the study of biofluids such as urine, (e.g., ‘normal’/ ‘cancer’)
saliva, serum or whole blood; the use of
biofluids is desirable in a clinical setting
as samples are obtained rapidly and relatively noninvasively, and environmental toxicology49–52, consumer safety53,54, taxonomy55–57,
minimal sample preparation is required. By using such methods, a and in the food industry58; a non-instrument– and non-­software–
spectral fingerprint of the biofluid can be obtained, which allows specific protocol for imaging and classification could be of
the subsequent classification of spectra from different categories ­considerable use to these areas of research.
with computational methods and possibly the identification of
biomarkers41–44. Experimental design: instrumental options
The main steps required to analyze a sample of interest are sam-
FTIR imaging of tissue and cells ple preparation, instrumental setting, acquisition of spectra and
Imaging of live cells is possible using both globar and synchro- data processing (Fig. 4). Before instrumental options are cho-
tron-based light sources, with the latter permitting greater sen, it is important for the user to understand the expectations
lateral spatial resolution and data quality owing to higher from the intended experiment. These include the desired spectral
flux21,45–47. Diffraction-limited resolution with ATR-FTIR imag- and spatial resolution and type of study (e.g., diagnostic versus
ing can also be advantageous as it allows analysis of live cells exploratory). In addition, proper consideration must be given to
in ­ aqueous systems21,48. In addition, the spatial resolution of potential sample restrictions such as acquiring appropriate sam-
the image can be increased by incorporating optics with a high ple thickness for respective modes.
refractive index21,34.
We describe a protocol that has three components: (i) speci- Sampling modes. Figure 2b shows a schematic representation of
men preparation and removal of possible sample contaminants; each sampling mode and details of each can be seen in Table 1;
(ii) acquisition of spectra with a sufficiently high signal-to-noise however, it is important to note that different manufacturer
ratio (SNR); and (iii) data processing for classification and imag- systems may vary slightly in some parameters, such as sampling
ing. As the precise steps in acquisition of spectra and data process- apertures. Transmission and transflection sampling modes have
ing are, respectively, dependent on the instrument and software been applied to a variety of biological specimens that can be sec-
available, this protocol covers (ii) and (iii) to deliver a general tioned into a thin layer allowing for accurate spectral data acqui-
understanding of the steps involved. Supplementary Methods 1–4 sition59. ATR-FTIR mode differs in that the IR beam is directed
correspond to four different examples of standard operating pro- through an internal reflection element (IRE) with a high refrac-
cedures (with troubleshooting) specific to common instruments tive index (e.g., diamond, zinc selenide, germanium or silicon) 60.
and acquisition/analysis software. Together, this protocol and the The evanescent wave extends beyond the IRE surface penetrating
material contained in Supplementary Methods 1–4 are designed the sample, which must be in direct contact with the IRE. The
to build researchers’ confidence in conducting their studies using penetration depth of this wave typically ranges from 1 to 2 µm
their own instrumentation and computational settings. within the 1,800–900 cm−1 region, but it should be remembered
that there is still ~5% intensity at a depth of 3 µm (refs. 18,61,62).
Application of this protocol to other research areas It has been shown that samples with thicknesses of <2 µm may
The application of this protocol is not limited to the biomedical give rise to spectral artifacts with IR-reflective substrates such as
field. IR spectroscopy has previously been used in the fields of MirrIR Low-E slides (Kevley Technologies); therefore, when these

nature protocols | VOL.9 NO.8 | 2014 | 1773

 protocol
Raw ATR-FTIR spectra

1,800 1,600 1,400 1,200 1,000

Wavenumber (cm–1)

Rubber band baseline correction First differentiation Second differentiation

0
0

0
1,800 1,600 1,400 1,200 1,000 1,600 1,400 1,200 1,000 1,600 1,400 1,200 1,000

Normalization to the amide I peak Normalization to the amide II peak Vector normalization Vector normalization
© 2014 Nature America, Inc. All rights reserved.

0
0

0 0
1,800 1,600 1,400 1,200 1,000 1,800 1,600 1,400 1,200 1,000 1,600 1,400 1,200 1,000 1,600 1,400 1,200 1,000

Figure 4 | Visual effect of different pre-processing steps on a set of FTIR spectra. Two common pre-processing sequences are rubber band baseline correction
followed by normalization to the amide I/II peak and first or second differentiation followed by vector normalization. Rubber band baseline correction
subtracts a rubber band, which is stretched ‘bottom-up’ at each spectrum, eliminating slopes. Amide I/II normalization forces all spectra to have the same
absorbance intensity at the amide I/II peak. Differentiation (Savitzki-Golay (SG) method) has the advantage of eliminating slopes while also resolving
overlapped bands, but has the drawback of altering the shape of the spectra (the y axis unit is no longer a.u. (arbitrary units), but ‘a.u. per wavenumber’
(first differentiation) or ‘a.u. per wavenumber squared’ (second differentiation)) and enhancing noise (note how second-differentiated spectra are visibly
more noisy). Vector normalization is typically applied after differentiation. This normalization technique does not require a reference peak as amide I/II
normalization does.

substrates are used with ATR-FTIR spectroscopy, a thicker sample ~1,000 µm in diameter, providing a uniformly illuminated aperture
is recommended18. of 20–100 µm of the diameter at the sample65. It has been shown
A magnification-limited digital camera may be used for vis- that single-cell investigations can be conducted using standard
ualization in order to guide manual navigation across a given globar IR sources to derive subcellular information66.
sample so as to locate a region of interest and help identify basic A synchrotron radiation light source is ~100–1,000-fold
microscopic features such as separation between cancer cells and brighter than current benchtop thermal ones, but it illuminates
stromal elements. An alternative setup for ATR involves placing a much smaller area. Thus, a synchrotron source has a natural
the sample directly onto the IRE aperture of the ATR accessory. sampling aperture of 10–20 µm in diameter with a high SNR67.
This is particularly useful for biofluid analysis as it bypasses any It is therefore possible to achieve single-cell and large organelle
potential contributions from any slide substrate that the sample (e.g., nucleus) lateral spatial resolution with these modern
could be placed on (Supplementary Method 1). This method- sources, allowing subcellular molecular distribution analysis 68,69.
ology may also help to reduce experimentation time owing to There are ~50 synchrotron facilities worldwide, all easily acces-
reduced sample preparation. sible for routine use as they operate on a call-for-projects basis 70.
Alternatively, other available sources that may be advantageous
Light sources. In IR microspectroscopy, the user has the to individual studies include optic parametric oscillator (OPO)
option of several light sources: a conventional thermal (glo- lasers, QCLs and free-electron lasers (FELs); traditionally they
bar) or synchrotron radiation source for FTIR interfero- have been primarily used for gas sensing because of intrinsically
metric measurements or alternative sources such as QCLs63 narrow linewidths71,72; however, modern QCLs can cover much
and filters64, which obviate the use of interferometers. broader wavelength regions (hundreds of cm−1).
The majority of benchtop instruments use conventional thermal
light sources often in conjunction with single-­element detectors. Mapping versus imaging. Broadly speaking, detectors can be
A globar source is composed of a silicon carbide rod that gener- separated into single-element, linear array and FPA detectors;
ates IR radiation, and can typically generate a collimated mean of the detector choice will be influenced by the requirement being

1774 | VOL.9 NO.8 | 2014 | nature protocols

 protocol
Table 1 | FTIR spectroscopy modes used for the interrogation of cellular materials.

Typical
Suitable interrogation
Mode samples Substrate area (mm) Pros Cons

ATR Tissues, Calcium 250 × 250 High SNR Can be destructive because of
cells and or barium Reduced scattering pressure
biofluids fluoride, zinc Analysis of large target area Air between sample and IRE will
selenide, affect spectra
Better for aqueous samples with
MirrIR Minimum sample thickness is
appropriate substrate
Low-E-coated required (~2.3 µm)
glass Highest spatial resolution (because
of the refractive index n, which is 3.5 Interactions of samples with
or even 4 in case of Si or Ge) the IRE leading to structural
alterations (e.g., secondary
protein structure)

Transmission Tissues, Calcium 5 × 5 to High spatial resolution Lower SNR than ATR
individual or barium 150 × 150 Nondestructive of prepared sample Maximum sample thickness is
cells, cellular fluoride and
© 2014 Nature America, Inc. All rights reserved.

Automated stage allows for spec- required

components zinc selenide tral acquisition at several different Sample thickness should be
and biofluids locations of choice with little user twice as large as for transflection
interaction to achieve the same absorbance
Longer sample and machine
preparation is required

Transflection Tissues, Calcium 5 × 5 to High spatial resolution May give rise to standing wave
individual or barium 150 × 150 Nondestructive of prepared sample artifacts
cells, cellular fluoride and Automated stage allows for spec- Lower SNR than ATR
components zinc selenide tral acquisition at several different Maximum sample thickness is
and biofluids locations of choice with little user required
interaction Longer sample and machine
Approximate sample thickness can be preparation is required
1–4 mm, whereas for transmission it Scattering effects such as
needs to be 2–8 mm RMieSc will be much more
intense in transflection type
measurements

imaging (i.e., FPA) or point spectra with high SNR (i.e., single In contrast to aperture-based systems, non-aperture-based
element). The use of a single-element detector allows for indi- instruments such as FPA and linear array detectors provide imaging
vidual point spectra to be obtained across a whole sample (for using spatially arranged detectors. Multielement detectors allow
instance, useful when analyzing biofluids); a particular application for simultaneous spectral acquisition, which, combined with suit-
has been to derive single-cell–specific fingerprint spectra across able optics, produce spectral images with good SNR and lateral
a heterogeneous tissue section. Acquiring large data sets contain- spatial resolution close to the diffraction limit75. Measurements
ing point spectra is a method regularly used in biomedical and using an FPA detector (typically 32 × 32, 64 × 64 or 128 × 128)
environmental studies coupled with multivariate data analysis40,73. are rapid as such detectors allow for the acquisition of thou-
Although time consuming, point spectra often have a high SNR, sands of spectra simultaneously 76; for a typical methodology
resulting in high-quality spectra, as spatial resolution is limited see Supplementary Method 2. The acquired spectral data can
by IR apertures74. Maps can be generated when point spectra are be used to generate pseudocolor images of the target area such
collected in a stepwise manner in a grid from a target area, which as shown in the characterization of prostate tissue77 and cervical
is useful for comparing the different cell types from that particu- biopsy samples28. The benefits of using a synchrotron radia-
lar area, e.g., gastrointestinal crypt23. Spectral maps take a much tion light source with FPAs also mean that much smaller pixel
longer time than individual point spectra and, thus, in order to sizes can be used (e.g., 0.54 µm × 0.54 µm at some synchrotron
make large maps feasible to run, the acquisition time for each point facilities) resulting in higher spatial-resolution images of the
can be reduced leading to a lower SNR. The absorbance intensity target area76.
at each spectral point within the map becomes an individual pixel ATR-FTIR spectroscopy coupled to an array detector can
in the resultant pseudocolor images, which can give details of how allow for sample imaging down to diffraction-limited resolution
different biomolecules vary across the target area. for the spectral range of interest 78. The spatial resolution of

nature protocols | VOL.9 NO.8 | 2014 | 1775

 protocol
non-aperture-based techniques is determined by the optics cho- that transmission or ATR spectroscopy measurements are more
sen, and it has been shown that a germanium optic is preferential, applicable to interrogation of biological material.
although ZnSe and diamond crystals can also be used34. Although
transmission and transflection imaging have been widely imple- Microfluidic devices. Traditionally, aqueous sampling environ-
mented in biological tissues, imaging in ATR mode is a versatile ments were unsuitable for IR spectroscopy because of the con-
option, because little sample preparation is required owing to tribution of water. Development of microfluidic devices and
minimal sample-thickness restrictions, which thus means that it processing to remove the water contribution has made it possible
has been implemented in biological fields such as pharmacology to achieve real-time, live-cell monitoring with IR spectroscopy.
and subcellular interrogation59,78,79. Nondestructive to cells, it better replicates physiological condi-
tions; no labeling is required and the resolution is such that single
Experimental design: sample preparation cells can be studied83. The nondestructive nature of these meth-
Sample formats. The main sample formats for clinical IR spec- ods has allowed studies to look at samples over time (e.g., stem
troscopy are fixed cell and tissue samples, biofluids and live cells. cells in situ as they differentiate and chemical reactions in flow
Spectroscopic approaches can be used to examine tissues of systems have been monitored84,85.
human extraction (all require the appropriate ethical approval The key challenge of IR spectroscopy using microfluidics is
before their use). The type of sample used greatly determines associated with the materials’ transparency over the spectral
which type of IR spectroscopy is appropriate and how it should range to be studied, and especially when live-cell monitoring is
be prepared for analysis. Table 2 shows the main types of samples desirable. Many potential window materials are unsuitable on
and how they should be prepared for analysis. the basis of their water solubility (e.g., KBr and NaCl), toxic-
© 2014 Nature America, Inc. All rights reserved.

ity toward the cells under observation (e.g., CdTe) or spectral

Sample thickness. Sufficient thickness of material needs to be dispersion (e.g., ZnS and BaF2)86. A flow chamber is used that
placed onto the support matrix to allow a sufficiently large absorb- combines IR transparency and robustness of diamond as window
ance intensity to be recorded. In transmission and transflection material. Although manufacture is complicated, the windows
modes, the specimen thickness needs to be adjusted appropriately: must be sufficiently thin (0.4–0.8 µm) to avoid multiple inter-
if it is too thick, the detector response function will be nonlinear nal reflections86. CaF2 is extensively used as a window material,
so that Beer-Lambert’s law cannot be applied anymore. This has and a simple flow cell with inlet and outlet flow is constructed
serious consequences for subsequent quantitative and classifica- by clamping two CaF2 plates together. One of the plates is etched
tion analyses. In contrast, to achieve an adequate SNR and to to form a 10-µm well, designed for the IR observation of live
avoid interactions of the evanescent wave with the underlying cells in aqueous media85. A similar device has been used for
substrate, samples must also not be too thin. For example, when synchrotron IR spectroscopy of living cells using a surface
using ATR-FTIR spectroscopy, it is ideal if the specimen is three- micro-etched silicon substrate87. Further advances in the field
or fourfold thicker than the penetration depth (that said, there have led to the development of sandwich devices and entirely
is no maximum thickness for ATR-FTIR, and samples that are polymeric devices.
even a millimeter thick can be analyzed). This is pertinent for
internal reflection measurements, which are commonly used for Experimental design: spectral acquisition
the disease diagnosis of biofluids; such samples can be naturally Instrumental and operational settings to maximize spectral
thinner in composition (especially with regard to cerebrospinal quality. When acquiring spectra, it is important to maximize as
fluid (CSF), although this is not so much the case with blood best as possible the SNR in order to produce high-quality spectral
or serum/plasma; serum, for example, is a solution containing a data (Table 3). There are a number of noise-related and signal-
high protein concentration, ~80 mg ml−1). The effect of substrate related parameters, with an effect on SNR, which can be altered
interference on spectra, especially in reference to transflection depending on the instrument mode being used (e.g., point mode
measurements, has now been shown independently in the last versus imaging)88–91. The instrumental and operational settings
year by several groups17,18,80. Given this, we would urge extreme will be specific to the user experimental setup; Table 1 compares
caution regarding the use of Low-E slides with transflection meas- properties of different sampling modes for optimized spectral
urements; with ATR-FTIR, it is unlikely that there will be optical acquisition. An initial noise-related parameter that can be altered
effects associated with substrate. is the sampling aperture in point or mapping mode; this will
reduce the SNR when the aperture size is reduced92. However, in
Substrate choice. Proper consideration of the substrate (the slide imaging mode there is no aperture. The interferometer mirror
or matrix) upon which the sample will be placed and any prepara- velocity may also have an effect on SNR3. Weighting the inter-
tion steps associated with this are essential in order to acquire the ferogram with an apodization function will also contribute to a
best and most-reproducible spectra. For transmission measure- reduction in SNR, as this smoothing effect can incorporate spec-
ments, this needs to be an IR-transparent material such as BaF2 or tral artifacts while one is attempting to optimize the information
CaF2 (the latter, in particular, for live-cell IR spectroscopy), whereas contained93. In general, the square root of the number of co-
for reflection or transflection measurements an IR-reflective additions is proportional to the SNR, and therefore an increased
substrate (e.g., Low-E slides) is required because glass alone number will enhance the SNR94.
absorbs the radiation and has a spectral signature in the mid-IR IR spectroscopy has a spatial resolution that is limited by the
region81,82. Previously, it had been recommended that biological diffraction limit; hence, as the resolution approaches this value,
materials be placed on IR-reflective substrates. However, there the SNR is reduced to a point where there is no further gain in
now appears to be a shift in the general consensus that suggests image quality95 A synchrotron radiation source (e.g., at the IR

1776 | VOL.9 NO.8 | 2014 | nature protocols

 protocol
Table 2 | Sample types and preparation.

Sample type Preparation Removal of contaminants Sample mount Considerations

Biofluids Biofluids such as blood, urine, When using blood-based Biofluids may be placed Dry film analysis (where
saliva and synovial fluid should be biofluids such as serum onto slides and dried, or the fluid is dried onto
collected as per hospital SOPs and plasma, spectra from dried directly onto the the slide) often results in
Samples that are not immediately erythrocytes may mask IRE44 large signals compared
used should be frozen and stored that of other biomolecules, with the wet biofluid,
at  − 80 °C so they should be removed but measurements may
Samples should be thawed fully if not being directly be impeded by uneven
before use investigated41 distribution
Only small sample
volumes are needed,
normally in the region of
a few nanoliters136

FFPE tissue FFPE tissue should be de-waxed for Samples must be de-waxed De-waxed tissue should If using tissue for
samples a minimum of 5 min in xylene and in order to probe the be floated onto slides imaging and extraction of
three washes should be performed full wavenumber range, tissue cell type, sample
© 2014 Nature America, Inc. All rights reserved.

Sample thickness should not exceed as paraffin is known to thickness is not just an
8–12 µm (transmission, less for have significant peaks at SNR issue. The thicker
transflection; see Table 1) in ~2,954 cm − 1, 2,920 cm − 1, the tissue, the greater
order to avoid a nonlinear detector 2,846 cm − 1, 1,462 cm − 1 the chance of probing
response (at absorbance values  > 1.2 and 1,373 cm − 1, which heterogeneous layers and
(for MCT) or  > 1.5 (for deuterated may mask solvent-resistant possibly multiple cell
triglycine sulfate)), to even total methylene components of types, rendering cell type
absorption native tissue128,129 signal less pure
Samples are then cleared
with acetone to remove
any final xylene
contamination
Another recent and
emerging alternative is
to model the paraffin
contribution and
numerically de-paraffinize
the sample36. In this way,
the sample is not
affected by chemical
de-paraffinization, and
intact tissue biochemical
information is used for
spectral histology

Cryosectioned Tissue must be thoroughly thawed Serial sections should be Snap-frozen tissue Although snap-freezing
tissue before IR analysis carefully isolated from the should be cut and negates the use of
samples Once a sample is thawed, compo- cryoblock to prevent OCT placed onto slides fixatives such as formalin
nents may start to degrade, so we compound contamination or the use of paraffin,
suggest imaging sections as soon as of the final tissue slice it may damage the
possible after thawing and drying, in structural integrity
a dark environment130 of the tissue
However, under dry conditions,
cryosections can be stored for
months without major problems
other than lipid oxidation, as seen
by the decrease of the ester carbonyl
bands (degrades within 2 weeks; this
can be avoided when samples are
stored in a N2 atmosphere)
(continued)

nature protocols | VOL.9 NO.8 | 2014 | 1777

 protocol

Table 2 | Sample types and preparation (continued).

Sample type Preparation Removal of contaminants Sample mount Considerations

Fixed cells Medium contaminants must be After formalin fixation, Cells can be grown onto If grown on slides, cells
removed before cells are placed in cells should be washed in IR substrates that have will typically be thin,
fixative such as ethanol or formalin HBSS before IR analysis to been first sterilized in as they grow and stretch
For formalin fixation, cells should be remove residual phosphate 70% (vol/vol) ethanol, over a 2D surface
washed twice in PBS before suspension ions as growing directly onto Cells fixed and then
in formalin for at least 30 min After ethanol fixation, the slides can preserve placed onto slides may
Slides should be dipped three times slides should be left to dry cell morphology be uneven in thickness,
in double-distilled water (this should for 24 h on the benchtop They can also be grown which may be resolved
not be extended beyond quick dips) and 24 h in a desiccator in a 3D culture matrix, using cytospinning,
as formalin fixation can be reversed so that all residual ethanol which can then be fixed which allows cells to be
in the presence of water92 evaporates or frozen and sectioned; proportionally dispersed
For ethanol fixation, cells should be this may provide the over the substrate
washed three times in ethanol most realistic environ-
(min. 70% (vol/vol)) before being ment in which cells can
left to stand in ethanol for at be studied
© 2014 Nature America, Inc. All rights reserved.

least 1 h

Live cells Cells that are to be analyzed in Cells in suspension must Spectra recorded The critical β-DNA
suspension should be detached from be washed with PBS to in an aqueous conformational marker
the growth substrate using trypsin remove residual medium or environment show bands are enhanced in
and then stored at 4 °C to prevent trypsin minimal dispersion the hydrated state2,
autolysis137 because the refractive and thus can be used
For ATR-FTIR measurements, cells index of aqueous to determine the
can be seeded and grown directly medium for the concentration of DNA
onto the ATR IRE using a cell background single- in simple cells138,139
chamber59 beam spectrum closely Single-cell micro­
matches that of the spectroscopy is
cell for the sample inherently difficult
spectrum47 because of the strong
Therefore, cell absorptivity of the water
suspension can be molecule, which can
placed onto the IR swamp the spectrum
slides as microdroplets especially when the
Cells can be grown sample path length
directly onto a is  > 10 µm (ref. 140)
detachable IRE such A bright source of IR
as diamond for photons is required to
ATR-FTIR analysis achieve a good SNR
Live cells can also be because the IR beam
analyzed in situ by must usually pass through
the use of microfluidic two IR transparent
devices21 windows, cell medium
and the hydrated cell,
causing attenuation of
the IR signal
Thus, most measurements
performed on single
living cells with an FTIR
microscope configuration
use a synchrotron
light source

Environmental Imaging Facility (IRENI) at the Synchrotron enhanced SNR spectra when using apertures approaching the
Radiation Centre (SRC)) in the mid-IR region is 1,000 times diffraction limit; however, when using an FPA detector, this can-
brighter than a thermal globar source and thus may generate not be exploited as the brightness is applied over a larger area. By

1778 | VOL.9 NO.8 | 2014 | nature protocols

 protocol
Table 3 | Typical conditions of the main variables affecting SNR in spectroscopy instruments.

Instrument options
FTIR ATR-FTIR
Single-element Single-element
Variable detector FPA detector FPA
Light source Globara Synchrotronb Globarc Synchrotrond Globar e Globarf

Sampling aperture 15 × 15 to 5 × 5 to 700 × 700 µm 50 × 50 to 250–250 µm 60 × 60 to 700 ×

150 × 150 µm 20 × 20 µm FOV 175 × 175 µm 700 µm

No. of co-additions 512 256 64 or 128 128 32 32

Spectral resolution 4 or 8 cm − 1 4 or 8 cm − 1 4 or 8 cm − 1 4 or 8 cm − 1 8 cm − 1 4 or 8 cm − 1

aRef. 141. bRefs. 119,142. cRefs. 142,143. dRefs. 76,144. eRefs. 145,146. fPatented by Agilent Technologies59.

using multiple beams, such as at IRENI, the single-beam disad- A background single-beam spectrum provides the source inten-
vantage when using an FPA may be overcome. sity, as modified by the instrument; placing a sample in the
© 2014 Nature America, Inc. All rights reserved.

It is important to consider that an optimized and well-aligned beam path and measuring the single beam again, theoretically,
benchtop instrument is not considered to be inferior with provides just the additional effect of the sample absorbance.
regard to SNR or image quality to a general synchrotron-based A logarithm (to the base 10) of the ratio of these quantities pro-
machine63. A number of options regarding the detector can also vides the absorbance, which is directly related to concentration by
have an effect on the SNR, such as the choice between a ther- Beer’s law. With point spectra, a background spectrum is typically
mal detector versus a quantum detector. A mercury cadmium retained for recording 5–10 sample spectra and with each different
telluride (MCT) quantum detector usually provides a superior sample to reduce the effects of constantly changing atmospheric
SNR than, for example, a thermal detector such as a deuterated conditions. As spectral maps are composed of a large number
triglycine sulfate detector96. An optimized cooling system in the of point spectra acquired in a stepwise manner, it is necessary
detector, such as thermoelectrical cooling, will also reduce the to set up background scans to be taken at set intervals (e.g., at
dark current produced by the detector, which has been shown to the end of every row) to account for the atmospheric variation
have a detrimental effect on SNR97,98. In addition, signal-related over the extended acquisition time66. When acquiring spectral
parameters can affect the SNR; for instance, an increase in the images, background spectra should be acquired over a defined
optical path length can reduce spectral quality, which has been time period, depending on the sample acquisition time.
particularly important in the analysis of aqueous samples such
as biofluids33. When producing spectral images with the help of Experimental design: data processing
multielement detectors, such as an FPA, one must consider opti- Data processing is carried out in a sequence of steps (Fig. 3) and
mizing the SNR. The authors point readers to the authoritative the most important factor determining its workflow is the analysis
reference on FTIR spectroscopy by Griffiths and De Haseth3 for goal; typical spectroscopy software programs used are shown in
theoretical and instrumental discussions; this book has supported Table 4. Here we describe two analysis goals: imaging and diag-
the authors since their undergraduate studies and continues to nosis. Other goals not covered here include pattern finding and
support them today3. biomarker identification101,102.
Imaging is defined as data analysis that uses an unsupervised
Water vapor and instrument purging. The presence of water data processing method to reveal tissue structure on a ‘spectral
vapor in the instrumentation and sample area can result in cube’ acquired by a mapping or imaging technique. Imaging allows
reduced transmission of IR light, potentially obscuring impor- for the study of shape and penetration of important histopatho-
tant spectral details even at low spectral resolutions often used logical features on the basis of the underlying chemistry28.
in biomedical IR spectroscopy. Water vapor interference can be In contrast, a diagnosis using IR spectroscopy requires a more
minimized by computational subtraction of a pure water vapor complex framework that uses supervised classification methods.
spectrum from the sample spectrum99. The efficacy of this com- A supervised data processing method is one that uses classes
pensation is limited and it is therefore considered crucial before assigned a priori to each IR spectrum as teaching information to
spectral acquisition to purge the instrumentation with dry air or build models that are used later to predict the classes of a data set
nitrogen and/or desiccants to remove any water vapor that may that does not have classes associated with its spectra 103,104. The
contaminate spectra between 1,350 and 1,950 cm−1, and between modeling process for diagnosis requires separate training and
3,600 and 3,900 cm−1 (ref. 100). By doing so, ambient CO2 is also testing stages and respective training and test data sets. The opti-
purged, thereby reducing its contribution to the spectra. mal size of a training data set (i.e., one that will maximize clas-
sification accuracy at a reasonable cost of data set generation) has
Acquisition of sample and background. Measurements of an FTIR been underinvestigated to date, but it has been suggested that it
absorption spectrum involve collecting a ‘single-beam’ spectrum. may be problem dependent105. For example, in a study, one could

nature protocols | VOL.9 NO.8 | 2014 | 1779

 protocol
Table 4 | Some existing FTIR spectroscopy data analysis software.

Software Website Description License

Cytospec http://www.cytospec.com Software for hyperspectral imaging Commercial;

(IR and Raman) free demo available

IRootLab https://code.google.com/p/irootlab/ MATLAB toolbox for biospectroscopy Open source

data analysis

OPUS http://www.bruker.com Spectral acquisition software with data Commercial

processing capabilities

Pirouette http://www.infometrix.com Chemometrics modeling software Commercial

Unscrambler X http://www.camo.com Multivariate data analysis and design of Commercial

experiments

PLS, MIA, EMSC toolboxes http://www.eigenvector.com MATLAB toolboxes for spectroscopy Commercial
data analysis

OMNIC http://www.thermoscientific.com Spectral acquisition software with data Commercial

© 2014 Nature America, Inc. All rights reserved.

processing capabilities

PyChem http://pychem.sourceforge.net/ Package for univariate and multivariate Open source

data analysis

ENVI, IDL http://www.exelisvis.com Integrated development, data analysis Commercial

and image processing suite

MCR-ALS toolbox http://www.cid.csic.es/homes/rtaqam/ MATLAB Toolbox implementing the Open source

tmp/WEB_MCR/welcome.htm MCR-ALS algorithm

start with ten samples (acquiring 5–10 spectra from each sample), methods may be divided into de-noising, spectral correction,
creating a trained model with eight samples and testing the model normalization and other manipulations; two or three methods are
using the remaining two samples; one could then repeat this pro- often combined (e.g., de-noising followed by spectral correction
cedure four more times, each time using two different samples and normalization). The choices of pre-processing methods may
for testing and the remaining eight samples for training (this is depend on the analysis goal, the physical state of the sample, and
called five-fold cross-validation). The number of times that the the time and computing power available.
classifier correctly guessed the class of the testing sample would De-noising of IR spectra may be carried out with Savitzki-
be counted to calculate a classification rate (i.e., the number of Golay (SG) smoothing, minimum noise fraction108 or wavelet
correct guesses divided by the total number of guesses). Next, one de-noising (WDN)101. The latter is known to be the best method
could acquire spectra from an additional five samples and repeat for eliminating high-frequency noise while still keeping intact
the cross-validation process, comparing the new classification rate high sharp peaks (this is essential in Raman spectra processing,
with the old one (it is expected to improve). The process of adding but WDN works well on IR spectra too). Another option is to
samples and repeating cross-validation could continue until the decompose the spectra by principal component analysis (PCA),
classification rate stops improving. and then reconstruct them from only a few of their principal
It is important to note that a diagnostic framework may be components (PCs), thus discarding those PCs that represent
set to use either point spectra or image maps; in the latter case, mostly noise85,109.
the trained classification system can be used to predict tissue Measurement characteristics that may require spectral correc-
structure. tion include:
We describe the following data analysis steps: pre-processing,
• Sloped or oscillatory baselines that result from scattering, with
feature extraction (FE), clustering (unsupervised classification)
resonant Mie scattering in biological materials being the most pro-
and supervised classification, and we exemplify some visualization nounced effect. The effects of sample (scattering centers, edges and
options in the ANTICIPATED RESULTS section. Quality control substrates) have often been lumped together and the effects of the
is another step that is not covered in this protocol, but there are same on spectra are termed ‘artifacts’. Although this terminol-
guidelines on this available in the literature105,106. ogy was initially acceptable, it is now clear that there is a rational
explanation for these effects and they arise merely from the cou-
Pre-processing. Pre-processing essentially aims to improve the pling of morphology and optics. Hence, we will refer to these
robustness and accuracy of subsequent multivariate analyses and as morphological effects on spectra. There are two major efforts
to increase the interpretability of the data by correcting issues in understanding and resolving these effects to recover absorp-
associated with spectral data acquisition107. Pre-processing tion spectra free from the effects of morphology. The first group

1780 | VOL.9 NO.8 | 2014 | nature protocols

 protocol
of methods is termed ‘physics based’. In this approach, explicit FE has an important or even essential role in both imaging
optical image–formation modeling from first principles is used and diagnosis. For imaging, FE is responsible for generating
to predict and correct data. Here each sample effect (boundary a single value based on the whole of an input IR spectrum.
scattering, scattering centers in the sample and ­substrate) needs This value can subsequently be used to set the color of a pixel
to be explicitly accounted for. The theory has been shown to be in the image; FE is repeated for all spectra, thus forming the pseu-
generally valid and there are methods now for correcting the docolor image. Popular FE methods for imaging include calculat-
same for films, spheres and fibers16,110,111. Extension to more ing the ratios between wavenumber absorbance intensities, area
complex samples is still the subject of ongoing research. A sec- under a subregion of the spectrum, selecting a single wavenumber
ond group of methods may be termed ‘model based’. In these or an ensemble of wavenumbers, or performing PCA. PCA may
methods, a model is assumed to explain all sample effects, typi- be applied to the spectral data set, followed by selection of a single
cally, Mie scattering. Subsequently, rigorous theory is used to
PCA factor for the color gradient.
recover spectra, e.g., including extended multiplicative scattering
For diagnosis, FE constitutes an important data reduction step
correction (EMSC)112, resonant Mie scattering correction (RMi-
in order to match the complexity of the subsequent supervised
eSC)113–115 and rubber band baseline correction116. An indirect
way to deal with baseline slope is to apply first or second deriva- classifier with the amount of data available so as to avoid over-
tive to spectra using the SG algorithm. This alters the shape of fitting or undertraining. PCA is one particular popular form of
the spectra, but may also resolve overlapped bands. Model-based unsupervised FE that is used for this purpose103. The number of
methods will generally be faster than explicit modeling methods PCA factors to retain may be subject to optimization. One way out
and may prove to be broadly useful but need to be validated in is to order the PCA factors from the most to the least discriminant
each case. A third approach, which was traditionally used but is on the basis of their P values as determined by a statistical test. The
© 2014 Nature America, Inc. All rights reserved.

now recognized to be of limited value, is to simply correct base- percentage of explained variance can also be taken into account.
lines with a piecewise linear approach. Obviously, this method is Within FE, the subgroup of feature selection (FS) methods
the fastest, as it requires the least effort to apply and no modeling. is particularly interesting because it can confer biological inter-
It is as yet unclear which of these methods works best. pretability (i.e., identify the wavenumbers most important for
• Spectral contributions may arise from atmospheric water vapor, classification) to the classification system. Popular FS methods
carbon dioxide, paraffin or other interfering compounds. Although include forward FS120 and COVAR121. Variance analyses may also
these artifacts may be compensated mathematically through be used to select spectral variables for elimination122. Another
EMSC117 or other least-squares–based technique118, the most approach to FS is to use spectral features that are obtained from
common actions are to remove contaminated spectral bands from a biochemical understanding of the problem123. These cases in
the data set, improve the control of atmospheric conditions or take which direct spectral interpretation is possible are termed metrics
background spectra more often.In this aspect,before pre-processing, for measures of biochemical activity in the samples. It is impor-
it is often useful to implement quality tests to verify SNR and tant to note that not all metrics may be useful biomarkers. Thus,
minimize water vapor contribution. By following this approach, even FE may be a multistep process, (i.e., one in which metrics
‘bad quality’ spectra are discarded as they can influence subse-
are converted to statistically relevant biomarkers).
quent analysis. The threshold values for defining ‘bad’ and ‘good’
spectra can be adjusted according to the biological application. Clustering (unsupervised classification). Clustering aims at sort-
• It is vital to normalize IR spectra to account for confounding ing different objects (i.e., spectra) into categories or clusters on
factors such as varying thickness of sample. Common normaliza- the basis of a so-called distance measure124. Clustering methods
tion methods are amide I/II peak normalization and vector nor- such as hierarchical cluster analysis (HCA) and k-means cluster-
malization. Amide I/II normalization is often used after baseline
ing (KMC) are frequently used in IR-imaging studies to identify
correction, whereas vector normalization is often used after dif-
tissue morphology23,125. HCA groups spectra into mutually exclu-
ferentiation of spectra (after correction by differentiation, there
sive clusters; in IR-imaging studies, HCA-based segmentation is
is no longer a consistent amide I/II peak in the spectra to allow
for amide I/II peak normalization). For imaging, leaving spectra achieved by assigning a distinct color to the spectra in one cluster.
non-normalized for chemical imaging or unsupervised cluster- Because each spectrum of an IR-imaging experiment has a unique
ing will reveal tissue structures primarily based on absorbance spatial (x,y) position, pseudocolor segmentation maps can be eas-
intensity, whereas normalization will highlight differences in ily generated by plotting specifically colored pixels as a function
biochemical structure. For diagnosis, some form of spectral nor- of the spatial coordinates.
malization is conducted.
Supervised classification. Supervised or concept-driven classi-
The optimal pre-processing method or sequence to apply is a fication techniques are machine-learning techniques for creat-
subject of discussion and no universal best approach exists for ing a classification function from training data. These methods
all samples. Often the choices are based on the problems visually involve a supervised learning procedure in which models are cre-
spotted in the spectra; a more objective criterion is to optimize ated that map input objects (spectra) to desired outputs (class
the pre-processing method (e.g., through a genetic algorithm)119. assignments). Popular supervised techniques are artificial neural
In this protocol, we offer several alternatives based on cues iden- networks, support vector machines (Supplementary Method 3),
tified by visual expression of raw (non-pre-processed) spectra, linear discriminant classifier11,103,126 and Bayesian inference-base
although objective validation will probably become more com- methods77. Among the many criteria guiding the choice of clas-
mon in the future. sifier, the most important is probably the accuracy (related to
sensitivity and specificity) when tested on an independent test
FE. FE methods process the IR spectra to form new variables data set. Other criteria include ease to train, computational time,
based on the original variables (which are absorbance intensities). spatial resolution considerations127 and software availability.

nature protocols | VOL.9 NO.8 | 2014 | 1781

 protocol
Classifiers such as artificial neural networks and support vector that requires only the fitting stage. A general rule of thumb is
machines may require a two-stage training, where the first stage that if two different classifiers are equally well performing on an
is dedicated to finding optimal tuning parameters or architecture independent test data set, the simplest one should be preferred
and the second stage fits the classifier model to the training data. over the more complex one, as simpler classifiers are more likely
Linear discriminant classifier (LDC) is a parameterless classifier to be better generalizers103.

MATERIALS
REAGENTS or Leica Surgipath DB80 blade (Leica Microsystems) ! CAUTION Blades are
 CRITICAL For sample preparation and analysis, please refer to Tables 1 extremely sharp; handle and dispose of them with care.
and 2 and the INTRODUCTION for further information. • Paraffin section mounting bath (40–75 °C; Electrothermal,
• FFPE blocks: see Reagent Setup for further information cat. no. MH8515)
• Sample preparation: advice regarding collection of biofluids, cryosectioned • Desiccator: these can be obtained from various manufacturers,
tissue samples, fixed cells and live cells can be found in the Reagent Setup e.g., desiccator (Duran Group) or WHEATON Dry-Seal vacuum desiccators
section ! CAUTION Human tissues (including biofluids, cytology or FFPE (Wheaton Industries)
blocks) should be obtained with appropriate local institutional review board • Labofuge 400e (Heraeus Instruments)
(e.g., in the UK, this is a Local Research Ethics Committee (LREC)) approval; REAGENT SETUP
generally, ethical permission will be granted for a carefully designed study FFPE blocks  These are prepared according to the standard methods used
in which patient participants sign a consent form. Worldwide, studies using routinely in all pathology laboratories; the overall steps are: immerse fresh
human tissues should adhere to the principles of the Declaration of Helsinki.
Similarly, for research using animals, appropriate approvals are required; The
© 2014 Nature America, Inc. All rights reserved.

Animals (Scientific Procedures) Act of 1986 is the legislation that regulates

the use of animals in scientific procedures in the United Kingdom and this is Table 5 | Instruments and corresponding data acquisition
enforced by the Home Office, which issues the licenses required. software.
Other reagents
• ThinPrep (PreservCyt Solution, Cytyc)
Manufacturer Instruments Software
• SurePath (TriPath Care Technologies)
• Formalin, 10% (vol/vol), neutral buffered (Sigma-Aldrich, cat. no.
Agilent Agilent 670-IR Resolutions Pro
HT501128)
! CAUTION It is a potential carcinogen, an irritant and an allergenic. Technologies spectrometer
Always work in a fume hood while handling it.
• Acetone (Fisher Scientific, cat. no. A/0600/17) ! CAUTION Its vapors may Cary 600 series FTIR
cause dizziness. Always work in a fume hood while handling it. spectrometers
• Ethanol, 2.5 liters (Fisher Scientific, cat. no. E/0600DF/17)
• Virkon (Antec, DuPont, cat. no. A00960632) ! CAUTION It is an irritant. Agilent 600 series FTIR
• Paraplast Plus paraffin wax (Thermo Fisher Scientific, cat. no. SKU502004) microscope
• Xylene (Sigma-Aldrich, cat. no. 534056) ! CAUTION It is a potential
carcinogen, an irritant and an allergenic. Always work in fume hood while Bruker Optics Bruker Tensor 27 spectrometer OPUS
handling it.
• Histoclear (Fisher Scientific, cat. no. HIS-010-010S) ! CAUTION It is an irritant. ALPHA FT-IR spectrometer
• Isopentane (Fisher Scientific, cat. no. P/1030/08) ! CAUTION It is an
extremely flammable, irritant, aspiration hazard and toxic reagent. HYPERION series FT-IR
Always work in fume hood while handling it. microscope
• Optimal cutting temperature (OCT) compound (Agar Scientific,
cat. no. AGR1180) LUMOS FT-IR microscope
• Liquid nitrogen (BOC, CAS no. 7727-37-9) ! CAUTION May cause
asphyxiation and contact with skin will cause burns. Wear cryoprotective JASCO UK JASCO FTIR-4100 series Spectra Manager
clothing and use it in a fume hood.
EQUIPMENT JASCO FTIR-6000 series
Electronic equipment
For a list of commercial instruments available, please refer to Table 5 IRT-5000 FTIR microscope
Substrate
• Low-E slides (Kevley Technologies, CFR) PerkinElmer PerkinElmer Frontier Spectrum 10
• BaF2 slides (Photox Optical Systems)
• Silicon multi-well plate (Bruker Optics) Spectrum Two
• Superfrost slides: these can be obtained from various manufacturers,
e.g., Menzel Glazer Superfrost slides (Menzel-Glaser, cat. no. AA00008132E); Spotlight FTIR microscope
Thermo Scientific SuperFrost slides (Thermo Fisher Scientific); system
or Fisherbrand Superfrost slides (Fisher Scientific)
Accessories Thermo Fisher Thermo Nicolet iS50 OMNIC 8
• Coverslips (Thermo Fisher Scientific, cat. no. 102440) Scientific spectrometer system
• Specac Golden Gate single-reflection diamond ATR accessory (Specac)
• Microtomes: these can be obtained from various manufacturers, Thermo Nicolet Scientific
e.g., Microtome (Surgipath Medical Industries); Leica rotary microtomes FTIR 5700 spectrometer with
(Leica Microsystems, Davy Avenue Knowlhill); or Bright Cryostat
continuum microscope
(Bright Instruments)
• Microtome blades: these can be obtained from various manufacturers,
Shimadzu IRTracer-100 spectrometer Lab Solutions IR
e.g., Feather disposable microtome blades S35 (VWR, cat. no. SURG08315E),
Edge-Rite disposable microtome blades (Thermo Fisher Scientific); IRAffinity-1S spectrometer

1782 | VOL.9 NO.8 | 2014 | nature protocols

 protocol
tissue in formalin solution that acts as a chemical fixative; dehydrate the Live cells  This is an emerging area within the field whereby viable cells
tissue in sequential washes of xylene and ethanol; and embed the tissue in can be spectrochemically analyzed, primarily in a constructed microfluidic
paraffin wax, which creates an airtight barrier. Tissue blocks can then be platform (for a typical method, see Supplementary Method 4).
stored indefinitely at room temperature (20–22 °C). EQUIPMENT SETUP
Biofluids  These are primarily blood plasma or serum, but can also Software  Two types of software are required: spectral acquisition and data
potentially include cerebrospinal fluid, saliva or urine. Typically, after analysis. Spectral acquisition software is normally provided by the instrument
acquisition, such samples should be stored in appropriate tubes at −85 °C manufacturer. Most instrumentation software also provides a number of
preprocessing and sometimes more advanced data analysis options. Various
until they are thawed before analysis.
data analysis software programs and packages exist, ranging from those
Cryosectioned tissue samples  Tissue samples can be snap-frozen and
for general-purpose use to those targeting specific data analysis tasks
stored at −80 °C before use. Tissue should be coated with optimal cutting
(e.g., multiplicative curve resolution–alternating least squares (MCR-ALS)).
temperature (OCT) compound before freezing, and it should be frozen with A popular development environment and programming language is
isopentane cooled with liquid nitrogen. MATLAB (http://www.mathworks.com) in which customized software can
Fixed cells  Typically, these would originate from cytology specimens placed be written for specific tasks. Python (http://www.python.org) is another
in a fixative buffer; an ideal example of this is cervical cytology. However, programming language that is becoming increasingly popular in the FTIR
it could be extended to any cell type isolated in the form of a suspension in a spectroscopy field, and it has the advantage of being open source. For a list of
preservative buffer solution. commonly used software and packages, please refer to Table 4.

PROCEDURE
© 2014 Nature America, Inc. All rights reserved.

Sample preparation
1| Prepare the samples by following the steps listed in one of the options given below. Perform the steps in option A
for FFPE tissue samples; option B for cryosectioned tissue samples; option C for cytological specimens; and option D for
biofluids.
Live cells may be prepared in three main ways for IR-transmission studies: grown directly onto IR substrates; grown in a
3D culture matrix (and then processed as described in options A and B); or fixed in suspension, e.g., as cervical cytological
specimens in fixative obtained from hospital pathology laboratories. Cells that are fixed in suspension should be processed
by following the steps in option C.
   To grow cells on IR substrates, sterilize the IR substrate for 1 h in 70% (vol/vol) ethanol before growing cells directly
onto the chosen IR substrate.

Cells grown onto IR substrates Sterilize the IR substrate for 1 h in 70% (vol/vol) ethanol before growing cells directly onto the
chosen IR substrate. Generally, cellular materials are then fixed in order to preserve their architectural
integrity, and the samples are stored in a desiccator prior to spectral acquisition (Step 2).

Cells grown in 3D culture matrix Cells may be grown on 3D culture matrices (a tissue culture environment or device in which live
cells can grow or interact with their surroundings in three dimensions), and subsequently fixed or
snap-frozen and sectioned as described for tissue samples in Step 1A and Step 1B

(A) FFPE tissue ● TIMING 50 min

(i) Obtain FFPE tissue blocks of interest from a pathology laboratory.
(ii) Place a FFPE block onto an ice block for 10 min. Use a microtome to trim into the block to expose the entire tissue
sample to the face of the block. This will ensure that a full tissue section is cut for analysis. Place trimmed blocks
back on ice for 10 min.
 CRITICAL STEP Make sure that the blocks are cold before cutting sections. This hardens the wax, reducing the
friction between the block surface and blade allowing a much smoother cut.
(iii) Cut a ribbon of 10-µm sections and float it onto a heated water bath (40–44 °C). Separate the individual sections
with forceps.
 CRITICAL STEP Optimal tissue thickness for the maximum SNR should be determined in-house by applying variable
thicknesses of sections (depending on the tissue type) to slides for IR interrogation, e.g., ~3 µm (e.g., for bone),
5 or 10 µm (e.g., for prostate tissue), and 15-µm serial sections to BaF2, CaF2 or Low-E slides. SNR is judged on the
quality of the raw spectra; in particular, the presence of many narrow, sharp peaks indicates high noise. If using tissue
for imaging and extraction of tissue cell type, sample thickness is not just an SNR issue. The thicker the tissue,
the greater the chance of probing heterogeneous layers and perhaps multiple cell types, rendering the cell type
signal less pure.
 CRITICAL STEP Depending on the melting point of the paraffin wax used for embedding tissue samples,
the temperature of the water bath will need to be adjusted to prevent melting of the wax.

nature protocols | VOL.9 NO.8 | 2014 | 1783

 protocol

(iv) Prepare tissue slides by re-floating a single 10-µm-thick tissue section onto a BaF2, CaF2 or Low-E slide for FTIR
microspectroscopy or ATR-FTIR spectroscopy. In our experience, a 5–10-µm section is the optimal thickness for
maximum SNR.
 CRITICAL STEP As BaF2 slides can be 1 cm × 1 cm in size to fit common slide holders, a H&E-stained parallel section
may be required to identify an area of interest for analysis. Once a section is floated onto the water bath, sections
can be picked up on normal microscope slides, dissected using a scalpel for the area of interest and floated back onto
water for application to BaF2 slide.
(v) Place the tissue slide in a 60 °C oven for 10 min.
(vi) De-wax the tissue slide by immersing it in xylene for 5 min at room temperature. Repeat this step twice with fresh
xylene. For small, round slides that are difficult to handle during solvent immersion, slides can be encased into plastic
histology cassettes that can be threaded round a large metal clip. The same procedure can be conducted using hexane.
 CRITICAL STEP For IR analysis, it is necessary to de-wax the tissue in order to probe unhindered the full
wavenumber range. This is paramount as paraffin is known to have significant peaks at ~2,954, 2,920, 2,846, 1,462
and 1,373 cm−1. If there is uncertainty about paraffin removal, these regions of the spectrum can be removed from
subsequent analysis. However, this comes at the cost of probing many solvent-resistant methylene components of
the native tissue128,129.
(vii) Sequentially, wash and clear the tissue slide by immersing it in acetone or 100% ethanol for 5 min at room temperature.
(viii) Allow the tissue slide to air-dry before placing it into an adequate-sized Petri dish for storage in a desiccator.
© 2014 Nature America, Inc. All rights reserved.

 PAUSE POINT Slides can be stored in a desiccator before IR interrogation; in our experience, storage should
be <1 year.
(B) Snap-frozen and cryosectioned tissue samples ● TIMING 120 min + drying time (3 h)
! CAUTION Snap-freezing should be carried out in a fume hood while you are wearing cryoprotective gloves,
clothing and a facemask.
(i) The fresh tissue should be no more than 2 cm in any one dimension; gently blot away any fluids from the surface,
place a cryomold and fill the mold with OCT compound.
(ii) Fill a plastic cryobucket with 3–4 cm of liquid nitrogen. Pour isopentane into the stainless steel beaker until it is
about 1–2 cm deep. Place the stainless steel beaker into the liquid nitrogen and allow temperatures to equilibrate
(3–5 min).
(iii) Take the cryomold containing the tissue sample in OCT compound and use long forceps to lower it into the isopentane;
hold until the OCT compound freezes (60–90 s).
(iv) Remove the cryomold and transfer it to the bucket of dry ice. Wrap snap-frozen tissue in aluminum foil and label it
before storing it in −80 °C freezer.
 PAUSE POINT Snap-frozen tissue can be stored in a −80 °C freezer for several months.
(v) Retrieve previously prepared snap-frozen tissue blocks from the −80 °C freezer and transfer them to the cryostat in dry
ice to prevent thawing.
(vi) Unwrap the frozen block from its protective foil covering and mount it into the cryostat. Allow the block to equilibrate
to the cryostat temperature for 30 min. The optimum cryostat cutting temperature will depend on the sample,
but −20 °C will be suitable for most tissues.
(vii) Cut sections with a cryostat until the region of interest is reached. Next, take serial sections of the tissue sample at
the desired thickness for your study.
(viii) Carefully mount the sections onto the substrate window. Immediately upon acquiring the cryosection, transfer the
slides to a slide box on dry ice, wrap them in foil and store them at −80 °C to preserve the biochemical content.
(ix) Before imaging, bring slides to room temperature in a dark slide box with desiccant for several hours (minimum 3 h)
until they are dry.
 CRITICAL STEP The tissue needs to be adequately thawed before IR analysis (freezing and thawing may also damage
the structural integrity of the tissue). During the thawing process, store the sample under dark, dry conditions at room
temperature. Light exposure is only advised during the short time required for the instrumental setup; this maintains
the stability of spectral acquisition. Room lights and bright-field microscope illumination should be switched off
during measurement collection130.
(C) Cytological specimens ● TIMING 30 min + desiccation time (24 h)
(i) For formalin fixation, cellular pellets should be washed twice in PBS to remove culture medium before resuspension in
formalin solution (in which they should remain for at least 30 min). Before IR analysis, cells should be washed with
HBSS to wash out the residual phosphate ions.
 CRITICAL STEP SurePath and ThinPrep fixative solutions, used in hospital pathology laboratories, have IR signatures
in the biochemical cell–fingerprint region and should therefore be removed from the sample by sequential washes
before analysis. Alcohol-based fixatives may remove some lipids from the sample.

1784 | VOL.9 NO.8 | 2014 | nature protocols

 protocol

(ii) Resuspend the remaining cell pellet in 0.5 ml of distilled water, transfer the cells to the appropriate IR slide and
allow them to air-dry before storing in a desiccator. Cells may be transferred to a slide as microdroplets, or they
can be cytospun.
(iii) For cytospinning, take a maximum volume of 200 µl of cells in suspension (spin-fixed cells at 800g (g force =
0.0000118 × radius of rotation (mm) × r.p.m.2) for 5 min). After spinning, leave the slide to air-dry for 24 h;
the centrifugal force will have squashed the cells onto the slide, but if you try to wash the slides with water straight
away you might lose them. After this time, wash the slide with 1-ml aliquots of deionized water three times (at around
5–10 s per wash, more water can be used if found necessary and cells stick adequately). The cells will remain on the
slide and can always be washed further if traces of salts remain.
 CRITICAL STEP When transferring the cellular material to the slide, ensure that an even deposit of cells is placed on
the slide. Cytospinning allows the cells to be proportionally dispersed over the substrate. If the cells are particularly
small, they may ‘bounce’ off the slide during the spinning instead of getting stuck down. In this case, do a 5-min spin
at 400g, then another 5-min spin at 800g to ensure firm plating.
(D) Biofluids ● TIMING 10 min
(i) Biofluids (i.e., urine, serum, plasma and saliva) should be immediately stored at −80 °C in cryovials after collection
from pathology laboratories and thawed at room temperature before use.
(ii) Samples of biofluids are painted directly onto the aperture (e.g., for ATR analysis) or a standard amount is pipetted
onto suitable IR substrates (50–250 µl would be typical, but depending on the biofluid, preliminary analysis would be
© 2014 Nature America, Inc. All rights reserved.

needed).
(iii) Samples are allowed to dry before analysis.
 CRITICAL STEP Contact of the sample with the crystal is a very important parameter for ATR-FTIR analysis.
If you are using an aperture ATR-FTIR accessory, 1 µl of sample has been shown to be dry within 8 min44.

Acquisition of spectra
2| Acquire spectra by ATR-FTIR spectroscopy (option A) or transmission FTIR microspectroscopy (option B) or FPA.
A standard operating procedure for direct-drop ATR-FTIR for biofluid analysis is included in Supplementary Method 1;
this would primarily be used when very small aliquots of the sample are available. A standard operating procedure for
FTIR-FPA imaging with an Agilent 670-IR spectrometer coupled with an Agilent 620-IR microscope and FPA detector is
included in Supplementary Method 2 (this file also contains a troubleshooting section).
(A) ATR-FTIR spectroscopy ● TIMING 20 min (10 spectra)
(i) Open the instrument-operating software.
(ii) Apply instrumental settings (guidelines are described in ‘Experimental design: spectral acquisition’).
(iii) Check the path where files are to be saved; set the file name according to a previously devised file-naming convention.
(iv) Visualize the sample through the instrument digital camera to locate the region of interest from which you wish to
acquire the spectrum.
 CRITICAL STEP If the instrument has been switched off, make sure you check the interferogram signal for the
correct location and amplitude. The system may need to be re-aligned if it has been moved or if components have
been changed.
(v) Clean the ATR IRE with distilled water and dry it with tissue.
 CRITICAL STEP Make sure that the crystal is thoroughly cleaned and dried before a background acquisition.
(vi) To acquire a background spectrum, the IRE should not be in contact with the sample or slide, and it should be open to
the surrounding environment. Record a background spectrum.
 CRITICAL STEP It is very important that a background spectrum is taken before every sample. Also, a background
spectrum should be taken if atmospheric changes occur (e.g., if a door has been suddenly opened).
(vii) Place the slide in contact with the IRE.
 CRITICAL STEP Ensure that the ATR IRE is completely covered by the sample and that the minimum sample
thickness is 3–4 times the depth of penetration to ensure that there is no interference from the substrate.
(viii) Acquire a spectrum.
(B) FTIR microspectroscopy ● TIMING 1 h per sample (~12 spectra) or 6 h per sample (image map, ~72 spectra)
(i) Switch on the microscope and instrument.
(ii) Fill the detector with liquid N2.
 CRITICAL STEP If you are using a MCT detector, filling it with liquid N2 is essential; allow the detector (and
therefore the signal) to stabilize (~10 min) to an optimal peak-to-peak value. Top up with N2 every 9 h (depending
on the instrument).
(iii) Open the instrument-operating software.
(iv) Apply settings.

nature protocols | VOL.9 NO.8 | 2014 | 1785

 protocol

(v) Use the software to get a view of the slide as seen through the microscope.
(vi) Load the sample onto the stage and focus the microscope.
(vii) To check the signal quality, move to a sample-free area of the slide and adjust the position to bring the surface of the
blank area of the substrate into focus.
(viii) In our experience, the optimal sample aperture for a benchtop FTIR spectrometer with a globar source is 20 µm ×
20–100 µm × 100 µm (dependent on sample quality and instrumental limitations). Apply the aperture size.
 CRITICAL STEP Optimization of the aperture size should be performed to confirm the smallest possible aperture
that can be used to acquire spectra with a high SNR.
? TROUBLESHOOTING
(ix) Use the joystick to move the sample around the microscope stage to identify points or areas to interrogate.
(x) Select a clean, sample-free point on the slide and acquire a background spectrum according to your device.
 CRITICAL STEP Acquire a background spectrum each time the detector is filled with liquid N2 and at regular
intervals (or before each sample) to account for atmospheric changes.
(xi) Acquire a sample measurement either as a point map or as an image map.

A point map Select a number and location of points of interest

Image map Use automatic allocation of adjacent points in a grid

© 2014 Nature America, Inc. All rights reserved.

 CRITICAL STEP Be sure to define a number of points (or map size) that does not exceed the scheduled time frame
of the liquid N2 top-up.
 CRITICAL STEP The integration time is essentially a measure of the time for which the shutter is open to collect the
incoming photons. The aim is to optimize the SNR without saturating the detector. If the integration time is too high,
the user will observe saturation effects in the FTIR images; if it is too low, the data quality and SNR will be reduced
as the FPA has not been fully illuminated. This calibration is a nonuniformity correction, and results are shown with
measures of high and low flux (in counts) and the number of out-of-range pixels.
(xii) Acquire spectra.
 PAUSE POINT Once the spectra are saved they can be stored in a database until data processing.

Data pre-processing ● TIMING 15 min–4 h (depending on the size of the data set)
 CRITICAL Steps 3–7 below all contain different options at each step; however, there are combinations of these steps that
may be more or less appropriate than others, depending on the sample type, instrumentation setup, noise level, need for
visualization of spectra, personal preference and classification performance among other factors (Table 4). Although they
are usually carried out in the sequence presented, none of the steps from 3 to 7 are mandatory. For guidelines on choosing
specific preprocessing steps and options, please refer to the ‘Experimental design data processing’ section. The reader may
also refer to the Supplementary Method 3 for an illustrated example of a pre-processing sequence applied to a real-world
data set using specific software.
3| De-noise the spectra (optional, depending on the SNR of the spectra). Consider using one of the following de-noising
algorithms: Savitzky-Golay de-noising, WDN (not commonly used, but is a nonlinear method with its own advantages),
PCA noise reduction or minimum noise fraction.

4| Perform spectral correction, which can be carried out using physical theory–based methods such as RMieSC or rubber
band baseline correction49,53,113–115.

5| Perform SG differentiation (first differentiation is most used; second differentiation is also common).

6| Perform data normalization. This can be done using min-max normalization (e.g., normalization to the amide I/II peak)
or vector normalization.

7| Scale the variables: this could be done by standardization (normalization of variables to zero mean and unit s.d.) or
by normalization to a 0–1 range.

Data analysis ● TIMING 1 h–2 d (depending on file size)

8| Choose a data analysis procedure appropriate to your analysis goal; here we cover diagnosis (supervised classification;
option A below) and imaging (options B–D). For option A, data sets obtained with a single-element IR detector are normally

1786 | VOL.9 NO.8 | 2014 | nature protocols

 protocol

used and the physical location of one spectrum within its sample is used in the data analysis. The procedures for creating an
image use a data set acquired by point or array mapping, or an FPA detector. Each spectrum has its Cartesian (x, y) location.
There are three different options for achieving this step (options B–D). Options B and C are suitable for chemical imaging
using the FE method and unsupervised classification, respectively. Option D (supervised classification) is suitable for spatial
diagnosis of tissues. It uses a training set of images to build a model to subsequently apply this model to unknown images.
After performing options B, C or D, follow the instructions in option E.
(A) Diagnosis (supervised classification)
 CRITICAL Training and test data sets are required. In rare cases, one single training and test data set pair could be
enough to obtain a meaningful estimation of real-world classification; however, most of the time such estimation is obtained
through cross-validation, in which the procedure below (training and testing) is repeated multiple times to get an average
performance (or error) estimation.
(i) FE training. Input pre-processed training data set into the FE algorithm of your choice (see ‘Experimental design: data
processing’) in training mode. Generate a model that will be able to subsequently extract features from a test data set.
(ii) Classifier training. Apply the FE model obtained in a previous step to the training data set. Next, input the
FE-processed data set to the classification algorithm of your choice (see ‘Experimental design: data processing’)
in training mode. Generate a model to be subsequently applied to test data set.
(iii) Testing. Apply the trained FE model to the test data set to obtain an FE-processed output data set; input the
FE-processed data set into classification model to obtain one-class estimation per spectrum. If there are several
© 2014 Nature America, Inc. All rights reserved.

spectra per sample, conduct a ‘majority vote’ procedure to obtain one class estimation per sample.
 CRITICAL STEP Training and testing should be repeated through a cross-validation procedure, depending on the
sample size.
(B) Chemical imaging using an FE method
(i) Map each spectrum into a single scalar value. Choose an FE technique to obtain a scalar value for each spectrum
(this value will be subsequently used to address a particular color within a gradient color map). Refer to ‘Experimental
design: data processing’ for guidelines on choosing the FE method.
(ii) Continue to Step 8E.
(C) Clustering (unsupervised classification)
(i) Apply a clustering algorithm (e.g., HCA or k-means) to organize the spectra into clusters.
(ii) Assign a different integer number to each cluster and continue to Step 8E.
(D) Supervised classification for imaging
(i) Conduct histological assessment of the training set to identify different regions within the training set images; this
will be used as teaching information for the supervised learning algorithms.
(ii) Apply Step 8A(i,ii), using the training data set obtained in the previous step to obtain a classification model.
(iii) Apply the model to the test data to obtain one class estimation per spectrum in the image.
(iv) Assign a different integer number to each class and continue to Step 8E.
(E) Mapping different scalar values to different color tones
(i) Map different scalar values obtained in previous steps into different color tones. For chemical imaging (Step 8B), a
gradient color map is used (e.g., red to yellow, rainbow and so on), whereas for Step 8C and Step 8D, an indexed color
map in which each cluster or class is represented by a color of choice is suitable. Although the idea is presented here
for understanding, this step is normally carried out by imaging software.

? TROUBLESHOOTING
Optimizing the sample aperture
To overcome the problem of over- and undersampling for fine imaging, the spatial sampling area should be at least two times
larger than the (spatial) frequency of the feature under study. The step size should be equal or smaller than the aperture size
divided by 2.
For Troubleshooting for FTIR imaging with an Agilent 620 IR microscope coupled with an Agilent 670/680 IR spectrometer,
see Supplementary Method 2.

● TIMING
Step 1A, FFPE tissue: 50 min
Step 1B, snap-frozen and cryosectioned tissue samples: 120 min + drying time (3 h)
Step 1C, cytological specimens: 30 min + desiccation time (24 h)
Step 1D, biofluids: 10 min
Step 2A, ATR-FTIR spectroscopy: 20 min (10 spectra)
Step 2B, FTIR microspectroscopy: 1 h per sample (~12 spectra) or 6 h per sample (Image map, ~72 spectra)

nature protocols | VOL.9 NO.8 | 2014 | 1787

 protocol
Figure 5 | Classification rates (% classification ± s.d.) of all possible Classifier
combinations between three different pre-processing, three different feature
Pre-processing Feature extraction LDC SVM
extraction and two different supervised classifier options. Pre-processing
options: rubber band baseline correction followed by normalization to the PCA 83.25 ± 11.50 79.75 ± 10.88
RBBC →
amide I peak; first Savitzky-Golay (SG) differentiation (7 points; second FFS(MANOVA) 82.00 ± 11.50 76.42 ± 19.42
Amide I norm
order) followed by vector normalization; and second SG differentiation Identity 75.42 ± 12.66 77.25 ± 13.28
followed by vector normalization. FE options: PCA (optimization of number of
PCA 77.58 ± 15.77 80.75 ± 12.29
PCs); forward feature selection (FFS) using multivariate analysis of variance 1st diff →
FFS(MANOVA) 78.83 ± 13.71 84.25 ± 16.15
(MANOVA) P values as a criterion to including the next variable (this is similar Vector norm
to the COVAR method for optimization of number of selected features); and Identity 81.08 ± 09.98 77.75 ± 19.02

‘Identity’ (FE skipped). Supervised classifier options: linear discriminant PCA 78.17 ± 15.12 72.50 ± 17.23
classifier (LDC); and support vector machine (SVM; using Gaussian kernel; 2nd diff →
FFS(MANOVA) 82.67 ± 11.78 76.83 ± 14.69
Vector norm
optimization of the C and γ parameters133,134). The figure’s cells are gradient- Identity 77.42 ± 17.05 76.42 ± 11.35
colored according to their respective classification rate inside (yellow to red).
RBBC, rubber band baseline correction.

Steps 3–7, data pre-processing: 15 min–4 h (depending on the size of the data set)
Step 8, data analysis: 1 h–2 d (depending on the file size)

ANTICIPATED RESULTS
Preprocessing options
© 2014 Nature America, Inc. All rights reserved.

Figure 4 is a basic example that shows a set of ATR-FTIR raw spectra (cut to the 1,800–900 cm−1 region) and their
appearance after being pre-processed by different methods. Rubber band baseline correction is one of the options to remove
sloped baselines. Normalization to the amide I/II peak shifts and scales all the spectra so that their vertical minimum is at
zero and the amide I/II peak of all spectra match at the same height. To resolve overlapping bands, mathematical derivatives
are used to narrow their full width at half height value (FWHH). Narrower bandwidths (i.e., higher resolution of differential
spectra) potentially allow for subtle differences between spectra to be more easily resolved. However, each differentiation
amplifies noise and therefore the SG differentiation algorithm (with implicit de-noising) is often used. Vector normalization
is applied after differentiation to normalize the Euclidean norm of each spectrum to unity.

Classification of blood plasma

This example shows a comparison of supervised classification performance between different combinations of pre-processing,
FE and supervised classification methodologies (Fig. 5). This data set consisted of blood serum and plasma samples of
patients with ovarian cancer or endometrial cancer (n = 30 for both) and control patients without ovarian cancer (n = 30)
analyzed with ATR-FTIR spectroscopy (7 spectra per sample)41,131. The classification rate, defined as the average between
sensitivity and specificity, was used as a classification performance measure to class patients on the basis of their disease
status (i.e., ‘normal’ versus ‘cancer’). The example illustrates
High that no single pre-processing, FE or supervised classification
methodology is the absolute best, but a combination
of these may be the best solution to the problem posed.
The counterpoint to this is that different data sets may
Absorbance

require different pre-processing, FE and/or supervised

classifier methodologies, as pointed to in the machine-
learning literature103. This is evidence that different
combinations of methodologies should be attempted and
Low a b compared in any diagnostic study.

Imaging of human colon mucosa in Cytospec using

agglomerative HC and KMC
An example of imaging of human colon mucosa sections
by using agglomerative hierarchical clustering and KMC
is shown in Figure 6. The image was produced using the
Cytospec software. Figure 6c shows the original histological
image from which FTIR spectra were recorded at a spatial
c d Figure 6 | IR image reconstruction of a section of human colon mucosa.
500 µm (a) Chemical map based on the integrated absorbance of the amide I band
(1,620–1,680 cm − 1). (b) IR imaging using agglomerative HCA (six clusters).
(c) Standard histological preparation of the colonic mucosa. (d) IR map
generated on the basis of k-means clustering (15 clusters). Adapted with
permission from ref. 135.

1788 | VOL.9 NO.8 | 2014 | nature protocols

 protocol

resolution of 232 × 233 pixels. Some images (Fig. 6b,d) have been reconstructed using the multivariate methods of
agglomerative hierarchical clustering (AHC) and KMC, respectively, with both demonstrating clear differentiation of the
histological structures of the sample analyzed.

Note: Any Supplementary Information and Source Data files are available in the 13. Kole, M.R., Reddy, R.K., Schulmerich, M.V., Gelber, M.K. & Bhargava, R.
online version of the paper. Discrete frequency infrared microspectroscopy and imaging with a tunable
quantum cascade laser. Anal. Chem. 84, 10366–10372 (2012).
Acknowledgments Over several years, work in F.L.M.’s laboratories has 14. Bhargava, R. & Levin, I.W. Spectrochemical Analysis Using Infrared
been funded by the UK Engineering and Physical Sciences Research Council Multichannel Detectors (Wiley-Blackwell, 2005).
(EPSRC), the Rosemere Cancer Foundation and the UK Biotechnology and 15. Davis, B.J., Carney, P.S. & Bhargava, R. Theory of mid-infrared absorption
Biological Sciences Research Council (BBSRC). Work in R.B.’s laboratories has microspectroscopy: II. Heterogeneous samples. Anal. Chem. 82, 3487–3499
been funded by the US National Institutes of Health (grants R01CA138882 and (2010).
R01EB009745). M.J.W. acknowledges the Department of Pathology, University 16. Davis, B.J., Carney, P.S. & Bhargava, R. Theory of mid-infrared absorption
of Illinois at Chicago for funding. P.R.F. and P.G. acknowledge the EPSRC. microspectroscopy: I. Homogeneous samples. Anal. Chem. 82, 3474–3486
M.J.B. acknowledges the Rosemere Cancer Foundation, the EPSRC, Brain Tumour (2010).
North West, the Sydney Driscoll Neuroscience Foundation and the Defence 17. Filik, J., Frogley, M.D., Pijanka, J.K., Wehbe, K. & Cinque, G. Electric field
Science and Technology (Dstl). standing wave artefacts in FTIR micro-spectroscopy of biological materials.
Analyst 137, 853–861 (2012).
18. Bassan, P., Sachdeva, A., Lee, J. & Gardner, P. Substrate contributions in
AUTHOR CONTRIBUTIONS F.L.M. is the principal investigator who conceived the
micro ATR of thin samples: Implications for analysis of cells, tissue and
© 2014 Nature America, Inc. All rights reserved.

idea for the manuscript; M.J.B. and K.M.D. provided information regarding FTIR
biological fluids. Analyst 38, 4139–4146 (2013).
FPA imaging; and P.R.F. provided information regarding microfluidic devices;
19. Miljković, M., Bird, B., Lenau, K., Mazur, A.I. & Diem, M. Spectral
J.T. wrote sections regarding Data processing, ANTICIPATED RESULTS and Figures,
cytopathology: new aspects of data collection, manipulation and
as well as maintaining a working manuscript; H.J.B. wrote the Instrumentation
confounding effects. Analyst 138, 3975–3982 (2013).
and Spectral acquisition sections; K.A.H. and B.O. wrote the Sample preparation
20. Cao, J. et al. Fourier transform infrared microspectroscopy reveals that
and MATERIALS sections; R.J.S. wrote the INTRODUCTION and PROCEDURE, and
tissue culture conditions affect the macromolecular phenotype of human
contributed to the ANTICIPATED RESULTS section; C.H. provided material for
embryonic stem cells. Analyst 38, 4147–4160 (2013).
Sample preparation, PCA–k-means clustering and cross-validation; P.L. provided
21. Chan, K.L.A. & Kazarian, S.G. Aberration-free FTIR spectroscopic imaging of
figure suggestions and information regarding water vapor; B.R.W. provided
live cells in microfluidic devices. Analyst 138, 4040–4047 (2013).
information regarding live-cell imaging; R.B. provided significant revisions to
22. Kallenbach-Thieltges, A. et al. Immunohistochemistry, histopathology and
the final manuscript; M.J.B., P.B., K.M.D., N.J.F., C.H., P.L., P.L.M.-H., G.D.S.,
infrared spectral histopathology of colon cancer tissue sections.
J.S.-S., M.J.W., S.W.F., B.R.W. and P.G. all provided feedback on the manuscript;
J. Biophotonics 6, 88–100 (2013).
and F.L.M. brought together the text and finalized the manuscript.
23. Lasch, P., Haensch, W., Naumann, D. & Diem, M. Imaging of colorectal
adenocarcinoma using FT-IR microspectroscopy and cluster analysis.
COMPETING FINANCIAL INTERESTS The authors declare competing financial Biochim. Biophys. Acta 1688, 176–186 (2004).
interests: details are available in the online version of the paper. 24. Bassan, P. et al. Whole organ cross-section chemical imaging using
label-free mega-mosaic FTIR microscopy. Analyst 138, 7066–7069 (2013).
Reprints and permissions information is available online at http://www.nature. 25. Nakamura, T. et al. Microspectroscopy of spectral biomarkers associated with
com/reprints/index.html. human corneal stem cells. Mol. Vis. 16, 359–368 (2010).
26. Hammiche, A., German, M.J., Hewitt, R., Pollock, H.M. & Martin, F.L.
1. Bellisola, G. & Sorio, C. Infrared spectroscopy and microscopy in cancer Monitoring cell cycle distributions in MCF-7 cells using near-field
research and diagnosis. Am. J. Cancer Res. 2, 1–21 (2012). photothermal microspectroscopy. Biophys. J. 88, 3699–3706 (2005).
2. Diem, M., Romeo, M., Boydston-White, S., Miljkovic, M. & Matthaus, C. 27. Walsh, M.J. et al. Tracking the cell hierarchy in the human intestine using
A decade of vibrational micro-spectroscopy of human cells and tissue biochemical signatures derived by mid-infrared microspectroscopy.
(1994–2004). Analyst 129, 880–885 (2004). Stem Cell Res. 3, 15–27 (2009).
3. Griffiths, P. & De Haseth, J.A. Fourier Transform Infrared Spectrometry 2nd 28. Wood, B.R., Bambery, K.R., Evans, C.J., Quinn, M.A. & McNaughton, D.
edn. (John Wiley & Sons, 2007). A three-dimensional multivariate image processing technique for
4. Walsh, M.J. et al. FTIR microspectroscopy coupled with two-class the analysis of FTIR spectroscopic images of multiple tissue sections.
discrimination segregates markers responsible for inter-and intra-category BMC Med. Imaging 6, 12 (2006).
variance in exfoliative cervical cytology. Biomark. Insights 3, 179–189 (2008). 29. Bird, B. et al. Cytology by infrared micro-spectroscopy: Automatic distinction
5. Bhargava, R., Wall, B.G. & Koenig, J.L. Comparison of the FT-IR mapping of cell types in urinary cytology. Vib. Spectrosc. 48, 101–106 (2008).
and imaging techniques applied to polymeric systems. Appl. Spectrosc. 54, 30. Baker, M.J. et al. FTIR-based spectroscopic analysis in the identification of
470–479 (2000). clinically aggressive prostate cancer. Br. J. Cancer 99, 1859–1866 (2008).
6. Bhargava, R. Infrared spectroscopic imaging: the next generation. 31. Bird, B. et al. Infrared micro-spectral imaging: distinction of tissue types in
Appl. Spectrosc. 66, 1091–1120 (2012). axillary lymph node histology. BMC Clin. Pathol. 8, 8 (2008).
7. Colarusso, P. et al. Infrared spectroscopic imaging: from planetary to 32. Bird, B. et al. Detection of breast micro-metastases in axillary lymph nodes
cellular systems. Appl. Spectrosc. 52, 106–120 (1998). by infrared micro-spectral imaging. Analyst 134, 1067–1076 (2009).
8. German, M.J. et al. Infrared spectroscopy with multivariate analysis 33. Naumann, D. et al. Cells and biofluids analyzed in aqueous environment by
potentially facilitates the segregation of different types of prostate cell. infrared spectroscopy. Biomed. Opt. 10, 609301 (2005).
Biophys. J. 90, 3783–3795 (2006). 34. Walsh, M.J., Holton, S.E., Kajdacsy-Balla, A. & Bhargava, R. Attenuated
9. Walsh, M.J. et al. Fourier transform infrared microspectroscopy identifies total reflectance Fourier-transform infrared spectroscopic imaging for
symmetric PO2− modifications as a marker of the putative stem cell region of breast histopathology. Vib. Spectrosc. 60, 23–28 (2012).
human intestinal crypts. Stem Cells 26, 108–118 (2008). 35. Ooi, G.J. et al. Fourier transform infrared imaging and small angle X-ray
10. Kelly, J.G. et al. Derivation of a subtype-specific biochemical signature of scattering as a combined biomolecular approach to diagnosis of breast
endometrial carcinoma using synchrotron-based Fourier-transform infrared cancer. Med. Phys. 35, 2151–2161 (2008).
microspectroscopy. Cancer Lett. 274, 208–217 (2009). 36. Nallala, J. et al. Infrared imaging as a cancer diagnostic tool: Introducing a
11. Romeo, M.J. et al. in Vibrational Spectroscopy for Medical Diagnosis new concept of spectral barcodes for identifying molecular changes in colon
(eds. Diem, M., Lasch, P., & Chalmers, J.) (John Wiley & Sons, 2008). tumors. Cytometry Part A 83, 294–300 (2013).
12. Diem, M. et al. Comparison of Fourier transform infrared (FTIR) spectra of 37. Bird, B. et al. Infrared spectral histopathology (SHP): a novel diagnostic
individual cells acquired using synchrotron and conventional sources. tool for the accurate classification of lung cancer. Lab. Invest. 92,
Infrared Phys. Technol. 45, 331–338 (2004). 1358–1373 (2012).

nature protocols | VOL.9 NO.8 | 2014 | 1789

 protocol
38. Baker, M.J. et al. Investigating FTIR-based histopathology for the diagnosis 64. Liu, J.-N., Schulmerich, M.V., Bhargava, R. & Cunningham, B.T. Optimally
of prostate cancer. J. Biophotonics 2, 104–113 (2009). designed narrowband guided-mode resonance reflectance filters for mid-
39. Gazi, E. et al. A correlation of FTIR spectra derived from prostate cancer infrared spectroscopy. Opt. Express 19, 24182–24197 (2011).
biopsies with Gleason grade and tumour stage. Eur. Urol. 50, 750–761 65. Miller, L.M. & Smith, R.J. Synchrotrons versus globars, point-detectors
(2006). versus focal plane arrays: Selecting the best source and detector for specific
40. Walsh, M.J. et al. IR microspectroscopy: potential applications in cervical infrared microspectroscopy and imaging applications. Vib. Spectrosc. 38,
cancer screening. Cancer Lett. 246, 1–11 (2007). 237–240 (2005).
41. Gajjar, K. et al. Fourier-transform infrared spectroscopy coupled with a 66. Lasch, P., Boese, M., Pacifico, A. & Diem, M. FT-IR spectroscopic
classification machine for the analysis of blood plasma or serum: a novel investigations of single cells on the subcellular level. Vib. Spectrosc. 28,
diagnostic approach for ovarian cancer. Analyst 138, 3917–3926 (2013). 147–157 (2002).
42. Ollesch, J. et al. FTIR spectroscopy of biofluids revisited: an automated 67. Duncan, W. & Williams, G.P. Infrared synchrotron radiation from electron
approach to spectral biomarker identification. Analyst 138, 4092–4102 storage rings. Appl. Optics 22, 2914–2923 (1983).
(2013). 68. Pijanka, J.K. et al. Spectroscopic signatures of single, isolated cancer cell
43. Scaglia, E. et al. Noninvasive assessment of hepatic fibrosis in patients with nuclei using synchrotron infrared microscopy. Analyst 134, 1176–1181
chronic hepatitis C using serum Fourier transform infrared spectroscopy. (2009).
Anal. Bioanal. Chem. 401, 2919–2925 (2011). 69. Dumas, P., Sockalingum, G.D. & Sule-Suso, J. Adding synchrotron radiation
44. Hands, J. et al. Investigating the rapid diagnosis of gliomas from serum to infrared microspectroscopy: what’s new in biomedical applications?
samples using infrared spectroscopy and cytokine and angiogenesis factors. Trends Biotechnol. 25, 40–44 (2007).
Anal. Bioanal. Chem. 405, 7347–7355 (2013). 70. Martin, F.L. Shining a new light into molecular workings. Nat. Methods 8,
45. Munro, K.L. et al. Synchrotron radiation infrared microspectroscopy of 385–387 (2011).
arsenic-induced changes to intracellular biomolecules in live leukemia cells. 71. Menzel, L. et al. Spectroscopic detection of biological NO with a quantum
Vib. Spectrosc. 53, 39–44 (2010). cascade laser. Appl. Phys. B 72, 859–863 (2001).
46. Tobin, M.J. et al. FTIR spectroscopy of single live cells in aqueous media by 72. Valle, J.J. et al. Free electron laser-Fourier transform ion cyclotron
synchrotron IR microscopy using microfabricated sample holders. resonance mass spectrometry facility for obtaining infrared multiphoton
© 2014 Nature America, Inc. All rights reserved.

Vib. Spectrosc. 53, 34–38 (2010). dissociation spectra of gaseous ions. Rev. Sci. Instrum. 76, 023103 (2005).
47. Whelan, D.R., Bambery, K.R., Puskar, L., McNaughton, D. & Wood, B.R. 73. Llabjani, V. et al. Differential effects in mammalian cells induced by
Synchrotron Fourier transform infrared (FTIR) analysis of single living cells chemical mixtures in environmental biota as profiled using infrared
progressing through the cell cycle. Analyst 138, 3891–3899 (2013). spectroscopy. Environ. Sci. Technol. 45, 10706–10712 (2011).
48. Kuimova, M.K., Chan, K.L.A. & Kazarian, S.G. Chemical imaging of live 74. Schubert, J.M., Mazur, A.I., Bird, B., Miljković, M. & Diem, M. Single point
cancer cells in the natural aqueous environment. Appl. Spectrosc. 63, vs. mapping approach for spectral cytopathology (SCP). J. Biophotonics 3,
164–171 (2009). 588–596 (2010).
49. Llabjani, V. et al. Polybrominated diphenyl ether-associated alterations in 75. Carter, E.A., Tam, K.K., Armstrong, R.S. & Lay, P.A. Vibrational spectroscopic
cell biochemistry as determined by attenuated total reflection Fourier- mapping and imaging of tissues and cells. Biophys. Rev. 1, 95–103 (2009).
transform infrared spectroscopy: a comparison with DNA-reactive and/or 76. Nasse, M.J. et al. High-resolution Fourier-transform infrared chemical
endocrine-disrupting agents. Environ. Sci. Technol. 43, 3356–3364 (2009). imaging with multiple synchrotron beams. Nat. Methods 8, 413–416 (2011).
50. Malins, D.C. et al. Biomarkers signal contaminant effects on the organs of 77. Fernandez, D.C., Bhargava, R., Hewitt, S.M. & Levin, I.W. Infrared
English sole (Parophrys vetulus) from Puget Sound. Environ. Health Perspect. spectroscopic imaging for histopathologic recognition. Nat. Biotechnol. 23,
114, 823–829 (2006). 469–474 (2005).
51. Cakmak, G., Togan, I. & Severcan, F. 17β-Estradiol induced compositional, 78. Chan, K. & Kazarian, S. New opportunities in micro-and macro-attenuated
structural and functional changes in rainbow trout liver, revealed by FT-IR total reflection infrared spectroscopic imaging: spatial resolution and
spectroscopy: a comparative study with nonylphenol. Aquat. Toxicol. 77, sampling versatility. Appl. Spectrosc. 57, 381–389 (2003).
53–63 (2006). 79. Holton, S.E., Walsh, M.J. & Bhargava, R. Subcellular localization of early
52. Llabjani, V. et al. Alterations in the infrared spectral signature of avian biochemical transformations in cancer-activated fibroblasts using infrared
feathers reflect potential chemical exposure: A pilot study comparing two spectroscopic imaging. Analyst 136, 2953–2958 (2011).
sites in Pakistan. Environ. Int. 48, 39–46 (2012). 80. Bassan, P. et al. The inherent problem of transflection-mode infrared
53. Trevisan, J. et al. Syrian hamster embryo (SHE) assay (pH 6.7) coupled with spectroscopic microscopy and the ramifications for biomedical single point
infrared spectroscopy and chemometrics towards toxicological assessment. and imaging applications. Analyst 138, 144–157 (2013).
Analyst 135, 3266–3272 (2010). 81. Goormaghtigh, E., Raussens, V. & Ruysschaert, J.M. Attenuated total
54. Ahmadzai, A.A. et al. The Syrian hamster embryo (SHE) assay (pH 6.7): reflection infrared spectroscopy of proteins and lipids in biological
mechanisms of cell transformation and application of vibrational membranes. Biochim. Biophys. Acta 1422, 105–185 (1999).
spectroscopy to objectively score endpoint alterations. Mutagenesis 27, 82. Wehbe, K., Filik, J., Frogley, M.D. & Cinque, G. The effect of optical
257–266 (2012). substrates on micro-FTIR analysis of single mammalian cells. Anal. Bioanal.
55. Ami, D., Natalello, A., Zullini, A. & Doglia, S.M. Fourier transform infrared Chem. 405, 1311–1324 (2013).
microspectroscopy as a new tool for nematode studies. FEBS Lett. 576, 83. Vaccari, L. et al. Synchrotron radiation infrared microspectroscopy of single
297–300 (2004). living cells in microfluidic devices: advantages, disadvantages and future
56. Hobro, A.J. & Lendl, B. Fourier-transform mid-infrared FPA imaging of a perspectives. J. Phys.: Conf. Ser. 359, 012007 (2012).
complex multicellular nematode. Vib. Spectrosc. 57, 213–219 (2011). 84. Chan, K.L.A. & Kazarian, S.G. FT-IR spectroscopic imaging of reactions in
57. Mariey, L., Signolle, J.P., Amiel, C. & Travert, J. Discrimination, multiphase flow in microfluidic channels. Anal. Chem. 84, 4052–4056
classification, identification of microorganisms using FTIR spectroscopy and (2012).
chemometrics. Vib. Spectrosc. 26, 151–159 (2001). 85. Marcsisin, E.J., Uttero, C.M., Miljkovic, M. & Diem, M. Infrared
58. Gómez-De-Anda, F. et al. Determination of Trichinella spiralis in pig muscles microspectroscopy of live cells in aqueous media. Analyst 135, 3227–3232
using mid-Fourier transform infrared spectroscopy (MID-FTIR) with (2010).
attenuated total reflectance (ATR) and soft independent modeling of class 86. Nasse, M., Ratti, S., Giordano, M. & Hirschmugl, C. Demountable liquid/flow
analogy (SIMCA). Meat Sci. 91, 240–246 (2012). cell for in vivo infrared microspectroscopy of biological specimens.
59. Kazarian, S.G. & Chan, K.A. ATR-FTIR spectroscopic imaging: recent advances Appl. Spectrosc. 63, 1181–1186 (2009).
and applications to biological systems. Analyst 138, 1940–1951 (2013). 87. Holman, H.-Y.N., Bechtel, H.A., Hao, Z. & Martin, M.C. Synchrotron IR
60. Glassford, S.E., Byrne, B. & Kazarian, S.G. Recent applications of ATR spectromicroscopy: chemistry of living cells. Anal. Chem. 82, 8757–8765
FTIR spectroscopy and imaging to proteins. Biochim. Biophys. Acta 1834, (2010).
2849–2858 (2013). 88. Bhargava, R. & Levin, I.W. Fourier transform infrared imaging: theory and
61. Stuart, B. Infrared Spectroscopy: Fundamentals and Applications. John Wiley practice. Anal. Chem. 73, 5157–5167 (2001).
and Sons (2005). 89. Bhargava, R. & Levin, I.W. Effective time averaging of multiplexed
62. Mantsch, H.H. & Chapman, D. Infrared Spectroscopy of Biomolecules measurements: A critical analysis. Anal. Chem. 74, 1429–1435
(Wiley-Liss, 1996). (2002).
63. Miller, L.M. & Dumas, P. Chemical imaging of biological tissue with 90. Bhargava, R., Ribar, T. & Koenig, J.L. Towards faster FT-IR imaging by
synchrotron infrared light. Biochim. Biophys. Acta 1758, 846–857 (2006). reducing noise. Appl. Spectrosc. 53, 1313–1322 (1999).

1790 | VOL.9 NO.8 | 2014 | nature protocols

 protocol
91. Bhargava, R., Schaeberle, M.D., Fernandez, D.C. & Levin, I.W. Novel route to 120. Guyon, I., Gunn, S., Nikravesh, M. & Zadeh, L. Feature Extraction,
faster Fourier transform infrared spectroscopic imaging. Appl. Spectrosc. 55, Foundations and Applications (Springer, 2006).
1079–1084 (2001). 121. Udelhoven, T., Novozhilov, M. & Schmitt, J. The NeuroDeveloper (R): a tool
92. Gazi, E. et al. Fixation protocols for subcellular imaging by synchrotron- for modular neural classification of spectroscopic data. Chemometr. Intell.
based Fourier transform infrared microspectroscopy. Biopolymers 77, 18–30 Lab. 66, 219–226 (2003).
(2005). 122. Kwak, J.T., Reddy, R., Sinha, S. & Bhargava, R. Analysis of variance in
93. Bretzlaff, R. & Bahder, T. Apodization effects in Fourier transform infrared spectroscopic imaging data from human tissues. Anal. Chem. 84, 1063–1069
difference spectra. Rev. Phys. Appl. 21, 833–844 (1986). (2011).
94. Tahtouh, M., Despland, P., Shimmon, R., Kalman, J.R. & Reedy, B.J. The 123. Bhargava, R., Fernandez, D.C., Hewitt, S.M. & Levin, I.W. High throughput
application of infrared chemical imaging to the detection and enhancement assessment of cells and tissues: Bayesian classification of spectral metrics
of latent fingerprints: method optimization and further findings. J. Forensic from infrared vibrational spectroscopic imaging data. Biochim. Biophys. Acta
Sci. 52, 1089–1096 (2007). 1758, 830–845 (2006).
95. Lasch, P. & Naumann, D. Spatial resolution in infrared microspectroscopic 124. Naumann, D. in Encyclopedia of Analytical Chemistry (John Wiley & Sons,
imaging of tissues. Biochim. Biophys. Acta 1758, 814–829 (2006). 2000).
96. Sun, D.-W. Infrared Spectroscopy for Food Quality Analysis and Control 125. Hughes, C. et al. FTIR microspectroscopy of selected rare diverse
(Academic Press, 2009). sub-variants of carcinoma of the urinary bladder. J. Biophotonics 6, 73–87
97. Snook, R.D., Harvey, T.J., Faria, E.C. & Gardner, P. Raman tweezers and their (2013).
application to the study of singly trapped eukaryotic cells. Integr. Biol. 1, 126. Hastie, T., Tibshirani, R. & Friedman, J. The Elements of Statistical
43–52 (2009). Learning: Data Mining, Inference and Prediction 2nd edn. (Springer, 2009).
98. Bhargava, R., Fernandez, D.C., Schaeberle, M.D. & Levin, I.W. Effect of focal 127. Bhargava, R. Towards a practical Fourier transform infrared chemical
plane array cold shield aperture size on Fourier transform infrared micro- imaging protocol for cancer histopathology. Anal. Bioanal. Chem. 389,
imaging spectrometer performance. Appl. Spectrosc. 54, 1743–1750 (2000). 1155–1169 (2007).
99. Bruun, S.W. et al. Correcting attenuated total reflection-Fourier transform 128. Berenbaum, M.C. The histochemistry of bound lipids. Q. J. Microsc. Sci.
infrared spectra for water vapor and carbon dioxide. Appl. Spectrosc. 60, s3-99, 231–242 (1958).
© 2014 Nature America, Inc. All rights reserved.

1029–1039 (2006). 129. Wigglesworth, V.B. Bound lipid in the tissues of mammal and insect:
100. Lasch, P. & Petrich, W. Data acquisition and analysis in biomedical a new histochemical method. J. Cell Sci. 8, 709–725 (1971).
vibrational spectroscopy. Biomed. Appl. Sync. Infrared Microspec. 11, 130. Stitt, D.M. et al. Tissue acquisition and storage associated oxidation
192–225 (2011). considerations for FTIR microspectroscopic imaging of polyunsaturated fatty
101. Trevisan, J., Angelov, P.P., Carmichael, P.L., Scott, A.D. & Martin, F.L. acids. Vib. Spectrosc. 60, 16–22 (2012).
Extracting biological information with computational analysis of Fourier- 131. Owens, G.L. et al. Vibrational biospectroscopy coupled with multivariate
transform infrared (FTIR) biospectroscopy datasets: current practices to analysis extracts potentially diagnostic features in blood plasma/serum of
future perspectives. Analyst 137, 3202–3215 (2012). ovarian cancer patients. J. Biophotonics 7, 200–209 (2014).
102. Martin, F.L. et al. Distinguishing cell types or populations based on the 132. Dorling, K.M. & Baker, M.J. Highlighting attenuated total reflection Fourier
computational analysis of their infrared spectra. Nat. Protoc. 5, 1748–1760 (2010). transform infrared spectroscopy for rapid serum analysis. Trends Biotechnol.
103. Duda, R.O., Hart, P.E. & Stork, D.G. Pattern Classification 2nd edn. 31, 327–328 (2013).
(Wiley Interscience, 2001). 133. Chang, C.-C. & Lin, C.-J. LIBSVM: a library for support vector machines.
104. Goodacre, R. Explanatory analysis of spectroscopic data using machine ACM TIST 2, 27 (2011).
learning of simple, interpretable rules. Vib. Spectrosc. 32, 33–45 (2003). 134. Trevisan, J. et al. Measuring similarity and improving stability in biomarker
105. Naumann, D. FTIR spectroscopy of microorganisms at the Robert Koch identification methods applied to Fourier-transform infrared (FTIR)
Institute: experiences gained during successful project. Proc. SPIE 6853, spectroscopy. J. Biophotonics 7, 254–265 (2014).
Biomedical Optical Spectroscopy 68530G (2008). 135. Lasch, P., Fabian, H., Thi, N.A.N. & Naumann, D. Infrarot-bildgebung für die
106. Lasch, P. & Petrich, W. in Biomedical Applications of Synchrotron Infrared pathohistologische diagnostik. Laborwelt 2, 8–12 (2004).
Microspectroscopy: a Practical Approach, RSC Analytical Spectroscopy Series 136. Lasch, P. & Kneipp, J. Biomedical Vibrational Spectroscopy. (Wiley-Blackwell,
Vol. 11, (ed. D. Moss), 192–225 (RSC Analytical Spectroscopy Series, 2011). 2008).
107. Lasch, P. Spectral pre-processing for biomedicalvibrational spectroscopy and 137. Lyng, F., Gazi, E. & Gardner, P. in Preparation of Tissues and Cells for
microspectroscopicimaging. Chemom. Intell. Lab. Syst. 117, 100–114 (2013). Infrared and Raman Spectroscopy (RSC Analytical Spectroscopy Monographs,
108. Bhargava, R., Wang, S.-Q. & Koenig, J.L. Route to higher fidelity FT-IR No. 11, ed. Moss, D.) 147–185 (Royal Society of Chemistry, 2011).
imaging. Appl. Spectrosc. 54, 486–495 (2000). 138. Whelan, D.R., Bambery, K.R., Puskar, L., McNaughton, D. & Wood, B.R.
109. Reddy, R.K. & Bhargava, R. Accurate histopathology from low signal-to- Quantification of DNA in simple eukaryotic cells using Fourier transform
noise ratio spectroscopic imaging data. Analyst 135, 2818–2825 (2010). infrared spectroscopy. J. Biophotonics 6, 775–784 (2013).
110. van Dijk, T., Mayerich, D., Bhargava, R. & Carney, P.S. Rapid spectral-domain 139. Whelan, D.R. et al. Monitoring the reversible B- to A-like transition of DNA
localization. Opt. Express 21, 12822–12830 (2013). in eukaryotic cells using Fourier transform infrared spectroscopy.
111. van Dijk, T., Mayerich, D., Carney, P.S. & Bhargava, R. Recovery of Nucleic Acids Res. 39, 5439–5448 (2011).
absorption spectra from Fourier transform infrared (FT-IR) 140. Rahmelow, K. & Hubner, W. Phase correction in Fourier transform
microspectroscopic measurements of intact spheres. Appl. Spectrosc. 67, spectroscopy: subsequent displacement correction and error limit.
546–552 (2013). Appl. Optics 36, 6678–6686 (1997).
112. Martens, H. & Stark, E. Extended multiplicative signal correction and 141. Gazi, E. et al. The combined application of FTIR microspectroscopy and
spectral interference subtraction - new pre-processing methods for ToF-SIMS imaging in the study of prostate cancer. Faraday Discuss. 126,
near-infrared spectroscopy. J. Pharmaceut. Biomed. 9, 625–635 (1991). 41–59 (2004).
113. Bassan, P. et al. Resonant Mie scattering in infrared spectroscopy of 142. Patel, I.I. et al. Isolating stem cells in the inter-follicular epidermis
biological materials - understanding the ‘dispersion artefact’. Analyst 134, employing synchrotron radiation-based Fourier-transform infrared
1586–1593 (2009). microspectroscopy and focal plane array imaging. Anal. Bioanal. Chem. 404,
114. Bassan, P. et al. Resonant Mie scattering (RMieS) correction of infrared spectra 1745–1758 (2012).
from highly scattering biological samples. Analyst 135, 268–277 (2010). 143. Bassan, P. et al. FTIR microscopy of biological cells and tissue: data analysis
115. Bassan, P. et al. RMieS-EMSC correction for infrared spectra of biological using resonant Mie scattering (RMieS) EMSC algorithm. Analyst 137,
cells: extension using full Mie theory and GPU computing. J. Biophotonics 3, 1370–1377 (2012).
609–620 (2010). 144. Kastyak-Ibrahim, M. et al. Biochemical label-free tissue imaging with
116. Opus 5 Reference Manual (Bruker Optik, 2004). subcellular-resolution synchrotron FTIR with focal plane array detector.
117. Ly, E. et al. Combination of FTIR spectral imaging and chemometrics for tumour NeuroImage 60, 376–383 (2012).
detection from paraffin-embedded biopsies. Analyst 133, 197–205 (2008). 145. Gajjar, K. et al. Diagnostic segregation of human brain tumours using
118. Beier, B.D. & Berger, A.J. Method for automated background subtraction Fourier-transform infrared and/or Raman spectroscopy coupled with
from Raman spectra containing known contaminants. Analyst 134, discriminant analysis. Anal. Methods 5, 89–102 (2013).
1198–1202 (2009). 146. Kelly, J.G., Martin-Hirsch, P.L. & Martin, F.L. Discrimination of base
119. Baker, M. et al. An investigation of the RWPE prostate derived family of cell differences in oligonucleotides using mid-infrared spectroscopy and
lines using FTIR spectroscopy. Analyst 135, 887–894 (2010). multivariate analysis. Anal. Chem. 81, 5314–5319 (2009).

nature protocols | VOL.9 NO.8 | 2014 | 1791