0003-2654(2013)138:14;1-X Analyst CRITICAL REVIEW

The discovery of the Raman effect in 1928 not only aided fundamental understanding about the quantum nature of light and matter but also opened up a completely novel area of optics and spectroscopic research that is accelerating at a greater rate during the last decade than at any time since its inception. This introductory overview focuses on some of the most recent developments within this exciting field and how this has enabled and enhanced disease diagnosis and biomedical applications. We highlight a small number of stimulating high-impact studies in imaging, endoscopy, stem cell research, and other recent developments such as spatially offset Raman scattering amongst others. We hope this stimulates further interest in this already exciting field, by 'illuminating' some of the current research being undertaken by the latest in a very long line of dedicated experimentalists interested in the properties and potential beneficial applications of light.


Introduction
The properties of light have been of interest to experimentalists for millennia.From the publication of Ibn al-Haytham's seven volume treatise the 'Book of Optics' (Kitāb al-Manā_ zir) in the early 11th century (leading to him being regarded as the father of modern optics 1 ), through to one of the very rst publications in a scientic journal, Isaac Newton's paper on the theory of the properties of light 2 itself followed some years later by his famous book, Opticks, in 1704. 3Whilst separated by seven centuries, these two polymaths and their respective bodies of work shared similarities, perhaps the most important being that both were the result of systematic and methodical experimentation.In the century following the publication of Newton's David Ellis was educated on the Welsh coast at the University of Wales, Aberystwyth, obtaining a BSc in Environmental Science and a PhD in Analytical Biotechnology/Microbiology.His research involving the rapid and quantitative detection of foodborne bacteria using FT-IR and machine learning has been widely publicised, featuring on BBC TV and radio, the Science Museum in London, as well as the national and international press (e.g.WIRED).He now works at the Manchester Institute of Biotechnology (MIB), as a Senior Experimental Officer undertaking research and managing the labs of Roy Goodacre (http://biospec.net)and Doug Kell (http:// dbkgroup.org), in the School of Chemistry, University of Manchester, UK.David Cowcher obtained an undergraduate Masters degree in Chemistry and Forensic Science in 2009 from the University of Manchester, which included a year working as an analytical chemist for Glax-oSmithKline.He is currently studying for a PhD with Roy Goodacre at the Manchester Institute of Biotechnology, University of Manchester, on the development of enhanced Raman scattering methods for biochemical analysis.
Opticks, as early as the mid-1800s, the interaction of light within tissue was being used by physicians to assist in disease recognition. 4t was during a sea voyage from India to England in 1921 that the brilliant Indian physicist C.V. Raman, another renowned experimentalist, undertook some on-board experiments which were later submitted to Nature in a letter called the 'The Colour of the Sea'. 5 Unable to accept Lord Rayleigh's 6 explanation that the colour of the sea was just a reection of the colour of the sky, 7 Raman's experiments showed that the colour of the sea was in fact a direct result of the molecular scattering of light and independent of absorption or the reection of light from the sky. 8This was very closely followed by another letter to Nature concerning the molecular scattering of light in liquids and solids. 9These experiments opened up a deep interest in C.V. Raman and a new eld of research on his return to Kolkata (Calcutta) on the scattering of light, 5 as well as the publication of another article on the molecular diffraction and quantum structure of light in the following year. 10aman and collaborators such as K.S. Krishnan began a series of seminal experiments concerning the scattering of light in a large number of liquids, as well as theories about the potential applications of their experiments, which culminated in their discovery of the inelastic scattering effect named aer Raman in 1928 on 28 February, 11 and his award of the Nobel Prize for Physics in 1930.This discovery was also independently observed by Landsberg and Mandelstam later in 1928 (see Fig. 1 for a timeline of events in Raman spectroscopy).There was a great deal of interest in the Raman effect and this not only aided the fundamental understanding about the quantum nature of light, and its interaction with matter at the molecular level, but also opened up a completely novel area of optics and spectroscopic research that is, particularly in terms of biological and biomedical applications, 12,13 accelerating at a greater rate during the last decade than at any time since its inception.
Whereas infrared (IR) spectroscopies measure the absorption of energy, Raman spectroscopy measures the exchange of energy with electromagnetic (EM) radiation of a particular wavelength, usually provided by a monochromatic light source such as a laser in the visible to near-IR portion of the EM; although it is also possible to conduct experiments in the nearand deep-UV.5][16] It is usually the Stokes shi which is measured, as Analyst Critical Review this has a higher probability of occurring than anti-Stokes. 17It is important to note that this shi is complementary to IR absorption and a spectroscopic 'ngerprint' 13,17,18 of the same sample can be analysed and constructed by both vibrational spectroscopies.Whilst mid-IR spectroscopy is known to be intensely sensitive to and a high absorber of water, this is generally not the case with Raman spectroscopy as water is a weak scatterer. 13For biomedical and routine clinical applications (and with low laser output power at the point of contact), this allows for the direct collection in vivo of Raman spectra.The utility of Raman spectroscopy has been demonstrated for a diverse and wide range of potential biological and biomedical applications, such as bacterial identication, 19,20 chemical hazards and illicit substance detection, 21,22 as well as food and product authentication, 23,24 with a great deal of interest and research into its potential for disease diagnosis and use in biomedical applications seen during the last decade (Fig. 2).Some of the previously documented limitations of Raman spectroscopy for biomedical applications have for example included issues such as weak scattering signals, subsequently  long spectral acquisition times, uorescence from biological samples, and interference from silica within bre-optics.However, within this exciting and highly research active era of biomedical Raman spectroscopy, and with the hard work and application of experimentalists (in the long tradition of those already mentioned above), these so-called limitations are being overcome by groups of scientists and engineers around the globe who are not content to remain within the bounds of current knowledge, or the connes of commercially available optics for that matter, and who are constantly pushing the eld forward.This introductory overview focuses on some of the recent developments within this exciting eld, highlighting a small number of high-impact studies in imaging, endoscopy, stem cell research, and other recent developments such as spatially offset Raman scattering (SORS), coherent anti-Stokes Raman scattering (CARS) and stimulated Raman scattering (SRS), amongst others.

Imaging
Biomedical imaging is an extremely useful and important eld which allows for the collection and processing of highly complex biochemical and physiological data, and the creation, manipulation, and in-depth analysis of three-dimensional colourised images to aid molecular sensing, 25 drug transport, 26 characterization of cells 27 and tissues, 28 and, perhaps the ultimate goal, the rapid diagnosis of disease (vide infra).Label-free optical imaging, particularly in vivo, would be highly advantageous as dyes or uorescent labels, which are needed as contrast agents, can be toxic or perturbative to the cell/tissue.Optical techniques have also been said to have the potential to be complementary to existing techniques such as magnetic resonance imaging (MRI), and they offer superior sensitivity and high spatial resolution (compared to MRI, see Fig. 3). 29Whilst mass spectrometry (MS) has mass appeal for metabolomics for example 30 and can be used for chemical imaging of tissues and cells 31,32 including in 3D 33 as well as being combined with vibrational spectroscopy, 34 imaging MS is a destructive analysis and the practical spatial resolution for MALDI-MS is 50-200 mm and for SIMS is 1 mm, thus Raman spectroscopy has considerable opportunities to shed light on disease.
In any review of biomedical imaging, and particularly one focussed on Raman spectroscopy, it would be a heedless omission not to include any one of a number of studies by the Sunney Xie group.A major contributor to biomedical imaging, based at Harvard University (http://bernstein.harvard.edu),this group's body of work includes both coherent Raman scattering techniques, 35 comprising of coherent anti-Stokes Raman scattering (CARS) microscopy [36][37][38] and stimulated Raman scattering (SRS) microscopy. 39Descriptions of these techniques can be found in Table 1 and it should also be noted that the signal from CARS is non-linearly proportional to species concentration whilst the signal from SRS is linearly proportional.

Analyst Critical Review
As briey mentioned above, the signal from spontaneous Raman scattering is known to be weak, with potentially long integration times, which impacts by signicantly reducing imaging speed (an important factor as in vivo samples are in constant motion at the micro scale).Although these nonlinear methods were rst invented in the 1960s, coherent Raman scattering techniques can be said to have very recently surmounted any technical difficulties and imaging speed hurdles by enhancing the Raman signal level by up to ve orders of magnitude, 39 and image collection speeds by three orders of magnitude.Indeed, on the subject of speed, one of the more memorable articles propelling CARS microscopy as a rapid method over a decade ago (for the identication of bacterial spores), had the unforgettable acronym FAST-CARS (femtosecond adaptive spectroscopic techniques for coherent anti-Stokes Raman spectroscopy). 40ore recent reports have highlighted some of the potential limitations of CARS, 29 which has been said to make quantitative interpretation and applications other than lipid imaging 41 unnecessarily challenging.These limitations can be particularly noticeable in the ngerprint region where spectral signals are said to be congested 39 and separating out weak from strong signals in this region is thus more problematic for CARS microscopic imaging.That being said, CARS microscopy has been successfully demonstrated as a biomedical imaging method for a range of ex vivo biological samples such as brain structures, 38 colon tissue, 42 and arterial tissue 27 for example, as well as in vivo studies including sciatic nerve tissue 43 and atherosclerotic plaque deposits 44 in animals, with a recent report also demonstrating the use of CARS microscopy in skin imaging of humans in vivo. 45ecent articles within the last half decade have explored SRS as an alternative imaging technology, which, unlike CARS, does not exhibit a nonresonant background and so is inherently more quantitative, though like other nonlinear multiphoton techniques (including CARS), does allow for sensitive, high spatial resolution 3D imaging. 46Whilst initial reports demonstrated stimulated Raman scattering's advantages over CARS (e.g. a lack of nonresonant background complications), it was said to not be suitable for bioimaging due to sample photodamage from excessive laser power 46 and had a limited spatial and spectral resolution as well as a slower image acquisition rate than CARS. 47These challenges were rapidly overcome by the Xie group the following year using a multifaceted approach which lowered peak rates by three orders of magnitude!(resulting in no photodamage), optimised spectral resolution, increased sensitivity by four orders of magnitude than the previous year's report and surpassed the detection limit previously stated for CARS microscopy.

Critical Review Analyst
Not content with these and other leaps forward, three bioimaging applications were then elegantly presented by Xie and co-workers.The rst application monitored and imaged the uptake of polyunsaturated omega-3 fatty acids by living human cancer cells, specically eicosapentaenoic acid (EPA).Using the Raman band at 3015 cm À1 (attributable to unsaturated fatty acids), and in contrast, the Raman band at 2920 cm À1 whose peak intensity is similar for saturated and unsaturated fatty acids, they were able to use SRS to follow the uptake of EPA by living cells, concluding that EPA is taken up by the cells and more strongly enriched in lipid droplets compared to other cellular organelles. 46The second application presented the potential for tissue imaging without the requirement for staining, highlighting the 3D sectioning capability and subcellular resolution of SRS, using the CH 2 stretching vibration at 2845 cm À1 (see Table 2 for Raman band frequencies of biological interest).
This was demonstrated in a variety of mouse tissues; from neuron bundles in corpus callosum (highlighting myelin sheaths), thick brain tissue, and from depth proles of ear tissue, in addition to comparative SRS and CARS images of stratum corneum (visually illustrating how the nonresonant background from CARS can complicate image interpretation).Finally, this seminal series of experiments showed the use of SRS to monitor drug delivery, namely deuterated dimethyl sulfoxide (DMSO), a skin penetration enhancer and retinoic acid (RA) used to treat a range of skin conditions and acute promyelocytic leukemia.Drug delivery into fresh mouse skin was monitored by tuning into the vibrations for DMSO at 670 cm À1 and RA at 1570 cm À1 (as well as lipids of subcutaneous fats at 2845 cm À1 ). 46This series of experiments elegantly demonstrated the potential for SRS as a new approach for studying pharmacokinetics in situ, sensitive labelfree imaging, and molecular sensing in 3D in living cells and tissue.
Whilst the potential of SRS had been so ably demonstrated, the series of experiments above had been undertaken ex vivo.Well aware of the importance of in vivo optical imaging in biomedicine, and perhaps acutely mindful of the challenges required to achieve this goal, as well as the fact that SRS imaging had not yet been accomplished in vivo (in animals or humans), an article was published in 2010 demonstrating that in vivo SRS was not only possible (both in mice and humans), but that it could also be performed at video-rate speeds. 29This high-speed imaging was achieved by modulating the intensity of the Stokes beam at 20 MHz and the group building their own all-analogue lock-in amplier with a response time of $100 ns (whereas this was previously limited by commercially available 100 ms lock-in ampliers). 29This new custom-built system raster scanned across a sample with a line rate of 8 kHz (100 ns per pixel at 512 Â 512 pixels with up to 25 frames per s), increasing imaging speed by three orders of magnitude.In addition, to signicantly increase the collection of backscattered light from in vivo samples further in-house modications were undertaken, such as changing the geometry of the photodetector and microscope objective.This involved exciting light from the objective through an aperture in the detector placed between the microscope objective and the sample, and specially designing a lter to block the modulated Stokes beam whilst transmitting the pump beam. 29

Analyst Critical Review
The effectiveness of this new video-rate SRS imaging technique was then demonstrated through a series of experiments such as skin imaging in living mice.The CH 3 stretching vibration at 2950 cm À1 was shown to mainly highlight proteins, as well as residual lipid structures and individual red blood cells in the viable epidermis, whilst the water signal at 3250 cm À1 was coincident with sebaceous glands in both positive and negative contrast.In vivo imaging of drug penetration of trans-retinol was demonstrated in living mouse skin, visualising with SRS imaging and 3D depth projection that penetration of the topically applied drug occurs along the hair sha.This pathway was not observed in previous experiments with excised fresh tissue, 46 which was said to further highlight the importance of in vivo imaging, as it can lead to insights into the transport mechanisms of small molecules in living organisms.
Finally, a series of in vivo SRS imaging experiments were performed on human skin of a volunteer's forearm showing cell layers of the viable epidermis to a depth of 50 mm, and the boundary (via observation of varying nuclear sizes) between the viable epidermis and stratum corneum when tuned into the CH 3 stretching vibration at 2950 cm À1 .The penetration-enhancing small molecule DMSO (which was deuterated) was applied to the skin, and its accumulation in the hair sha and lack of complete penetration into the hair itself was imaged using the C-D stretching vibration at 2125 cm À1 .With image acquisition times of 150 ms and 37 ms respectively, these blur-free high-speed in vivo vibrational imaging experiments demonstrated the diagnostic potential of this technology in humans. 29

Stem cells
Stem cells, 48 stem cell therapy, 49,50 and stem cell engineering 51 are extremely important and highly topical areas of scientic research with huge potential benets for the treatment of a wide range of diseases and biomedical applications.The ability to identify the phenotypic purity of live cells is absolutely crucial (and a noninvasive optical method would be highly benecial), as excessive proliferation of unwanted phenotypes (i.e., uncontrolled differentiation) following transplantation could result in undesirable consequences, such as tissue overgrowth and tumour formation for example. 52,53One of the recent studies by the Notingher group applied Raman spectroscopy to determine if it could be used as a label-free, noninvasive method to detect and image intrinsic chemical differences that could be used as molecular markers among highly heterogeneous stem cells populations; specically, cardiomyocytes (CMs) derived from human embryonic stem cells (hESCs). 52sing a custom-built instrument, Raman detection and imaging of molecular markers specic to hESC-derived CMs was carried out, along with retrospective phenotypic identication of all cells via immunouorescence imaging integrated with the Raman microscope.Multivariate statistical analysis of Raman spectra and cross-validation methods were used to develop a model (from 50 CMs and 40 non-CMs within the same heterogeneous populations) to determine the true accuracy of phenotypic identication of CMs, the sensitivity and specicity parameters, and select discriminatory Raman bands.Raman spectral images corresponding to the Raman bands identied by both the multivariate model and immunostaining of the same cells allowed for the accurate assignment of Raman molecular markers.The conclusions drawn from the results were that spectral differences were mainly attributable to glycogen and myobrils, with glycogen being responsible for discrimination of CMs (with a band assignment at 860 cm À1 ), and myobril proteins providing a lesser contribution (with a band assignment at 938 cm À1 ). 52This study demonstrated the potential of Raman spectroscopy for noninvasive phenotypic identication of stem cells, though the authors themselves stated that it was not yet practical for medical applications due to their long spectral acquisition times.
However, a later study by the same group on the same hESCderived cardiomyocytes reduced spectral acquisition times by a hundredfold, from minutes per cell to 5 s per cell, without the need for raster scanning.When incorporating high-powered commercial lasers, this could be further reduced to cell sorting speeds of approximately 10 cells per s. 54In addition, a recent study by the same group has successfully applied Raman to identify, image and quantify the differentiation status of live neural stem cells in vitro, where this time the spectral differences were said to be related to cytoplasmic RNA. 55A number of other interesting studies involving Raman analysis of stem cell populations have also appeared in the literature very recently [56][57][58][59] and this area of interest, not surprisingly, continues to grow and ourish.
As well as cellular differentiation, the location of specic drugs inside cells is also of interest as this may allow the elucidation of the pharmaceutical's site of action.A recent study addressing this enabled mapping the site of action of the HIV protease inhibitors indinavir and lopinavir in cervical carcinoma cells expressing the E6 oncogene from human papilloma virus (HPV). 60This study demonstrated that indinavir undergoes enhanced nuclear accumulation in E6 expressing cells, indicating this as the site of action for this compound against the HPV.Further interesting studies showing the plethora of Raman imaging applications include a report on stimulated Raman photoacoustic imaging, 61 surface enhanced Raman scattering (SERS) imaging using nanotags in live mice, as a potential multiplexed imaging detection method for multiple biomarkers in living subjects associated with a specic disease, 62 quantitative multiplex SRS imaging, 63 noninvasive time-course imaging of apoptotic cells, 64 and multivariate image reconstruction methods for Raman hyperspectral datasets. 65For those specically interested in coherent nonlinear optical imaging, which includes stimulated Raman scattering, the reader is directed to a recent review by Min et al., 66 others include a review of CARS microscopy, 37 as well as gold nanoparticles and imaging in medicine. 67

Endoscopy
The clinical potential for in vivo Raman endoscopy has been the subject of research for over a decade, since the rst published report of in vivo Raman spectra of human gastrointestinal tissue measured during routine clinical endoscopy in 2000. 68 The culmination of this study was a custom-built confocal Raman probe constructed by the group.This novel probe had been reported by the group previously, 72 although by the time of the latest study, the probe had undergone some modications which had improved its performance for spectral acquisition during oesophageal endoscopy.The 90 cm long, mm diameter bre-optic probe was designed to t into the instrument channel of a standard clinical endoscope and to have direct contact with oesophageal epithelial tissue.The optics had been modied by incorporating a graded index lens at the tip (and the output power regulated to 60 mW), so that a sampling depth of 100 to 200 mm could be ensured.Collecting spectra from this depth meant that signals from deeper tissue structures would not obscure those collected from mucosal abnormalities such as early neoplastic changes, 73 and that these abnormalities could be quickly classied in timescales suitable to a clinical setting.The performance of the probe ex vivo had been evaluated for translational use for in vivo sample collection, with low laser power at the probe tip and short spectral acquisition times said to enable its routine use for oesophageal endoscopy and paving the way for in vivo clinical trials. 73This group have recently published a review on in vitro and in vivo Raman spectroscopy as a potential routine tool for the rapid, noninvasive, early diagnosis of lesions and preventing development of cancer in the oesophagus. 74nother group who have made a signicant contribution to the eld of Raman endoscopy in recent years is headed by Zhiwei Huang and based at the Optical Imaging Laboratory in Singapore (http://www.bioeng.nus.edu.sg/optbioimaging/huang/).Several of these studies utilised image-guided Raman endoscopy, which the group reported for the rst time in 2009. 75hilst image-guided endoscopy is by no means novel, 76,77 the integration of image-guided techniques with Raman endoscopy is relatively recent.The rst report on this technique involved integrating Raman spectroscopy with trimodal imaging techniques (white-light reectance, autouorescence and narrowband) and the development of a novel 1.8 mm Raman probe which ltered out interference from uorescence as well as interference from silica from within optical bres.This was demonstrated via the rapid collection of Raman spectra (<1 s) and the corresponding endoscopic images of different locations of the upper gastrointestinal tract of a healthy volunteer in realtime and in vivo. 75nce then this group have published several articles using image-guided Raman endoscopy, demonstrating its potential as an in vivo real-time detection method for a variety of diagnostic applications.These studies have included, perhaps not surprisingly, the in vivo diagnosis of oesophageal cancer using this technique in conjunction with biomolecular modelling. 78his involved collecting spectra from 75 oesophageal tissue sites from 27 patients of normal tissue (squamous mucosa) and malignant tumours.The cancerous tissue was said to show distinct Raman signals mainly associated with cell proliferation, lipid reduction, abnormal nuclear activity and neovasculation.To estimate the biochemical composition of oesophageal tissue, biomolecular modelling was employed using six basis reference spectra from actin, collagen type I, DNA, histones, triolein and glycogen.This allowed for the construction of a linear discriminant analysis (LDA) model with a sensitivity of 97% and specicity of 95.2% for the in vivo diagnosis of oesophageal cancer. 78These results have since been said to be extremely promising, but that image-guided Raman endoscopy is yet to be used to detect dysplasia or the early onset of cancer. 73evertheless, the Huang group have previously demonstrated the potential of Raman spectroscopy for the detection of dysplasia, with a ball-lens bre-optic probe, in the high wavenumber region (HW) (2800-3700 cm À1 ) for the in vivo detection of cervical dysplasia (a HW Raman probe was rst presented by Gerwin Puppels and co-workers for in vitro measurements of brain tissue in 2005 79 ).The perceived main advantages of HW Raman were said to be a signicant reduction in uorescence and background signal from optical bres, more intense Raman signals (compared to the ngerprint region), and the possibility of an unltered single bre Raman probe design for in vivo clinical use. 80][83] The in vivo detection of epithelial neoplasia in the stomach using image-guided Raman endoscopy has also been demonstrated, and signicant differences between normal and cancerous gastric tissue were reported. 84More recently, a study has been published showing the development of an online automated spectral diagnostics system integrated with imageguided Raman endoscopy for real-time in vivo diagnosis during endoscopic examination. 85This system was built on a database of 2465 normal and 283 cancerous gastric tissue spectra acquired from 305 patients, with the system employing a variety of diagnostic algorithms.Other so-called in vivo realtime applications recently reported have involved transnasal image-guided Raman endoscopy of the larynx and nasopharynx, which the authors hoped would pave the way for realizing early diagnosis and detection of cancers and precancers of the head and neck. 86All of these studies show promise and many have similarities in terms of rapid spectral acquisition times (absolutely crucial in in vivo studies), and novel developments in optical engineering, but there remains more to accomplish, such as the ability not only to discriminate cancer, but to classify and grade both cancerous and precancerous cells and tissue.The central difference in SORS is that Raman spectra are collected from different locations, spatially separated (offset) from the point of laser excitation on the sample (Fig. 4).As a consequence of photon diffusion processes within tissue the resultant Raman spectra contain different relative contributions from different depths within the sample and thus allows for the highly accurate chemical analysis of subsurface objects of interest. 89These spectra acquired from different spatial offsets are processed to reveal pure Raman spectra of subcomponents from separate depth locations within tissue. 91SORS can be applied in a number of Raman collection and beam delivery geometries including single point collection, ring, and other pattern illumination. 91In respect to biomedical imaging this approach is said to be particularly useful when congured as inverse SORS, where the sample is illuminated by an adjustable ring-shaped laser beam (generated by a conical (axicon) lens) and the Raman light collected via bres in the centre of the ring. 92The radius of this ring is said to dene the spatial offset and as it is adjustable, this can be optimised to suit both the scattering properties and dimensions of each sample and covers a wider illumination zone on a sample surface than conventional SORS. 91SORS is said to be effective at tissue depths in excess of 500 mm, which is well beyond the accessible range of conventional confocal Raman spectroscopy.Applications can be as diverse and wide-ranging as noninvasive detection of pharmaceuticals through packaging, 93 detection of hidden explosives and drug precursors behind opaque plastics and garments, 21,[94][95][96] as well as agricultural and food product analysis. 97rom the outset, Matousek and collaborators immediately realised the diagnostic potential of SORS with its ability for noninvasive subsurface probing of for example bone through skin 89 and dermatology studies. 92This foresight was quickly realised with the rst SORS spectra of ex vivo bone collected in 2006 98,99 from animal and human cadavers, and in vivo SORS demonstrated by Matousek during the same year. 92There have been several studies since including accurate in vivo assessment of bone composition using the carbonate (1070 cm À1 ) -tophosphate (958 cm À1 ) ratio through the skin of live mice, 100 in vivo evaluation of bone gras, 101,102 transcutaneous in vivo monitoring of glucocorticoid induced osteoarthritis 103 and in vivo measurement and evaluation of subtle changes in bone composition. 104 signicant body of pioneering work forwarding Raman spectroscopy as a tool for breast cancer diagnosis was undertaken in the George R. Harrison Spectroscopy Lab at MIT (http:// web.mit.edu/spectroscopy/)under the directorship of Michael S. Feld.This included identifying chemical differences in microcalcications from benign and malignant breast lesions, 105,106 demonstrating the real time capabilities of an in vivo Raman system during femoral bypass and breast lumpectomy surgeries, 107 and also involved testing of spectral diagnostic algorithms for breast cancer diagnosis. 108The results from the Raman research on the chemical composition and identication of the different types of microcalcications has led directly to others investigating the potential of SORS as a possible clinical adjunct to mammography, for the noninvasive diagnosis of breast cancer.As the changing concentration of carbonate substitution for phosphate ions in the calcium hydroxide lattice in microcalcications may relate to the process of tumour cell metastasis, and the ability to measure the magnitude of this (as well as so tissue signals) by Raman, could indicate the potential progression of this cancer. 109esearch by Stone, Matousek, and collaborators, demonstrated the proof-of-principle of SORS for potential in vivo breast cancer diagnosis in a model system using three calcication standards overlaid with various preparations of animal tissue (i.e.chicken breast tissue, with and without skin).For this work they utilised a continuous wave 827 nm laser with a spatial offset for collection of 3 mm, which enabled the probing and collection of Raman spectra from calcications through up to 10 mm of tissue. 109Previous results by this group using Kerrgated Raman techniques had achieved penetration depths of 1 mm in comparison. 90This SORS study demonstrated the collection of high quality Raman spectra and biochemical information measured through 8.7 mm of tissue, identifying the difference between three calcication standards.
Subsequent experiments by the same group applied transmission Raman spectroscopy in combination with chemometrics in similar model systems collecting Raman calcication signals from depths of 20 mm, this time through porcine tissue, as a breast tissue 'phantom'. 110This was said to reach the lower range for clinically relevant breast tissue thicknesses from Critical Review Analyst mammographic screening (1.9 cm), with further work said to be required to reach the top of this range (5 cm). 91Whilst these studies utilised the carbonate-to-phosphate band ratio (similar to the SORS bone studies above 100 ), a more recent study showed that the intense Raman phosphate band at 960 cm À1 broadens and shis as carbonate concentration (from calcium hydroxyapatites) increases in the calcications. 111This study was said to pave the way toward a new generation of noninvasive breast cancer screening methods based around SORS and transmission Raman spectroscopy. 111Some very recent studies by the Mahadevan-Jansen group have investigated SORS for the realtime, intraoperative assessment of breast cancer tumour margins. 112This same group then developed and tested a SORS probe with multiple source detector offset limits specic to this type of analysis, and published results acquired from 35 freezethaw breast cancer samples in vitro. 113ariants of SORS have also recently been used to augment analysis and include in vivo transcutaneous glucose sensing in rats 114 and further demonstrated the accuracy and functionality of this in a later study. 115These studies employed what they term surface enhanced SORS (SESORS), rst published in 2010, 116 and presented at the Federation of Analytical Chemistry and Spectroscopy Societies (FACSS) meeting in the USA during the same year.SESORS marry SERS techniques using nanoparticles and nanosurfaces with the subsurface probing of SORS, with a further study exploring this technique's potential for Raman imaging. 117n the subject of SERS, which has not been mentioned in any great detail here, many studies have appeared in the literature during the last few years related to disease detection and we would like to highlight just a modest selection of these.They of course include the multiplexed imaging of SERS nanotags in vivo already mentioned above, 62 as well as other multiplexing studies, such as multiplex single nucleotide polymorphism (SNPs) genotyping coupled to SERS, 118 multiplexed in vivo cancer detection using SERS NIR nanotags, demonstrating the excellent sensitivity, stability and tumour specicity of three bioconjugated nanotags, 119 recent reviews on the area of SERS multiplexed detection for disease diagnostics, 120 as well as SERS cancer detection and imaging and the potential of SERS agents for targeted drug delivery and photothermal therapy. 121everal SERS studies involving immunoassays, including cancer detection, 122 detection of a potential pancreatic cancer marker in serum, 123 and on-chip immunoassays using hollow gold nanospheres. 124Fluorescent SERS gold co-functionalized nanorod probes for in vivo imaging of lymph node mapping and tumor targeting in mice, 125 high sensitivity in vivo detection of inammation using gold nanoclusters conjugated to monoclonal antibodies 126 and the use of functionalised nanoparticles and SERS for the detection of DNA relating to disease. 127With the recent interest concerning the potential resurgence in microbial disease, as well as bioterrorism, an article demonstrating a portable quantitative SERS system for detection of Bacillus spores at levels signicantly lower than those previously reported for SERS has been published. 128Fig. 5 shows SERS spectra of a biomarker from the spores of Bacillus, a species of which Bacillus anthracis is the cause of the acute, and mostly lethal, disease anthrax.The detection limit of this dipicolinic acid biomarker is estimated at 5 ppb (30 nM). 128n addition to those primary articles and reviews already mentioned above, other recently published reviews and books on an array of Raman-based topics which may be of interest include those concerning in vivo and in vitro analysis, 129 Raman scattering in pathology, 130 optical tumour margin identication in the larynx, 131 data-classication algorithms for spectral analysis, 132 recent advances in gold nanoparticle based assays for detecting and identifying microbes, 133 and emerging Raman applications and techniques in biomedical and pharmaceutical elds. 134

Concluding remarks
We hope that whilst only an introductory overview to some of the more recent work in this eld, the range and scope of the studies shown here elegantly demonstrate the development and exploitation of Raman spectroscopy as a medical diagnostic tool.We believe this truly is an exciting eld, where those at the forefront are propelling it forwards and are simply not content to remain within the connes of current knowledge or commercially available optics.These studies highlight the necessary interdisciplinary nature of this eld, with inspirational contributions from a range of analytical scientists and technologists including clinicians, biologists, chemists, optical engineers, as well as statisticians and chemometricians.These multidisciplinary inputs and fresh approaches are invigorating this eld, unlocking the doors to new insights, 135 adding to knowledge, and opening up new dimensions and avenues of study, as well as potential clinical applications, for biomedical Raman spectroscopy.
Ashton studied for her BSc with the Open University before joining the University of Manchester where she obtained her PhD in Biomolecular Sciences and Raman Optical Activity.She has since worked as a post doctoral research associate specialising in Raman Spectroscopy and two-dimensional correlation analysis for the characterisation of conformational transitions in biomolecules.Lorna is currently working as part of a BBSRC funded project for rapid evolution of enzymes and synthetic microorganisms developing Raman spectroscopy as a high-throughput analytical technique for industrial bioprocesses.Steve O'Hagan gained BSc and MSc degrees in Chemistry at the University of Manchester.The MSc research was to characterise 'Molecular Beam Sources' using mass spectrometry and TOF kinetic energy measurements; computer simulations were also used.Steve gained a PhD in Chemistry at the University of Warwick; several mass spectrometry and chemometric techniques were employed to analyse engine oils and additives.Aer several years working for commercial laboratories, he joined the University of Manchester as a Computer Officer, working with Doug Kell's and Roy Goodacre's research groups.Interests include: genetic programming; laboratory automation; analytical laboratory data analysis; chemometrics; as well as scientic visualization.Roy Goodacre is a PhD graduate from the University of Bristol (UK) where he studied mass spectrometry of microbial systems.Aer a postdoc, Wellcome Trust fellowship and lectureship in the University of Wales, Aberystwyth, he is now Professor of Biological Chemistry at the University of Manchester (UK).His group's main areas of research (http:// www.biospec.net/)are broadly within analytical biotechnology, metabolomics and systems biology.His expertise involves many forms of Raman spectroscopy (including deep UV resonance Raman and SERS), FT-IR spectroscopy, and mass spectrometry, as well as advanced chemometrics, machine learning and evolutionary computational methods.He is Editor-in-Chief of the journal Metabolomics, on the Editorial Advisory Boards of Analyst and Journal of Analytical and Applied Pyrolysis, a founding director of the Metabolomics Society and a director of the Metabolic Proling Forum.3872 | Analyst, 2013, 138, 3871-3884 This journal is ª The Royal Society of Chemistry 2013

Fig. 1 A
Fig. 1 A timeline of events and discoveries in the history of Raman spectroscopy, Nobel Prizes marked in red (photo of Sir C.V. Raman in the sculpture park of Nehru Science Centre, Mumbai, India, by Prof. Paul O'Brien FRS, School of Chemistry, University of Manchester).

Fig. 2
Fig. 2 Results of bibliometric analysis of the number of publications (A) and citations (B/C/D/E) per year listed on ISI Thomson Web of Scienceâ (http:// wos.mimas.ac.uk/) for the period 1992-2012, using the search terms (A) Raman AND disease, (B) spatially offset Raman AND disease, (C) SERS AND disease, (D) Raman endoscopy, (E) Raman imaging AND disease.

Fig. 3 A
Fig.3A figurative (as opposed to fibre-optic) bundle of some of the key terms and demonstrated capabilities of in vivo Raman imaging mentioned throughout this article.

Fig. 4
Fig. 4 Simplified graphical representation of (A) spontaneous Raman compared to (B) spatially offset Raman scattering, illustrating the spatial offset and deeper subsurface probing in skin tissue.L ¼ laser light, R ¼ Raman light.

Fig. 5
Fig.5SERS spectra of dipicolinic acid, a biomarker for Bacillus spores, measured using citrate-reduced silver colloid and a portable 633 nm Raman spectrometer.The concentration range 0-100 ppb in the sample volumes used is of equivalent magnitude to the infective dose of inhalation anthrax.Highlighted are: (A) the pyridine ring breathing vibration and structure of dipicolinate at 1006 cm À1 , and (B) the C-H stretching vibration and structure of glutaric acid at 2934 cm À1 , which is included as an internal standard to allow accurate quantification.

Table 1
Definition of some of the main approaches of Raman spectroscopy discussed in this review /adsorption onto a roughened metal surface, a colloidal solution or a roughened electrode (usually Ag or Au).Enhancement explained by two processes; an electromagnetic enhancement effect (thought to dominate) and a charge transfer mechanism, known as chemical enhancement.Has a uorescence quenching effect.Can tune to a specic chromophore for additional resonance enhancement (known as SERRS).Enhancements over normal Raman scattering of typically 10 3 to 10 6 SORSRaman spectra are collected from locations spatially separated from the point of laser illumination on the sample surface.SORS allows for the isolation of chemically rich spectral information from distinct substructures or layers and through other barriers, not accessible via spontaneous Raman.Typical wavelengths used for biological tissue are 785 or 830 nm.Ideal for detecting disease in cells and tissue underlying other tissue types, such as bone through skin, cancer cells through muscle and lipid tissue for example 25, 26, 39, 61, 63 and 139 This journal is ª The Royal Society of Chemistry 2013 Analyst, 2013, 138, 3871-3884 | 3875

Table 2
[140][141][142][143][144] frequency assignments of biological interest[140][141][142][143][144] 3876 | Analyst, 2013, 138, 3871-3884 This journal is ª The Royal Society of Chemistry 2013 In the same year work by the Stone group incorporated Raman spectroscopy and clinical endoscopy to discriminate between This journal is ª The Royal Society of Chemistry 2013 Analyst, 2013, 138, 3871-3884 | 3877 Critical Review Analyst normal, dysplastic and cancerous laryngeal tissue. 69Nick Stone's group, based in the UK (http://www.exeter.ac.uk/) have played a major role in pioneering the eld of clinical optical diagnostics with (in addition to other vibrational spectroscopies) well received studies using Raman spectroscopy/ endoscopy to analyse a range of diseases/disorders such as Barrett's oesophagus 70 and bladder and prostate cancer. 71One of the recent studies by the Stone group involved the evaluation of the suitability of a custom-built bre-optic Raman probe for the potential in vivo diagnosis of early onset oesophageal neoplasia.Whilst this involved ex vivo sampling, the results clearly demonstrated the potential for the rapid and accurate differentiation between benign tissue, Barrett's oesophagus and both dysplastic and malignant tissue.
90,883878 | Analyst, 2013,138, 3871-3884This journal is ª The Royal Society of Chemistry 2013In terms of noninvasive biomedical Raman other very exciting developments are spatially offset Raman spectroscopy (SORS)89as well as transmission Raman, both techniques directly resulting from research involving depth proling using Raman Kerr-gating methods.90SORSwas invented and developed in the Central Laser Facility of the Rutherford Appleton Laboratory (http://www.clf.rl.ac.uk) in 2005 by Pavel Matousek and readers are directed to an excellent overview of this fast moving eld by Matousek and Stone.91