29.03.2024 08:29 - About Us - Mediadaten - Imprint & Contact - succidia AG
L&Mint-int-2-2014 > The potential and limits of Raman spectroscopy methods in medical diagnostics

The potential and limits of Raman spectroscopy methods in medical diagnostics

Quick, comfortable diagnosis

As the proportion of elderly people in the population continues to rise, we are facing the growing challenge of providing affordable and ­sustainable healthcare systems. An impending collapse of these ­systems can be avoided only by developing new methods and ­equipment that enable disease to be detected and tackled as early ­as possible. Ideally, this would be no later than the occurrence of initial disease-driven changes at a molecular level.

On account of the special properties of light, photonic solutions offer highly promising results – and none more so than ­Raman spectroscopy and its various techniques. As one example, Raman spectroscopy differs from several other photonic solutions as it enables contactless measurement that requires no exogenous labels. This is particularly beneficial since such labels face the same regulatory hurdles as medicines. In addition, the fact that Raman spectroscopy is comparatively fast and precise means it offers a diagnostic approach that is especially comfortable and direct for the patient. Its qualities as an imaging method mean that it offers high specificity while simultaneously remaining low- or non-invasive when combined with other optical methods and chemometrical methods in particular. Other advantages of Raman spectroscopy include its high spatial resolution, low effort for sample preparation and the weak Raman spectrum given by water, a property offering the possibility of working with aqueous samples.

This article will present some examples of possible applications, selected for their relevance to medicine and clinical practice.

Cell diagnostics

In pathogen diagnostics, the gold standard is the incubation and analysis of a bacterial culture. This requires both time – as much as a week, in some cases – and experienced lab personnel. In the example of a patient with sepsis, however, this quantity of time is simply unavailable. Not least because the rate of survival after the onset of septic shock drops to less than 1 in 5 after merely twelve hours without specific treatment [1]. Raman spectroscopy has the potentially to help identify the pathogen responsible within just a few hours. Every species of bacteria possesses its own personal Raman signature, and the spectrum from just a single bacterium can be sufficient for identification [2]. Since spectral differences between species are often subtle, the application of chemometrical methods is imperative to ensure their proper assignment.


Fig. 1 I) Raman spectra of the biological constituents of bacteria: water, proteins, nucleic acids (DNA), carbohydrates and fats. In the examples of Raman spectra given, separate Staphylococcus strains can be identified from the various bands produced by individual constituents. II) Bio Particle Explorer (a rapID product).

A bacterial spectrum is made up of spectral signatures from its constituent substances, such as water, proteins, fats, ­nucleic acids, carbohydrates, etc. (fig.1-I). Accordingly, subtle differences can even be present in spectra from members of the same bacterial species, as a result of differences in age, nutritional condition and environmental factors. By utilising databases that store spectra from bacteria in different physical conditions, chemometrical methods are then able to assign the Raman spectra to the corresponding species. ­Effectively, the process involves splitting a spectrum into especially data-rich areas, which are then compared with the corresponding areas from spectra held in the database. The process enables almost 99% of bacteria to be assigned to the correct species [2]. For identifying bacteria in less complex matrices, e.g. in cleanroom air, a corresponding solution is already commercially available (RapID Bio Particle Explorer, fig.1-II). This approach uses fluorescence spectroscopy to distinguish between non-living particles and bacteria. The latter are then targeted for identification by Raman spectroscopy. In more complex media, however, such as saliva, urine or (especially) blood, bacteria must first be separated from these media, since the medium's own Raman signature otherwise makes identification difficult or impossible. This separation stage can be performed by microfluidic chips, which may apply the principle of dielectrophoresis, for example, so as to trap and hold bacteria, and thus make them available for measurement [3]. This obviates the need for actual physical separation of the medium: instead, the ­bacteria can also be measured directly in solution. Apart from merely identifying the bacteria, the latter step also provides data on bacterial susceptibility or resistance to antibiotics. This involves using Raman spectroscopy to measure bacterial growth curves under the influence of antibiotics. Within two hours, the presence of resistance can be determined with a sensitivity and specificity each of 90%.

Alongside bacterial detection, diagnosis of tumour cells in blood also has a major role to play. Such cells can be shed by cancerous tissue, enter the bloodstream and then cause metastases. In the bloodstream, isolated tumour cells are fairly easily accessible. A procedure using Raman spectroscopy can be used to detect them in a way similar to flow cytometry. The blood specimen first passes through a microfluidic chip. Within this chip, single cells are held by optical traps, and are then analysed and classified with the aid of Raman spectroscopy. Based on this classification, the cells are then sorted for further processing. Due to vibrational spectroscopic characterisation, Raman spectroscopy permits considerably more precise diagnostic results to be achieved at the level of the single cell, compared to flow cytometry. The process does require more time, however, resulting in a considerably reduced flow (5–6 cells/minute). Yet technical improvements to equipment will no doubt lead to considerably higher rates of flow in the future. Figure 2 shows a microfluidic chip of this kind made from quartz [4]. The Raman spectra recorded still exhibit artefacts from the spectral properties of the optical filter, the trapping laser and the substrate material. Accordingly, these still need to be eliminated in order to ensure successful classification, and render visible the spectral fingerprints of white blood cells (green) and tumour cells (orange, brown, blue).


Fig. 2 Microfluidic chip for Raman-activated cell sorting

Tissue diagnostics

For some tumours, radical removal may not be desirable or even possible. A brain ­tumour is just one example. The aim here is to remove as little healthy tissue as possible: a difficult task, since tumour and healthy tissue are often not clearly differentiated. In general terms, an objective method would be desirable, which could then be used directly in the operating theatre. The ideal instrument would be an operating microscope, able to image tumour bound­aries directly. To implement such a microscope, it would be advantageous to combine multiple imaging techniques that increase contrast while largely avoiding ­labels. Our experience has shown us that especially promising techniques here ­include combining Raman with coherent anti-Stokes Raman scattering (CARS), two-photon excited fluorescence (TPEF) while using endogenous markers and second harmonic generation (SHG) [5]. With Raman spectroscopy, all (Raman-active) vibrational modes are excited simultaneously, whereas in CARS, the overlaying of three separate, spatially oriented light pulses causes a ­chosen vibration to be isolated and coherently excited, resulting in the generation of a fourth, spatially aligned and coherent light pulse. In contrast to Raman imaging, the strongly enhanced scattering cross-section enables the recording of an image in a much shorter time frame (factor of 104). While Raman and CARS provide us with chemical data, SHG and TPEF augment these to include morphological details. SHG emphasises highly ordered, non-centrosymmetric structures such as collagen, while TPEF is particularly amenable to endogenously fluorescent substances such as NAD(P)H, flavins, elastins, etc. Figure 3 presents a comparison of TPEF, CARS and Raman microscope images of an undyed brain tumour thin-section with a light ­microscope image of the prepared specimen subsequently dyed with haemotoxylin and eosin. The cell nuclei, which are resolved by all methods, are particularly important for the histopathological evaluation. By combining morphological and functional information, this approach has the potential not only to detect and classify tumours at an early stage, but also to localise tumour boundaries with an adequate level of precision and reliability.


Fig. 3 Comparison of TPEF, CARS and Raman microscope images of an undyed brain tumour thin-section with a light microscope image of the prepared specimen subsequently dyed with haemotoxylin and eosin (from left to right: Raman, light microscope, TPEF, CARS).


Fig. 4 In vivo investigation of rabbit arteries ­using Raman endoscopy.

Organ diagnostics

For the endoscopic investigation of arterial plaque, a morphological evaluation alone is not sufficient, since this is unable to assess whether deposits are harmless or whether these may detach themselves from the vessel wall, producing obstructions that are capable of causing heart failure or a stroke. Since the Raman spectra of calcium phosphate, connective tissue, triglycerides and cholesterol are characteristic, endoscopic Raman spectroscopy could be used to determine plaque composition and thus the hazard that these plaques represent. The value of this diagnostic approach has already been confirmed in animal experiments. Figure 4 shows how a probe 1 mm in diameter was combined with a central excitation fibre and 12 detection fibres for use in ex vivo measurements in rabbits [6]. Measurement conditions were chosen to accurately simulate in vivo conditions. In terms of their intensity and spectral position, the signals from plaque deposits are clearly distinguishable from arterial wall lipids with bands of collagen and from blood with bands of red blood cells. The combination of chemical and morphological data should be equally advantageous in this scenario. We are therefore planning to combine Raman spectroscopy with optical coherence tomography and/or ultrasound. Miniaturisation also offers the possibility of making even finely-structured arteries accessible to measurement.

Bibliography

[1] Kumar, A. et al. (2006) Crit. Care Med., 34, 1589?–1596
[2] Harz, M. et al. (2009) Cytometry A, 75A(2), 104?–113
[3] Schröder, U.-Ch. et al. (2013) Infection, 41(Suppl.1), P036
[4] Dochow, S. et al. (2013) Anal. Bioanal. Chem., 405, 2743??–6
[5] Meyer, T. et al. (2011) J. Biomed. Optics, 16, 021113
[6] Matthäus, C. et al. (2012) Anal. Chem., 84, 7845?–51

Picture: © 123rf.com | mikekiev

L&M int. 2 / 2014

The articles are publishes in issue L&M int. 2 / 2014.
Free download here: download here

The Author: