Christoph Krafft

Disease recognition by infrared and Raman spectroscopy

Infrared (IR) and Raman spectroscopy are emerging biophotonic tools to recognize various diseases. The current review gives an overview of the experimental techniques, data-classification algorithms and applications to assess soft tissues, hard tissues and body fluids. The methodology section presents the principles to combine vibrational spectroscopy with microscopy, lateral information and fiber-optic probes. A crucial step is the classification of spectral data by a variety of algorithms. We discuss unsupervised algorithms such as cluster analysis or principal component analysis and supervised algorithms such as linear discriminant analysis, soft independent modeling of class analogies, artificial neural networks support vector machines, Bayesian classification, partial least-squares regression and ensemble methods. The selected topics include tumors of epithelial tissue, brain tumors, prion diseases, bone diseases, atherosclerosis, kidney stones and gallstones, skin tumors, diabetes and osteoarthritis.

Raman and CARS microspectroscopy of cells and tissues

Raman spectroscopy has been recognized to be a powerful tool to study cells and tissues because the method provides molecular information without external markers such as stains or radioactive labels. To overcome the disadvantage of low signal intensities from most biomolecules, enhancement effects are utilized. A non-linear variant of Raman spectroscopy called coherent anti-Stokes Raman spectroscopy (CARS) belongs to the most promising techniques because it combines signal enhancement due to the coherent nature of the process with further advantages such as directional emission, narrow spectral bandwidth and no disturbing interference with autofluorescence. This review describes briefly the principles of the methods and summarizes applications to cells and tissues that are expected to gain significance in the future such as the combination with imaging approaches, microscopy, optical traps and fiber-optic probes.

Mapping of single cells by near infrared Raman microspectroscopy

Objective of our work is the development of new diagnostic methods for detection of tissue states (e.g. tumors, necrosis) in vivo. Therefore, it is important to understand the spectral features of pure individual components of tissues, that means cells and subcellular components. Raman spectroscopy has promising potential as an analytical tool for clinical applications because it can probe the chemical composition and molecular structure of such complex systems. Furthermore, the spatial resolution of Raman microspectroscopy in the low micrometer scale and its ability to probe samples under in vivo conditions allow new insights into living single cells without the need for fixatives, markers or stains. In a first set of experiments we prepared human embryonic lung epithelial fibroblasts, human osteogenic sarcoma cells and human astrocytoma cells. Raman mapping data sets were acquired with 1 μm step size and 1 min exposure time per spectrum using 785 nm excitation wavelength. Principal component analysis (PCA) was used for evaluation of the spectral maps. Bands in individual spectra were assigned to proteins, lipids, cholesterol and nucleic acids. Based on this information, the main cellular constituents of freeze dried cells and living cells in media were identified in score plots of principal components.

Nonlinear microscopy, infrared, and Raman microspectroscopy for brain tumor analysis

Contemporary brain tumor research focuses on two challenges: First, tumor typing and grading by analyzing excised tissue is of utmost importance for choosing a therapy. Second, for prognostication the tumor has to be removed as completely as possible. Nowadays, histopathology of excised tissue using haematoxylin-eosine staining is the gold standard for the definitive diagnosis of surgical pathology specimens. However, it is neither applicable in vivo, nor does it allow for precise tumor typing in those cases when only nonrepresentative specimens are procured. Infrared and Raman spectroscopy allow for very precise cancer analysis due to their molecular specificity, while nonlinear microscopy is a suitable tool for rapid imaging of large tissue sections. Here, unstained samples from the brain of a domestic pig have been investigated by a multimodal nonlinear imaging approach combining coherent anti-Stokes Raman scattering, second harmonic generation, and two photon excited fluorescence microscopy. Furthermore, a brain tumor specimen was additionally analyzed by linear Raman and Fourier transform infrared imaging for a detailed assessment of the tissue types that is required for classification and to validate the multimodal imaging approach. Hence label-free vibrational microspectroscopic imaging is a promising tool for fast and precise in vivo diagnostics of brain tumors.

Sample size planning for classification models

In biospectroscopy, suitably annotated and statistically independent samples (e.g. patients, batches, etc.) for classifier training and testing are scarce and costly. Learning curves show the model performance as function of the training sample size and can help to determine the sample size needed to train good classifiers. However, building a good model is actually not enough: the performance must also be proven. We discuss learning curves for typical small sample size situations with 5-25 independent samples per class. Although the classification models achieve acceptable performance, the learning curve can be completely masked by the random testing uncertainty due to the equally limited test sample size. In consequence, we determine test sample sizes necessary to achieve reasonable precision in the validation and find that 75-100 samples will usually be needed to test a good but not perfect classifier. Such a data set will then allow refined sample size planning on the basis of the achieved performance. We also demonstrate how to calculate necessary sample sizes in order to show the superiority of one classifier over another: this often requires hundreds of statistically independent test samples or is even theoretically impossible. We demonstrate our findings with a data set of ca. 2550 Raman spectra of single cells (five classes: erythrocytes, leukocytes and three tumour cell lines BT-20, MCF-7 and OCI-AML3) as well as by an extensive simulation that allows precise determination of the actual performance of the models in question.

Molecular pathology via IR and Raman spectral imaging

During the last 15 years, vibrational spectroscopic methods have been developed that can be viewed as molecular pathology methods that depend on sampling the entire genome, proteome and metabolome of cells and tissues, rather than probing for the presence of selected markers. First, this review introduces the background and fundamentals of the spectroscopies underlying the new methodologies, namely infrared and Raman spectroscopy. Then, results are presented in the context of spectral histopathology of tissues for detection of metastases in lymph nodes, squamous cell carcinoma, adenocarcinomas, brain tumors and brain metastases. Results from spectral cytopathology of cells are discussed for screening of oral and cervical mucosa, and circulating tumor cells. It is concluded that infrared and Raman spectroscopy can complement histopathology and reveal information that is available in classical methods only by costly and time-consuming steps such as immunohistochemistry, polymerase chain reaction or gene arrays. Due to the inherent sensitivity toward changes in the bio-molecular composition of different cell and tissue types, vibrational spectroscopy can even provide information that is in some cases superior to that of any one of the conventional techniques.

Near infrared Raman spectroscopic mapping of native brain tissue and intracranial tumors

This study assessed the diagnostic potential of Raman spectroscopic mapping by evaluating its ability to distinguish between normal brain tissue and the human intracranial tumors gliomas and meningeomas. Seven Raman maps of native specimens were collected ex vivo by a Raman spectrometer with 785 nm excitation coupled to a microscope with a motorized stage. Variations within each Raman map were analyzed by cluster analysis. The dependence of tissue composition on the tissue type in cluster averaged Raman spectra was shown by linear combinations of reference spectra. Normal brain tissue was found to contain higher levels of lipids, intracranial tumors have more hemoglobin and lower lipid to protein ratios, meningeomas contain more collagen with maximum collagen content in normal meninges. One sample was studied without freezing. Whereas tumor regions did not change significantly, spectral changes were observed in the hemoglobin component after snap freezing and thawing to room temperature. The results constitute a basis for subsequent Raman studies to develop classification models for diagnosis of brain tissue.

Identification and differentiation of single cells from peripheral blood by Raman spectroscopic imaging

Medical diagnosis can be improved significantly by fast, highly sensitive and quantitative cell identification from easily accessible body fluids. Prominent examples are disseminated tumor cells circulating in the peripheral blood of cancer patients. These cells are extremely rare and therefore difficult to detect. In this contribution we present the Raman spectroscopic characterization of different cells that can be found in peripheral blood such as leukocytes, leukemic cells and solid tumor cells. Leukocytes were isolated from the peripheral blood from healthy donors. Breast carcinoma derived tumor cells (MCF-7, BT-20) and myeloid leukaemia cells (OCI-AML3) were prepared from cell cultures. Raman images were collected from dried cells on calcium fluoride slides using 785 nm laser excitation. Unsupervised statistical methods (hierarchical cluster analysis and principal component analysis) were used to visualize spectral differences and cluster formation according to the cell type. With the help of supervised statistical methods (support vector machines) a classification model with 99.7% accuracy rates for the differentiation of the cells was built. The model was successfully applied to identify single cells from an independent mixture of cells based on their vibrational spectra. The classification was confirmed by fluorescence staining of the cells after the Raman measurement.

Raman and coherent anti-Stokes Raman scattering microspectroscopy for biomedical applications

A tutorial article is presented for the use of linear and nonlinear Raman microspectroscopies in biomedical diagnostics. Coherent anti-Stokes Raman scattering (CARS) is the most frequently applied nonlinear variant of Raman spectroscopy. The basic concepts of Raman and CARS are introduced first, and subsequent biomedical applications of Raman and CARS are described. Raman microspectroscopy is applied to both in-vivo and in-vitro tissue diagnostics, and the characterization and identification of individual mammalian cells. These applications benefit from the fact that Raman spectra provide specific information on the chemical composition and molecular structure in a label-free and nondestructive manner. Combining the chemical specificity of Raman spectroscopy with the spatial resolution of an optical microscope allows recording hyperspectral images with molecular contrast. We also elaborate on interfacing Raman spectroscopic tools with other technologies such as optical tweezing, microfluidics and fiber optic probes. Thereby, we aim at presenting a guide into one exciting branch of modern biophotonics research.

A comparative Raman and CARS imaging study of colon tissue

An experimental evaluation of the information content of two complimentary techniques, linear Raman and coherent anti-Stokes Raman scattering (CARS) microscopy, is presented. CARS is a nonlinear variant of Raman spectroscopy that enables rapid acquisition of images within seconds in combination with laser scanning microscopes. CARS images were recorded from thin colon tissue sections at 2850, 1660, 1450 and 1000 cm(-1) and compared with Raman images. Raman images were obtained from univariate and multivariate (k-means clustering) methods, whereas all CARS images represent univariate results. Variances within tissue sections could be visualized in chemical maps of CARS and Raman images. However, identification of tissue types and characterization of variances between different tissue sections were only possible by analysis of cluster mean spectra, obtained from k-means cluster analysis. This first comparison establishes the foundation for further development of the CARS technology to assess tissue.