Infrared spectroscopy is nowadays frequently employed for applications in clinical chemistry. Since blood serves as the primary metabolic transport system in the body, its composition is the preferred indicator with respect to the pathophysiological condition of the patient. An important class of substances are the metabolites, including glucose, which are accessible by direct spectroscopic measurement without sample treatment. Multicomponent assays based on such technology are reagentless, fast and readily automated. Different in-vitro assays using mid- or near-infrared spectral data are presented including results from ex-vivo measurements using microdialysis and ATR spectroscopy for continuous blood glucose monitoring. Non-invasive sensing systems are under development for the determination of blood glucose, especially for diabetic patients or for monitoring in intensive care and surgery. Near-infrared spectrometry of skin tissue has been proposed, which allows a certain tissue volume to be integrally probed. On the other hand, fast measurements, such as used in pulse oximetry, can enable intravascular probing, i.e. collecting information on the arterial part of the vascular system (near-infrared plethysmography). Results and prospects for applications in non-invasive blood component assays are discussed.

IR spectra of breast tumor cell lines and breast tumor tissues have been measured. IR measurements of tumor cells revealed that approximately 15 cells are necessary to obtain spectra of good signal-to-noise ratio using an IR microspectrometer equipped with a conventional IR thermal source. Comparative studies of human breast tumor cell line suspensions demonstrated that MCF-7 cells and drug-resistant NCI/ADR cells can be differentiated based on their IR spectra. The most striking differences between MCF-7 and NCI/ADR were found in features assigned to CH₂ and CH₃ stretching vibrations of lipid acyl chains and PO₂ stretching vibrations of nucleic acids. To assess the potential of IR spectroscopy for the diagnosis of breast tumor tissues, thin sections of tissue were mapped by FTIR microspectroscopy. The spectra of these maps were analyzed using functional group mapping techniques and cluster analysis, and the output values of the different approaches were then reassembled into IR images of the tissue. A comparison of the IR images with the standard light microscopic images of the corresponding areas suggested that: (i) chemical mapping based on single band intensities is an easy way to detect microscopic fat droplets within tissue; (ii) the comparison of IR images based on band intensities at 1054 and 1339 cm^-1 provides information on tissue areas containing tumor cells; (iii) cluster analysis of the spectra is superior to the single band approach and more appropriate for differentiation between tissue types.

IR spectroscopy is proving to be a powerful tool for the study and diagnosis of cancer. The application of IR spectroscopy to the analysis of cultured tumor cells and grading of breast cancer sections is outlined. Potential sources of error in spectral interpretation due to variations in sample histology and artifacts associated with sample storage and preparation are discussed. The application of statistical techniques to assess differences between spectra and to non-subjectively classify spectra is demonstrated.

ATR-FTIR spectroscopy and Raman spectroscopy were employed to obtain information about the molecular composition and hydration of skin in vivo. Both techniques enable the in vivo acquisition of high quality spectra within 10-30s at a spectral resolution of 8cm^-1. The penetration depth of ATR-FTIR is about 1.5 (Mu) m. Raman spectra could be obtained with a resolution of about 5 micrometers . ATR-FTIR spectra of hydrated stratum corneum were analyzed using a band fitting algorithm. By means of this algorithm the signal contributions of water relative to protein signal contributions could be determined. The results of Raman microspectroscopic experiments on frozen sections and isolated skin components were used for the interpretation of Raman spectra obtained in vivo. Information was obtained about lipid components present in the stratum corneum. These were shown to vary widely between individuals and between different locations on the body. The combination of these spectroscopic techniques may prove to be valuable for applications in dermatology and skin care.

In this work, we have studied the cancer cell lines sharing a common feature: the multi-drug resistance where P- glycoprotein is responsible for the active efflux of the drug out of the cell. For this, we have used two types of cells, MDR-human leukemic K562 cells and non-MDR acute promyelocytic leukemic HL60 cells. The comparison between normalized micro FT-Raman spectra of resistant and sensitive K652 cells shows a decrease in the intensity of the amide I and III bands and a down shift of the amide I band. On the other hand, control experiments with HL60 cells do not show any remarkable changes. Analysis of micro-FT-Raman spectra by resolution enhancement methods and by chemometrics tools reveal further information concerning the conformational changes of the cell constituents involved in the expression of the MDR-phenotype.

The fluorescence of natural constituents of bio-material may conceal its Raman spectra. This fluorescence is reduced by shifting the existing radiation to longer wavelengths. For several reasons the optimum is the excitation with 1064 nm radiation, produced by the Nd:YAG laser. In order to explore the applicability to medical diagnostics we installed an NIR-FT-Raman spectrometer in the 'Universitaetsklinikum Essen'. The results complied within 5 dissertations show that there are some useful potential applications in this field. However, much more work and, especially, international cooperation, is needed to develop this tool further for routine application.

A noninvasive method to measure levels of important tissue culture nutrients and metabolic by-products would provide the potential for computer control of the growth process, decreasing the amount of human attention necessary while improving efficiency and repeatability of experiments by precise maintenance of important parameters without contaminating the culture environment. Our approach to meeting this need is the development of optical sensor that extract chemical information based on the absorbance of near-IR radiation. Near-IR spectroscopy of the 2.0-2.5 micrometers region was used to collect absorbance data of aqueous mixtures containing glucose, lactate, ammonia, glutamate, and glutamine, as well as cell culture media samples from a three-day growth period. Multivariate calibration was used with these data to prove that complicated spectra can yield reasonable prediction errors. Calibration models using different combinations of spectra from aqueous mixtures and culture media were built and evaluated for glucose, ammonia, and lactate. Prediction errors of 2 percent, 8 percent, and 15 percent were obtained for glucose, lactate, and ammonia, respectively.

A method of multicomponent analysis for blood substrates based on FTIR spectroscopy is described. Venous and capillary blood was used as a matrix, with sample volumes as small as 1 (mu) l. The spectra were obtained by transmission and diffuse reflection measurements from dried samples on a disposable carrier. The mean square prediction errors for 90 blood samples form different patients, calculated by cross- validation, were found to be 18 mg/dl for glucose, 17.5 mg/dl for cholesterol, 23.5 mg/dl for triglycerides, and 0.77 g/dl for haemoglobin. The mean square prediction errors for serum analyses are smaller: for a population of more than 200 serum samples we found 8 mg/dl for cholesterol, 14.9 mg/dl for triglycerides, 0.23 g/dl for total proteins, and 8.7 mg/dl for urea. The in-series standard error for the glucose concentration was low as 2.2 mg/dl. The combined spectroscopic and biological error for glucose was found to be 7 mg/dl. This value includes the 'matrix effect'. In independent serum samples glucose concentrations were measured with a total error of 16 mg/dl.

In IR immunoassay, organometallic markers containing metal- carbonyl moieties are employed in conjunction with FTIR spectroscopy as a detection method. These metal-carbonyl markers have characteristic and intense absorption bands in a region of the IR spectrum that is devoid of other strong absorptions and can be detected at pixomole levels. In addition, the ability to individually detect different metal-carbonyl markers in the presence of each other affords the possibility of determining multiple antigens in a single test. In the present work, we have undertaken the simplification and automation of the IR immunoassay method in order to increase its potential utility for routine use. A solid-phase IR immunoassay test has been developed in which antibodies specific to the analytes of interest are adsorbed on the surface of an IR-transparent membrane. In an alternative homogenous immunoassay format, antigens are adsorbed on the surface of an internal reflection element (IRE), and the binding of these antigens to the corresponding antibodies in a solution placed in contact with the IRE is monitored by attenuated total reflectance spectroscopy. Competition between the adsorbed antigens and free antigens in the solution for the antibody binding sites then provides the basis for a homogenous immunoassay.

pH electrodes have been used during open heart surgery to ensure adequate delivery of blood and oxygen to the myocardium during the surgical procedure. The electrodes are cumbersome and suffer from motion artifacts. Near infrared spectroscopy was evaluated as a noninvasive method of measuring myocardial pH during regional ischemia in seven beating dog hearts. Two pH microelectrodes were implanted in the distribution area of the left anterior descending (LAD) coronary artery. The LAD was occluded to stop the myocardial blood flow and to initialize regional ischemia. Ischemia was maintained for 20 minutes before the LAD was released to resume blood flow. A fiber-optic probe was used to collect the reflected NIR light over the spectral region of 575 nm to 1100 nm from the heart muscle. Partial least-squares multivariate calibration technique was applied to relate the myocardial pH changes to the NIR spectral changes in the region of 700 to 1100 nm. Calibration models based on data collected on each individual dog heart had an average of 7 factors with an R² of 0.84. The standard error of prediction (SEP) averaged 0.09 pH units for a mean pH change of 0.73 pH units, adequate for monitoring pH changes during cardiac surgery.

Light in the near-IR (NIR) spectral region can penetrate relatively deep into soft tissue. In this region, the light absorption property is determined by tissue constituents, especially water, fat, and collagen, and their combination ratio. If the light absorption spectra of tissue constituents were known, the combination ratio could be determined by quantifying the light path length in different tissue constituents. In order to obtain the accurate absorption property, the absorption spectra were measured by a Shimadzu 3101-PC spectrophotometer. Since animal fat contains many kinds of fatty acid, five kinds of major fatty acid found in human fat were mixed with proper ratio as a standard sample. The results show that temperature has a stronger effect on the absorption property of water than on that of fatty acid mixture. The absorption spectrum of hog eye lens was measured to obtain the absorption property of collagen. Its absorption spectrum is quite similar to that of dry bovine gelatin. NIR spectroscopy might be useful to characterize or identify different types of soft tissue based on their major chemical composition, such as detecting a fat plaque in a muscular tissue or a tumor in a high fat content tissue.

The importance and effects of data preprocessing and wavelength selection were investigated in predicting total hemoglobin concentrations form absorption spectra. Spectra of the 1 nm interval between 500-900nm were measured from the whole blood samples taken form 165 patients whose hemoglobin concentrations ranged between 7-17 g/dl. The concentrations were predicted using the partial least squares regression. A total of 18 different combinations of preprocessing were tested. The partial least squares regression analysis provided quite different results depending on preprocessing methods and a wide range of prediction accuracy was obtained. For example, the sum of squares of difference ranged from 6-18.6, R² varied from 0.8333 to 0.9477 and the root mean squared errors were from 0.5504-0.966 g/dl. The best results was obtained from the data processed by linear regression baseline fitting, unit area correction, mean centering and variance scaling. Instead of using all wavelengths in the broad-band spectra, a discrete number of wavelengths were selected to predict the concentrations using our algorithm, which will be advantageous in developing compact and less expensive commercial devices. It proves that a careful selection of wavelengths can provide a comparable accuracy obtained from using the broad-band spectra. For our particular experimental data, the measurement form only three discrete wavelengths could provide excellent results.

NIR spectra of biological tissue consist of a number of broad, overlapping absorbance bands on a sloping vaseline. The interpretation and processing of such spectra are complicated by multiple-scattering interactions that distort the shapes of the absorbance bands and introduce wavelength- dependent scattering losses. In this paper we explain the dependence of the shape of the diffuse-reflection log(1/R) spectrum of a turbid medium on the scattering coefficient and probe geometry. From measurements on tissue phantoms and biological tissue, we observe that the separation distance between source and detector probes affects the sensitivity of the reflectance to changes in the density of scattering centers and alters the wavelength dependence of the baseline slope of the log(1/R) spectra. A new method, called fractional derivative processing (FDP), is introduced for extracting information from broad absorption bands corrupted by residual baseline variations and high-frequency noise. FDP was evaluated on spectra obtained from living tissue and tissue phantoms. Possible applications include NIR spectroscopy of hemoglobin, water, and other absorbers in human skin.

The inhomogeneity of tissue structure greatly affects the sensitivity of tissue oxygenation measurement by reflectance NIRS. We have proposed a method for correcting the influence of a subcutaneous fat layer on muscle oxygenation measurement. In this study, this method was validated by measuring the peak-to-peak variation of muscle oxygenation in periodic exercise tests on the vastus lateralis and the falling rate of oxygenation in ischemia tests on the forearm. A newly developed multisensor probe with source- detector distances of 7-40 mm was used. THe probe, consisting of a two-wavelength LED and four photodiodes, was connected to a 4-channel tissue oxygen monitor. The fat layer thickness was also measured by ultrasonography. Results of the tests clearly showed that the presence of a fat layer greatly decreases the sensitivity of measurement and increases the light intensity at a detector. The correction factors of sensitivity were determined from this relationship and Monte Carlo simulation. The corrected oxygenation levels were quantitatively compared among subjects in spite of different fat layer thicknesses.

Conventional Fourier transform IR spectroscopy has proven to be an invaluable research and diagnostic approach for the study of a wide range of biomedical problems. In this article we describe a new biomedical imaging method which integrates high-resolution IR spectroscopy with high- definition digital imaging. The continuing development and commercialization of long-wavelength IR cameras or focal- plane arrays has been a key enabling technology. These imaging systems are capable of rapidly generating chemically specific images from a variety of unstained biological tissue and cells. Image contrast is intrinsic to the sample and is determined only by its biochemical composition. In addition, data from a single experiment can be digitally manipulated to produce numerous images of the same sample, for which different spatial and biochemical properties are emphasized. We present data demonstrating the potential of the technique to generate spectroscopic signatures and images from single human breast cells.

FTIR microspectroscopic maps of unstained colon carcinoma thin sections were obtained on a conventional IR microscope equipped with an automatic x, y stage, or alternatively by using a MCT focal plane array detector system. IR data were analyzed by different image re-assembling techniques. One main goal of the present study was to test the influence of different spectra data compression approaches on the quality of the FTIR images. The images, re-assembled by Principal component analysis (PCA) on the basis of spectral information available from the fingerprint region exhibited an excellent image contrast confirming standard histo- pathological examinations. The second approach included a systematic search for spectral windows which were supposed to contain the relevant information, necessary for spectra classification and identification. Data from these spectral windows were analyzed by an ANN and output data were utilized for image construction. In contrast to the PCA approach, the image contrast was lower although the main morphological structures were exactly classified. From the spectroscopic point of view, the spectral feature selection method delivered useful information which could be discussed in terms of structural alternations upon carcinogenesis.

Near IR diffuse reflectance spectroscopy and imaging are used to assess tissue status following reconstructive surgery. Utilizing the differential absorption of oxy- and deoxy-hemoglobin between 670-1100 nm, tissue hemoglobin oxygen saturation changes were monitored in reverse McFarlane dorsal rat skin flaps. Significant changes in these parameters were observed upon surgical elevation of the skin flap. A significant regional variation along the skin flap was also observed. The magnitude of the drop is tissue oxygen saturation, as observed immediately following surgery, correlated with the final clinical outcome of the flap tissue. These results indicate the potential of near IR spectroscopy and imaging to monitor tissue oxygenation status and assess tissue viability following reconstructive surgery.

Healthy and abnormal human skin has been examined using Fourier transform (FT) Raman spectroscopy. The molecular basis of alternations in this tissue have been proved with the aim of providing a tool to aid in clinical diagnosis of skin disorders. Intact human stratum corneum show spectral features of keratin and the lipids. Spectra from callus and psoriatic tissue show that the keratin component is essentially intact and is similar to that in the normal tissue. However, the abnormal skin shows that the keratin component is essentially intact and is similar to that in the normal tissue. However, the abnormal skin shows a significant depletion of the lipoidal component, which correlates with clinical observations of an increase in permeability and the hyper-proliferative nature of these conditions. Verrucal tissue again shows some alterations to the lipoidal fraction of the tissue.

Vibrational spectroscopies hold great promise for applications in medical diagnosis, especially if they can be applied in vivo. Recent advances in flexible fiber probe design, enable good quality Raman spectra of tissue to be obtained in vivo. Here we illustrate this with Raman spectra of rat tissues, obtained ex vivo and in vivo.

FTIR spectra were obtained from 15 methicillin-resistant and 22 methicillin-susceptible Staphylococcus aureus strains using our DRASTIC approach. Cluster analysis showed that the major source of variation between the IR spectra was not due to their resistance or susceptibility to methicillin; indeed early studies suing pyrolysis mass spectrometry had shown that this unsupervised analysis gave information on the phage group of the bacteria. By contrast, artificial neural networks, based on a supervised learning, could be trained to recognize those aspects of the IR spectra which differentiated methicillin-resistant from methicillin- susceptible strains. These results give the first demonstration that the combination of FTIR with neural networks can provide a very rapid and accurate antibiotic susceptibility testing technique.

ATR-FTIR spectroscopic data of well characterized bacterial and yeast strains have been analyzed using sophisticated methods of numerical data analysis. These new emerging methods have helped in differentiating between clinical strains exhibiting two different acquired resistance mechanisms and presenting very subtle differences on a biological basis. The feasibility of this methodology has also been verified on a set of industrial yeasts.

FTIR and FT-NIR Raman spectra of intact microbial cells are highly specific, fingerprint-like signatures which can be used to (i) discriminate between diverse microbial species and strains, (ii) detect in situ intracellular components or structures such as inclusion bodies, storage materials or endospores, (iii) detect and quantify metabolically released CO₂ in response to various different substrate, and (iv) characterize growth-dependent phenomena and cell-drug interactions. The characteristic information is extracted from the spectral contours by applying resolution enhancement techniques, difference spectroscopy, and pattern recognition methods such as factor-, cluster-, linear discriminant analysis, and artificial neural networks. Particularly interesting applications arise by means of a light microscope coupled to the spectrometer. FTIR spectra of micro-colonies containing less than 10³ cells can be obtained from colony replica by a stamping technique that transfers micro-colonies growing on culture plates to a special IR-sample holder. Using a computer controlled x, y- stage together with mapping and video techniques, the fundamental tasks of microbiological analysis, namely detection, enumeration, and differentiation of micro- organisms can be integrated in one single apparatus. FTIR and NIR-FT-Raman spectroscopy can also be used in tandem to characterize medically important microorganisms. Currently novel methodologies are tested to take advantage of the complementary information of IR and Raman spectra. Representative examples on medically important microorganisms will be given that highlight the new possibilities of vibrational spectroscopies.

Disease pattern recognition (DPR) is being developed as a reagent-free measurement technique for the diagnosis of blood samples. The technological basis of this method is mid IR spectroscopy. For the analysis, 1 (mu) l blood is deposited on a disposable and dried before measurement. The IR spectra give rise to characteristic patterns in narrow wavenumber regions, which are modified in the presence of relatively small pathophysiological changes. the spectral changes depending on the disease are always higher than those caused by the deviations resulting from instrumental or handling errors. The information content of the spectra is reflected by the standard error, which varies between differently 'diseased' and 'healthy' persons. The standard error of the first derivative spectra is three times higher for 'diseased' compared to 'healthy' persons. This has been demonstrated for a comparable population of 'diseased' and 'healthy' persons. In total, more than 2000 spectra from different individuals were analyzed. Preliminary result using various mathematical algorithms indicate, that clear distinctions can be found for a variety of different disease compared to patterns of 'healthy' spectra. This qualitative information about the blood sample may be used as a quick and comprehensive diagnostic tool.

IR spectroscopy shows the promise of developing into a clinically relevant methodology for both screening and quantitative applications. The principles are straightforward - the IR spectrum of appropriate biological fluids or tissues may be interpreted to reveal diagnostic information either directly through inspection, or indirectly through the use of appropriate classification or quantitation algorithms. Implementing these methods in practice is not as straightforward. This report summarizes our experiences to date in developing methods for the quantitation of several serum analytes from the IR spectra of dried serum films, and for the IR-based detection/grading of abnormalities in exfoliated cervical cell specimens.

The new method of fiber-optical evanescent wave Fourier transform IR (FEW-FTIR) spectroscopy has been applied to the diagnostics of normal tissue, as well as precancerous and cancerous conditions. The FEW-FTIR technique is nondestructive and sensitive to changes of vibrational spectra in the IR region, without heating and damaging human and animal skin tissue. Therefore this method and technique is an ideal diagnostic tool for tumor and cancer characterization at an early stage of development on a molecular level. The application of fiber optic technology in the middle IR region is relatively inexpensive and can be adapted easily to any commercially available tabletop FTIR spectrometers. This method of diagnostics is fast, remote, and can be applied to many fields Noninvasive medical diagnostics of skin cancer and other skin diseases in vivo, ex vivo, and in vitro allow for the development convenient, remote clinical applications in dermatology and related fields. The spectral variations from normal to pathological skin tissue and environmental influence on skin have been measured and assigned in the regions of 850-4000 cm^-1. The lipid structure changes are discussed. We are able to develop the spectral histopathology as a fast and informative tool of analysis.

The Raman scattering spectra of two different samples of Leptospiral Glycoipoprotein (GLP-1 and GLP-2) which have different toxic effects have been obtained and investigated. Leptospirosis is one of the most harmful zoonosis. It is a serious public health issue in some area of Sichusan province. The two samples offer different structural informations of GLP molecules, it would be important to find the difference in contents, structures and the amino acid side - chains environment of the molecules between the two samples of GLP for understanding the different toxic effects. The intense Am I at 1651 cm^-1 and weak Am III at 1283 cm^-1 show that GLP-1 has a predominantly (alpha) -helix secondary structure. The intense Am I at 1674 cm^-1 and intense Am III at 1246 cm^-1 show that the conformation of GLP-2 has a high content of (Beta) - sheet and a low content of random - coil secondary structure. Strong Raman scattering occurs in the range 920- 980 cm^-1, belong to the C-COO vibration and the stretching of the peptide backbone. The molecules of GLP-1 has trans-gauche-trans configuration of the C-S-S-C-C linkage and the molecules of GLP-2 has trans-gauche-gauche configuration of the C-C-S-S-C-C linkage. The intensity ratio of the two tyrosine liens at 830 cm^-1 and 850 cm^-1 is 1.1 and 1.23, indicate their tyrosine reduces environment respectively. Other side-chain environment in the two samples were discussed.

For characterization of skin cancer, an artificial neural network method has been developed to diagnose normal tissue, benign tumor and melanoma. The pattern recognition is based on a three-layer neural network fuzzy learning system. In this study, the input neuron data set is the Fourier transform IR spectrum obtained by a new fiberoptic evanescent wave Fourier transform IR spectroscopy method in the range of 1480 to 1850 cm^-1. Ten input features are extracted from the absorbency values in this region. A single hidden layer of neural nodes with sigmoids activation functions clusters the feature space into small subclasses and the output nodes are separated in different nonconvex classes to permit nonlinear discrimination of disease states. The output is classified as three classes: normal tissue, benign tumor and melanoma. The results obtained from the neural network pattern recognition are shown to be consistent with traditional medical diagnosis. Input features have also been extracted from the absorbency spectra using chemical factor analysis. These abstract features or factors are also used in the classification.

Retinoids are potent molecules that can affect a variety of fundamental biological processes including cell differentiation and proliferation and apoptosis. These molecules elicit their biological effects by activating a family of nuclear receptors which act as ligand-inducible transcription factors belonging to the steroid/thyroid receptor superfamily. Retinoic acid receptors form heterodimers in which response to ligand binding, both partners contribute to transactivation and/or DNA binding in vivo. Surface-enhanced Raman scattering (SERS), Fourier transform-SERS (FT-SERS), fluorescence and circular dichroism are proposed to rapidly give information on the interaction of the different RARs and RXRs with their specific ligands at physiological concentrations. FT-SERS data reveal a significant attenuation in intensity of the bands originating from the retinoic polyenic chain upon complexation. The spectrum is dominantly of the (Beta) - ionone ring. Fluorescence measurements supported the hydrophobic character of the ligand binding pocket and the circular dichroic data indicate that the protein helices extend upon ligand binding. These novel spectroscopic information are fully consistent with published x-ray crystallographic results and suggest that these techniques may be valuable additional tools to characterize the interactions of agonists and antagonists with residues of the ligand binding pocket retinoid receptor homo- and hetero-dimers.

Native vesicles containing the nicotinic acetylcholine receptor (nAChR) prepared from the electric organ of the ray Torpedo marmorata were used to obtain fluorescence signal sin dependence of different concentrations of the local anesthetics procaine. Nonlinear concentration dependent spectral changes are found using ethidium bromide as a fluorescence marker. Structural changes are found for the proteins including the nAChR in the vesicles during immobilization onto surfaces such as IR transparent germanium (GE) crystal, Ge crystal coated with silver (Ag) cluster to use the SEIRA effect and/or crystals covered with a lipid subphase. A comparison between Ge and Ge coated with Ag (Ge/Ag) clusters reveals increased structural changes in the spectral regions around 1670 cm^-1 upon adsorption of the vesicles on the latter surface. Is the Ge/Ag crystal precoated with a lipid subphase an almost similar spectral contour for the amide I band envelope as in the spectra recorded on a neat Ge crystal is found.

The probability of transplanted skin remaining viable is often difficult to assess visually. The adverse circulatory changes following the surgical elevation of a skin flap limits the supply of oxygen to the flap tissue. Regions of tissue which experience prolonged and severe deprivation of oxygen will not survive. A dorsal rat skin flap model was used to demonstrate the potential of visible/near IR multispectral imaging to detect tissues under hypoxic stress. Images were acquired before and immediately after surgery. Image pre-processing methods were used to enhance tissue contrast and eliminate image artifacts. Principal component analysis of these images further enhanced contrast along the length of the flap while varimax rotation simplified data interpretation. Significant hemodynamic changes were detected (i) between pre- and post-operative images, and (ii) within the post-elevation flap image itself. K-means and fuzzy C-means image segmentation methods were applied to the post-operative multispectral images and proved to be reliable means of predicting regions of tissue that would go on to become visibly necrotic after a 72 h monitoring period. The result suggest that statistical analysis of visible/near IR multispectral images can be used to extract clinically relevant information pertaining to tissue hemodynamics following reconstructive surgery.

Two-layered phantom experiments were performed to examine the influence of a fat layer on measurement of muscle oxygenation using near-IR spectroscopy (NIRS). The phantom consisted of a fat-like layer and a muscle-like layer which were a mixture of agar and TiO2 powder and a suspension of washed bovine blood into 0.55 percent intralipid solution. An LED including 760 and 840 nm elements was used as the optical source, and the reflectance light was detected by photodiodes at source-detector distances of 20, 30 and 40 mm. Curves of optical density changes versus blood volume ratio were obtained with fat-like layer thickness of 0, 5, 10 and 15 mm. It was found that the change in optical density is significantly decreased and that the linearity of measurement characteristics clearly deteriorated by the presence of a fat layer. This strongly suggests that a new algorithm is needed for muscle oxygenation measurement to eliminate the influence of a fat layer. In addition to the phantom experiments, Monte Carlo simulations corresponding to the experiments were performed. Although the simulations showed similar results concerning the influence of a fat layer, it was noted that the changes in optical density obtained from simulations were lower than those of the phantom experiments. This discrepancy was though to be due to the light scattering caused by blood cells.

IR spectroscopy is being widely used to study the biochemical changes associated with cancer. In particular, based upon the hypothesis that biochemical changes associated with cancer precede morphological manifestations of the disease, IR spectroscopy is being evaluated as a potential early diagnostic and prognostic tool. In the current study, IR spectroscopy was applied to the study of colon tissue from rats treated with the specific colon carcinogen azoxymethane, to determine whether tumor induction was associated with identifiable spectroscopic changes in the colon. Characteristic spectra were found for each layer of the colon. Spectra of normal-appearing mucosa and tumors form treated animals then compared to spectra of control mucosa. Differences between tumors and control mucosa were apparent, indicating changes in cellular biochemistry associated with tumor development. In particular, differences in absorptions attributed to nucleic acids were seen, indicating alterations in the structure of cellular DNA in malignant and carcinogen treated tissues. Interestingly, spectra of carcinogen treated rates exhibit characteristics intermediate between those of normal mucosa and tumors. Application of multivariate analysis allowed non-subjective classification of the spectra into three distinct classes with and accuracy of 86.7 percent. The separate classification of control and treated mucosa suggests that IR spectroscopy, when combined with the appropriate classifier, can indeed detect biochemical changes in tissue before physical manifestation of the disease process.

A method for predicting clinically relevant levels of glucose concentration in aqueous solutions from NIR absorbance spectra will be described. The method makes use of the discrete wavelet transform (DWT). The general concepts of the DWT will be briefly reviewed, with emphasis on the properties of the DWT that make it a suitable transform for the analysis of spectroscopic data. The wavelet analysis prediction method will then be described. Results obtained from applying this method to a set of spectra obtained from solutions with varying glucose and protein concentrations will be presented. These result will be compared with the results obtained from using partial least-squares regression.

We report the use of a pseudo-random modulated (PRM), low power laser diode based and dual wavelength NIR instrument for non-invasive, time-resolved spectroscopy (TRS) monitoring and quantification of the cerebral tissue blood supply and oxygenation. In vivo experiment have been conducted with dogs under hypoxia, hypercarbia and hemodilution conditions. Cerebral tissue blood supply and oxygenation are monitored through TRS of photon migration at wavelengths of 670 nm and 810 nm respectively. By (chi) ² curve fitting the measured TRS with effective scattering and absorption coefficients as two variables, both effective scattering and absorption coefficients at each wavelength are extracted. It is found that under hypoxia condition, both scattering and absorption coefficients increases with the hypoxic level. This agrees with the physiology that the blood supply increase while the blood oxygenation decreases under hypoxia. Under hypercarbia condition, both scattering and absorption coefficients of 670 nm decreases with the increase of the hypercarbia level whereas those of 810 nm increases. These correspond to the increased blood supply and oxygenation. Under hemodilution condition, both scattering and absorption coefficients at both wavelengths decreases when the anemia level increases. Therefore, the cerebral tissue blood supply and oxygenation can be monitored and quantified in real-time and non-invasive manner.

Biomedical applications of vibrational spectroscopy developed for routine analysis require methods for data evaluation. Artificial neural networks open a new perspective for the spectra differentiation and identification of biological samples with their small spectra variance. In the present study, the stacked spectral data processing and the following use of neural networks for spectral identification was investigated. 6 different neural network architectures were tested in their capability to built spectral libraries for different bacterial genera and for yeasts, using FTIR and FT-Raman spectra. After developing these libraries, they were connected to a large library, what we called 'multilayered neural networks'. This combines the advantages that the wavelength can be chosen more selective for a given differentiation problem and the network architecture and training function can be more adapted to a special task.

Raman spectroscopy has been sued for the analysis of biological tissue. Preliminary studies, which have been performed ex vivo, indicate that potentially useful diagnostic information may be obtained from the spectra. A new fiber optic-based in vivo Raman system has been constructed which can obtain spectra in vivo from tissue in less than 30 s. Unfortunately, tissue spectroscopy is hindered by the fluorescence and Raman signal generated in the silica fiber optic cables that are used for delivery of the excitation light and collection of the scattered light from the sample. Various fiber optic probes have been evaluated to quantify the extent to which each is able to suppress the fiber optic fluorescence and silica Raman during in vivo measurements. These included (1) standard 'cut-end' bifurcated silica probes, (2) large probe head attachments that utilized filters to eliminate the fiber signal and (3) fiber optic probes which have optimized collection efficiencies and use 'in-the-tip' (ITT) filters to eliminate the spectral contamination. The collection efficiency and the fiber optic suppression capabilities of these probes were compared. Signal response functions were measured and tissue spectra were collected to evaluate the performance of each probe.s