Vibrational spectroscopy, known as biospectroscopy in biological systems, is a powerful analytical tool widely used to address biological questions in a novel approach (Martin et al 2010; Baker et al 2014; Butler et al 2015). This allows for the non-invasive and non-destructive analysis of biological samples without the need for prior fixation or labelling of samples. FTIR and/or Raman microspectroscopy has been shown to be a powerful analytical technique in the study of biological materials, and it allows rapid, non-invasive and high spatial resolution acquisition of biochemical and structural information through the generation of point spectra or spectral images. It generates fingerprint spectra of a biological sample through a sensor, based on the chemical bonds in the sample resulting in bond vibrations. These spectral fingerprints are characteristic of the chemical composition, structure and function of the sample. These ‘biochemical fingerprints’ are complex, reflecting the heterogeneous nature of biological material, and quantitative. A computational framework is used to reduce the level of complexity allowing subtle differences in composition between samples, to be clearly visualized allowing spectral biomarkers to be identified within samples (Martin et al 2010). Data import, pre-treatment, and construction of chemometric bio-imaging generation (PCA, HCA, MCR-ALS) is implemented in MATLAB R2010a software (Mathworks Inc, Natick, MA, USA) (Patel et al 2011).
These approaches have been applied in a wide range of studies using biofluids or tissues for oncology, neurodegenerative disease, environmental bio-monitoring and toxicology studies. Biospectroscopy approaches are capable of deriving biochemical information from tissue structures. Not requiring specific contrast agents such as immunohistochemical tags for feature identification, bio-molecular contrast images or fingerprint spectra can be derived following spectral mapping of a tissue at a diffraction-limited spatial resolution. Vibrational spectroscopy approaches have the capability to cheaply, rapidly and non-destructively analyze biological tissue, having been previously used to track stem cell lineages or to investigate cancer cell biology. As it is a scattering technique, Raman micro-spectroscopy has the inherent advantage of being unaffected by aqueous, permitting in vivo and live-cell imaging. In Raman spectroscopy chemical bonds are excited to a virtual state through excitation with photons; these bonds then relax into a different vibrational state causing inelastic photon scattering and frequency shift. These scattered photons cause a shift in frequency indicative of specific vibrational modes; from this light scattering effect, a biochemical signature of the analyzed cell can be derived. This methodology has been exploited as a potential diagnostic tool for many diseases, including breast and gastric cancer, and malaria.
Each FTIR or Raman spectrum can be considered as a highly complex biochemical dataset, with typical acquisitions of several thousand spectra per experiment. These require intricate algorithms for classification of spectroscopy data or image re- construction from vibrational spectra. Clustering analyses partition a dataset into a pre-defined number of segments (clusters), where the spectra in a single cluster are similar and those of other clusters progressively differ with distance. Commonly-used clustering analyses are hierarchical cluster analysis (HCA), KMCA and fuzzy C-means cluster analysis. Component analysis such as principal component analysis (PCA) decomposes a data matrix into a number of latent variables, called principal components (PCs). The data is transformed into a new coordinate system, with the greatest variance displayed in the 1st PC, and so on. Each spectrum is substituted for values of scores on each PC in the new coordinate system, allowing a single spectrum to be visualized on multiple PC axes in what can be known as a ‘scores’ plot. A contrast image can be re-constructed according to the variance captured in these variables.
- Martin, F.L. et al. (2010) Distinguishing cell types or populations based on the computational analyses of their infrared spectra. Nature Protocols 5: 1748-1760.
- Baker, M.J. et al. (2014) Application of Fourier transform infrared spectroscopy for analysis of biological materials towards understanding cell functionality, tissue imaging and disease diagnosis. Nature Protocols 9: 1771-1791.
- Butler, H.J. et el. (2016) Using Raman spectroscopy to characterize biological materials. Nature Protocols 11: 664-687.
- Patel, I.I. et al. (2011) High contrast images of uterine tissue derived using Raman microspectroscopy with the empty modelling approach of multivariate curve resolution-alternating least squares. Analyst 136: 4950-4959.