The large content of mitochondria in metabolizing cells, coupled with intrinsic NADH and flavoprotein signals makes these signals ideal for characterizing tissue metabolic states in health and disease. The first few millimeters of tissue are reached by the fluorescence excitation in the exposed surfaces of the cervix, bladder, rectum and esophagus, etc. Thus, extensive use has been made of fluorescent signals by a large number of investigators for tumor diagnosis from an empirical standpoint where the fluorescent signals are generally diminished in precancerous and cancerous tissue. This article reviews the biochemical basis for the fluorescent signals and points to a 'gold standard' for fluorescent signal examination involving freeze trapping and low temperature two- or three-dimensional high resolution fluorescence spectroscopy.

Fluorescence imaging in situ may provide highly specific identification of cell types and altered metabolic activity near the surface of tissue. Most approaches to developing the necessary analytical framework for quantitative 3-D use are based on numerical solutions of some form of transport equation. These are highly computer-intensive and can only be carried out for specified parameters. We apply a random walk model for photon migration which enables us to find an exact expression for the frequency-dependent fluorescent signal emitted from the site of a single fluorophore. Our general expression allows for broad variation of the degree of absorptivity, and is potentially important in providing a basis for the development of fluorescence image reconstruction algorithms.

A study was conducted to determine whether laser-induced fluorescence could detect high grade dysplasia in Barrett's esophagus. Four-hundred-ten nm laser light was used to induce autofluorescence of Barrett's mucosa in 36 patients during routine endoscopy. The spectra were analyzed using the Differential Normalized Fluorescence (DNF) Index technique to differentiate high grade dysplasia from either low grade or non-dysplastic mucosa. Each spectrum was classified as either premalignant or benign using two different DNF indices. Analyzing the fluorescence spectra from all patients using one DNF Index, 96% of non- dysplastic Barrett's samples classified as benign tissue. All low grade dysplasia samples classified as benign. Ninety percent of high grade dysplasia samples classified as premalignant. Twenty-eight percent of mixed low grade/focal high grade dysplasia samples classified as premalignant. In summary, high grade dysplasia in Barrett's esophagus patients can be detected by endoscopic laser-induced fluorescence spectroscopy using differential normalized fluorescence technique.

Laser-induced fluorescence was used for direct in-vivo diagnosis of colon malignancy without requiring biopsy. The methodology was applied in a clinical study in order to differentiate adenomatous polyps from hyperplastic polyps in the colon. The measurements were performed in vivo during routine colonoscopy. Detection of the fluorescence signal from the tissue was performed using laser excitation. This report describes the differential normalized fluorescence (DNF) procedure using the amplified spectral differences between the normalized fluorescence of polyps and normal tissue. Data related to various grades of pathology of colonic tissues are discussed. In this preliminary study, the DNF procedure provides a general trend which corresponds to severity of dysplasia associated with colon malignancy.

Oxidative modification of the major cholesterol carrying lipoprotein, low-density lipoprotein (LDL), renders it more atherogenic as well as inducing unique fluorescence spectral characteristics that distinguish it from native (non-oxidized) LDL. This fluorescence feature has been identified using a microspectrofluorometry system capable of recording autofluorescence of individual cultured macrophages incubated with oxidized LDL. Differences in fluorescence spectra between individual control cells and oxidized LDL loaded cells could also be elicited using dye-enhanced fluorescence with neutral lipid probes such as nile red. Autofluorescence spectroscopy applied to the detection of intracellular oxidized LDL accumulation in circulating monocytes may be useful for identifying a novel risk factor in the assessment of atherosclerosis.

A spectroscopic system with flexible three optical fiber sensor had been developed to study tissue fluorescence for a clinical use. Autofluorescence spectra at 413 nm and 10 mW excitation light power from different tissues in oral cavity had been measured in vivo in 25 subjects. The correlation coefficient in spectral shape between individual spectra and the mean emission spectrum of each site was about 0.9 and fluorescence intensity variation ranged between 20% and 45% according to the examined site. The variation in fluorescence intensity of the main emission wavelength at about 520 nm between spectra of the lower part of tongue, gingiva, lips, floor of cavity, cheek and palate was not statistically significant. But the spectrum of the upper part of tongue had been characterized by an additional peak around 635 nm. Otherwise, autofluorescence spectra at 410 nm and 0.5 mW excitation light power of 8 carcinoma of buccal and lung tissues were measured. The fluorescence ratio at 520 emission peak between normal tissue and carcinoma was evaluated at a maximum value of 13 for a lung cancer (ex vivo measurement) and a minimum of 3.3 for a cancer of the oro-pharynx (in vivo measurement). On the other hand, a fluorescence peak at 635 nm had characterized the carcinoma of the floor of cavity and the upper part of tongue.

Fluorescence spectroscopy was applied to characterize normal, malignant and adipose breast tissues. Excitation, emission, and synchronized diffusive reflectance spectral scans were measured on over one hundred specimens for the purpose of developing an improved spectroscopic diagnostic technique. These techniques were able to successfully distinguish malignant tissue from adipose glandular fibrous and normal tissue. A sensitivity of 91% for fifty-six (56) malignant specimens with specificity of 91% for forty-six (46) benign tissue specimens has been achieved, using pathology as the golden standard.

Fluorescence spectroscopy of diseased tissues, including chemical-induced rat liver, kidney and testis lesions, as well as murine mammary tumor, was studied. The rat liver, kidney and testis tissues were excited by radiation of 350 and 366 nm, which appeared to provide the optimal differentiation between normal and lesion tissues; the tumor tissues were excited by both 350 nm and 775 nm wavelengths. In comparison with normal liver tissue, at (lambda) ex equals 366 nm, the fluorescent spectrum of liver lesion showed a clear red shift around the emission peak of 470 nm, the major native fluorescent peak of organized tissue. When excited by 350 nm wavelength, all the chemically induced lesion tissues (liver, kidney and testis) appeared to cause a significant reduction of emission intensity at the 470 nm peak. While the 775 nm excitation did not reveal any significant difference among tumor, muscle and skin tissues, the 350 nm excitation did provide some interesting features among the tumor tissues at different stages. Compared with muscle tissue, the viable tumor showed an overall reduction of emission intensity around 470 nm. In addition, the viable tumor tissue showed a secondary emission peak at 390 nm with necrotic tumor tissue having a reduced intensity. The histology of both viable and necrotic tumor tissue was examined and appeared to correlate with the results of the fluorescent spectroscopy observation.

Ultraviolet resonance Raman (UVRR) spectroscopy was used to characterize normal and diseased colon mucosa in vitro. A tunable mode-locked Titanium:Sapphire laser operating at 76 MHz was used to irradiate normal and diseased colon tissue samples with 251 nm light generated from the third harmonic of the fundamental radiation. The Raman scattered light was collected and analyzed using a 1 meter spectrometer fitted with a UV coated, liquid nitrogen cooled CCD detector. The measured spectra show prominent bands that correspond to those of known tissue constituents including nucleic acids, aromatic amino acids and lipids. Using the Raman lineshapes measured from pure solutions of nucleotides, tryptophan, tyrosine, FAD, and from lipid-rich serosal fat, the colon spectra were modeled by a least square fitting algorithm whereby the colon spectra were assumed to be a linear combination of the pure biochemical lineshapes. The relative Raman scattering cross section of each biochemical was determined so that the relative concentration of each compound with respect to the others, could be extracted from a given tissue spectrum.

In this study a rigorous approach to tissue fluorescence is presented, based on the study of tissue fluorescence as an electromagnetic scattering problem. Fluorescence scattered wave is treated by taking a continuous spectrum distribution in a region of frequencies lower and equal to the excitation frequency. The existence of inelastic field components can be considered as a result of the particular form, that the polarization of the irradiated medium has. In order to provide the most general formulation, the polarization vector P for the observed light frequency (omega) can be written as P(r,(omega) ) equals ^(omega )(integral) _(omega )^o d(omega) 'E(r,(omega) ')(tau) ((omega) ,(omega) '), where E(r,(omega) ') is the electric field at the excitation frequency (omega) ' and (tau) ((omega) ,(omega) ') the transfer permittivity function from (omega) ' at the spatial point r, to the emission frequency (omega) , measured at the same point. Substitution of the polarization vector into the electromagnetic field equations leads to a formulation of the inelastic field components. The model used is based on considering tissue as a single dielectric layer, under pulse excitation. The theoretical background for such an evaluation, together with the mathematical technique used and the theoretical results, are presented.

During the past several years a range of spectroscopies, including fluorescence and elastic- scatter spectroscopy, have been investigated for optically based detection of cancer and other tissue pathologies. Both elastic-scatter and fluorescence signals depend, in part, on scattering and absorption properties of the cells in the tissue. Therefore an understanding of the scattering and absorption properties of cells is a necessary prerequisite for understanding and developing these techniques. Cell suspensions provide a simple model with which to begin studying the absorption and scattering properties of cells. In this study we have made preliminary measurements of the scattering and absorption properties of suspensions of mouse mammary carcinoma cells (EMT6) over a broad wavelength range (380 nm to 800 nm).

Optical measurements of the cervical transformation zone (sometimes referred to as the transition zone) using elastic-scattering spectroscopy, demonstrate sensitivity to the epithelial cell-type differences.

The optical properties of biological tissue should be determined in vivo whenever possible. However, for those instances when in vivo studies are impractical, too expensive or inappropriate, and when blood flow is not an issue, the ability to perform in vitro studies then becomes invaluable. Optical absorption spectroscopy shows that it may be possible to obtain meaningful information about the optical properties of human breast tissue from in vitro samples if strict preparation and measuring protocols are used. That a strict protocol for storing and handling tissue is critical can be seen from our observations of changes in the optical absorption spectra that occur in response to formalin fixation, the passage of time, application of stains and dyes, and storage in growth medium of the excised tissue. In vivo optical absorption spectroscopy measurements have been made on human breast cancer xenografts and compared with in vitro measurements on breast biopsies prepared according to precise collection and treatment protocols. There is a 'window of opportunity' before time dependent changes in the UV optical absorption spectra of the excised tissue specimens occur. This time window of opportunity widens at longer wavelengths with the least changes occurring in the optical spectra in the NIR.

Phosphorescence and fluorescence spectra of human breast tissues excited at two wavelengths were measured and compared at low temperature. A blue shift of fluorescence spectral peak was observed at low temperature for the excitation wavelength at 300 or 340 nm. The phosphorescence spectra excited at 300 nm were found to be centered at 412 nm while the fluorescence spectra consisted of two bands located at about 340 nm and 440 nm. For 340 nm excitation, the phosphorescence spectral profile was found to be relatively smooth but shifted to the red centered at 480 nm while the fluorescence consisted of two bands at 375 nm and 440 nm.

The aim of this study was to develop quantitative methods for relating the microstructure of a tissue to the magnitude and wavelength dependence of its scattering coefficient. Two methods, cell counting and spatial frequency analysis, were used to estimate the distribution of sizes of structures imaged by light and electron microscopy. We found that scatterers in the epidermal layer of the skin exhibit a log-normal size distribution, whereas the spatial fluctuations in the index of refraction of dense fibrous tissues, such as the dermis, follow a power law. The correlations in the refractive indices of a variety of tissues exhibit characteristics of a random fractal with a Hurst coefficient between 0.3 and 0.5. Calculated from the measured distributions and volume fractions, the magnitudes of the scattering coefficient and anisotropy parameters of the tissue were found to be within the range 10 less than (mu) _s less than 35 mm^-1 and 0.7 less than g less than 0.97, depending on wavelength and tissue structure. Our results suggest that analysis of histological images of tissues is a viable method for estimating the optical parameters of tissues and their wavelength dependence.

Native fluorescence spectroscopy of biomolecules has emerged as a new modality to the medical community in characterizing the various physiological conditions of tissues. In the past several years, many groups have been working to introduce the spectroscopic methods to diagnose cancer. Researchers have successfully used native fluorescence to distinguish cancerous from normal tissue samples in rat and human tissue. We have developed three generations of instruments, called the CD-scan, CD-ratiometer and CD-map, to allow the medical community to use optics for diagnosing tissue. Using ultraviolet excitation and emission spectral measurements on both normal and cancerous tissue of the breast, gynecology, colon, and aerodigestive tract can be separated. For example, from emission intensities at 340 nm to 440 nm (300 nm excitation), a statistically consistent difference between malignant tissue and normal or benign tissue is observed. In order to utilize optical biopsy techniques in a clinical setting, the CD-scan instrument was developed, which allows for rapid and reliable in-vitro and in-vivo florescence measurements of the aerodigestive tract with high accuracy. The instrumentation employs high sensitivity detection techniques which allows for lamp excitation, small diameter optical fiber probes; the higher spatial resolution afforded by the small diameter probes can increase the ability to detect smaller tumors. The fiber optic probes allow for usage in the aerodigestive tract, cervix and colon. Needle based fiber probes have been developed for in-vivo detection of breast cancer.

Details of the interaction of photons with tissue phantoms are elucidated using Monte Carlo simulations. In particular, photon sampling volumes and photon pathlengths are determined for a variety of scattering and absorption parameters. The Monte Carlo simulations are specifically designed to model light delivery and collection geometries relevant to clinical applications of optical biopsy techniques. The Monte Carlo simulations assume that light is delivered and collected by two, nearly-adjacent optical fibers and take into account the numerical aperture of the fibers as well as reflectance and refraction at interfaces between different media. To determine the validity of the Monte Carlo simulations for modeling the interactions between the photons and the tissue phantom in these geometries, the simulations were compared to measurements of aqueous suspensions of polystyrene microspheres in the wavelength range 450 - 750 nm.

We have constructed a new semiconductor laser device that may be useful in high speed characterization of cell morphology for diagnosis of disease. This laser device has critical advantages over conventional cell florescence detection methods since it provides intense, monochromatic, low-divergence light signals that are emitted from lasing modes confined by a cell. Further, the device integrates biological structures with semiconductor materials at the wafer level to reduce device size and simplify cell preparation. In this paper we discuss operational characteristics of the prototype cytometer and present preliminary data for blood cells and dielectric spheres.

We report on an analysis of diffuse reflectance spectra measured in conjunction with the fluorescence from normal human breast tissues and malignant breast tumors. The diffuse reflectance spectra from excised, air-equilibrated, human breast tissue samples show lower fractions of oxygenated hemoglobin and higher content of ferric (Fe³⁺) heme in malignant breast tumor samples than in normal breast tissues. Normal tissues are found to be easily deoxygenated and reoxygenated, but malignant tumors usually do not change their state as much. An analysis of tissue oxygenation parameters is discussed with respect to an enhancement of predictive power of fluorescence diagnostic method. The oxygenation state of tissues may be used as an additional marker in cancer diagnostics.

Fluorescence decay characteristics are used to identify biologic fluorophores and to characterize interactions with the fluorophore environment. In many studies, fluorescence lifetimes are assessed by iterative reconvolution techniques. We investigated the use of a new approach: the Laguerre expansion of kernels technique (Marmarelis, V.Z., Ann. Biomed., Eng. 1993; 21, 573-589) which yields the fluorescence impulse response function by least- squares fitting of a discrete-time Laguerre functions expansion. Nitrogen (4 ns FWHM) and excimer (120 ns FWHM) laser pulses were used to excite the fluorescence of an anthracene and of type II collagen powder. After filtering (monochromator) and detection (MCP-PMT), the fluorescence response was digitized (digital storage oscilloscope) and transferred to a personal computer. Input and output data were deconvolved by the Laguerre expansion technique to compute the impulse response function which was then fitted to a multiexponential function for determination of the decay constants. A single exponential (time constant: 4.24 ns) best approximated the fluorescence decay of anthracene, whereas the Type II collagen response was best approximated by a double exponential (time constants: 2.24 and 9.92 ns) in agreement with previously reported data. The results of the Laguerre expansion technique were compared to the least-squares iterative reconvolution technique. The Laguerre expansion technique appeared computationally efficient and robust to experimental noise in the data. Furthermore, the proposed method does not impose a set multiexponential form to the decay.

The progression of disease is certainly accompanied by biochemical changes. Since early detection promises a greater efficacy for therapeutic intervention, non-invasive biomedical optics may offer the opportunity for detecting biochemical changes thereby improving prognosis by providing early diagnosis. Magnetic resonance imaging (MRI) is an example of a modality that has successfully monitored the relaxation of spin states of paramagnetic nuclei in order to provide biomedical imaging and biochemical spectroscopy of tissues. In this paper, we discuss an optical analog of MRI, called fluorescence lifetime imaging. instead of monitoring the relaxation of a spin state, lifetime imaging depends upon monitoring the relaxation of an electronically activated state which is brought about by the absorption of a photon. Figure 1 is the Jablonski diagram outlining the electronic transitions which occur following the absorption of light. Relaxation from the activated state to the ground state can occur via either non-radiative or fluorescent decay, depending upon the local environment. The mean time between the events of the absorption of an excitation photon and the radiative relaxation process which produces a fluorescent photon is known as the lifetime, (tau) , of the activated state. Typically, endogenous fluorophores have lifetimes on the order of nanoseconds while exogenous compounds have lifetimes ranging from sub nanosecond to tens of nanoseconds. If the spin state of the electron in the activated state is 'flipped' (referred to as a 'intersystem crossing') then radiative relaxation requires extra time since paired electrons of the same spin are forbidden. Consequently, the resulting phosphorescence lifetime is on the order of milliseconds. The lifetime of the activated probe is dependent on its environment much the same as the time for relaxation of the spin states of paramagnetic nuclei in MRI and NMR depends upon local environment. There are two mechanisms responsible for reducing the lifetime of an activated probe: (1) energy transfer from the activated state to a donor molecule(s), and (2) collisional quenching of the activated state. For an activated probe experiencing collisional quenching, the measured lifetime, (tau) , can be used to determine metabolite or quencher concentration [Q], from the empirical Stern-Volmer relationship. Consequently, a map of lifetime provides a map of metabolite or quencher concentration. In this paper, we report computational experiments which point to the feasibility of the optical analog to MRI: fluorescence lifetime imaging in tissues and other scattering media.

Fluorescence signal during tissue LIF-analysis depends on both excitation conditions and tissue optics, registration optics and location of the probe relative to tissue sample as well. To develop reliable fiber optic probes and optimize their position the spatial distribution of tissue fluorescence should be studied. Fluorescence indicatrices of skin of rat were measured in angular range of 80 degrees. Excited light from He-Cd (20 mW, 442 nm) laser was delivered on the cutaneous surface at the angular range from 0 to 60 degrees with the tissue surface. Fluorescence was registered in the spectrum between 530 nm and 700 nm with 1.5 nm resolution. Autofluorescence of the skin of 5 white rats was studied in-vivo. Local application of sensitizer hypericin was used for stimulated fluorescence studies. Fluorescence indicatrices were not corresponded to scattering ones under the same conditions and depended on incident angle of excitative laser beam. No influence of polarization of excitative beam on outside fluorescence distribution was observed. Maximum in-vivo fluorescence yield was registered at the normal incidence. There were observed marked differences between spatial distribution of normal and photosensitized rat skin tissues.

The use of reflection spectrophotometry to measure the spectra of oxy-hemoglobin and deoxy- hemoglobin, strong absorbers of light in the visible region of the spectrum, is a well established method for determining tissue oxygenation. This type of spectral measurement is typically made with a point-spectrometer and provides information only at a single point. An imaging spectrometer, on the other, can measure the hemoglobin spectra at every pixel in the image, thus providing a two-dimensional (spatial) map of tissue ischemia. A novel spectral bio-imaging system based on the SpectraCube^TM technology, an optical method based on proven Fourier transform (FT) spectroscopy, has been applied successfully in intact rat brain to measure oxy- and deoxy-hemoglobin spectra. Spectral images containing 10,000 spectra were acquired in a rat ventilated with 30% O₂, and repeated when the inspired gas mixture was switched for 45 seconds to 100% nitrogen. Differences in hemoglobin spectra corresponding to real differences in tissue oxygenation are readily apparent under these two conditions. There is also some evidence that information concerning cytochromes is present in these spectral images, and algorithms are currently being developed to extract the signatures of cytochromes. Details of the spectral bio-imaging system and the results of the measurements made in intact rat brain are discussed.

Cancer is a disease whose cause has been eluding researchers. This paper details a new method for the diagnosis of cancer using characteristic autofluorescence from tissues. The experiments were carried out using a nitrogen laser, spectrofluorometer and tissue samples of normal and cancer tissues. The fluorescence spectra of various samples were obtained and observed both by in-vivo and in-vitro studies. Results from these studies have been detailed. The paper also presents a hypothesis on the origin of cancer.

Implementation of optical imagery in a diffuse inhomogeneous medium such as biological tissue requires an understanding of photon migration and multiple scattering processes which act to randomize pathlength and degrade image quality. The nature of transmitted light from soft tissue ranges from the quasi-coherent properties of the minimally scattered component to the random incoherent light of the diffuse component. Recent experimental approaches have emphasized dynamic path-sensitive imaging measurements with either ultrashort laser pulses (ballistic photons) or amplitude modulated laser light launched into tissue (photon density waves) to increase image resolution and transmissive penetration depth. Ballistic imaging seeks to compensate for these 'fog-like' effects by temporally isolating the weak early-arriving image-bearing component from the diffusely scattered background using a subpicosecond optical gate superimposed on the transmitted photon time-of-flight distribution. The authors have developed a broadly wavelength tunable (470 nm - 2.4 micrometer), ultrashort amplifying optical gate for transillumination spectral imaging based on optical parametric amplification in a nonlinear crystal. The time-gated image amplification process exhibits low noise and high sensitivity, with gains greater than 10⁴ achievable for low light levels. We report preliminary benchmark experiments in which this system was used to reconstruct, spectrally upcovert, and enhance near-infrared two-dimensional images with feature sizes of 65 micrometer/mm² in background optical attenuations exceeding 10¹². Phase images of test objects exhibiting both absorptive contrast and diffuse scatter were acquired using a self-referencing Shack-Hartmann wavefront sensor in combination with short-pulse quasi-ballistic gating. The sensor employed a lenslet array based on binary optics technology and was sensitive to optical path distortions approaching lambda/100.

Cellular differentiation is a characteristic that defines normal and neoplastic tissue. Epithelial tumors express keratins that are quantitatively and qualitatively different from normal tissue. Therefore, a study was designed to test the hypothesis that native cellular fluorescence (NCF) can be used to detect in vivo differentiation changes associated with higher keratin content in epidermal thick skin and callus when compared with non-keratinized forearm skin. Nineteen control subjects (males, N equals 12; females, N equals 7) were tested using a xenon based fluorescence spectrophotometer (Mediscience Corp.). Seven emission and five excitation spectra were recorded from two sites on each subject: the ventral forearm and palmar hand at the metacarpal-phalangeal joint. Wavelength ratios were used to quantitate three emission and three excitation scans. Two sample t-test analysis was performed for forearm (FA) vs. hand thick skin (HTS) and callus (CAL). The emission scans ((lambda) Ex 300 nm, (lambda) Em 320 - 580 nm; (lambda) Ex 340 nm, (lambda) Em 360 - 660 nm; (lambda) Ex 365 nm, (lambda) Em 400 - 700 nm) are statistically significant (P less than 0.0001) at the wavelength ratios 340 nm/440 nm, 420 nm/540 nm, and 440 nm/540 nm, respectively. The excitation scans ((lambda) Ex 200 - 320 nm, (lambda) Em 340 nm; (lambda) Ex 200 - 360 nm, (lambda) Em 380 nm; (lambda) Ex 270 - 500 nm, (lambda) Em 520 nm) are statistically significant (P less than 0.0001) at the wavelength ratios 260 nm/290 nm, 290 nm/340 nm, and 320 nm/345 nm, respectively. In vivo tissue fluorescence can be used to detect the influence of keratin on the spectral profiles of the epidermis.

A rapid micro-lightguide spectrometer (EMPHO II) coupled to an automatic three axis positioning system enables very precise and fast 2D-scans at the surface of human skin. The positioning accuracy amounts to 1 micrometer. This allows measurements with excellent spatial reproducibility. With this system examinations of local distribution of HbO₂ and Hb have been performed in human skin. For this purpose at the back of the hand areas of 5 by 5 mm to 5 by 10 mm were scanned in defined steps of 100 micrometers. Functional images of local hemoglobin concentration and hemoglobin oxygenation of microscopical structures have been resolved by use of 250 micrometer lightguide sensors. Two-dimensional-images of local oxygen supply parameters corresponding directly to morphological structures of human skin have been gained. The local pattern matches the distribution of the papillas of the corium. In the papillas the capillary loops supplying the lower part of the epidermis are situated. The measured parameters describe very exactly the local oxygen supply situation of the area under investigation.

Determining the concentration of glucose in vivo by near-infrared spectroscopy is complicated by the effects of optical changes caused by fluctuations in temperature, tissue water content and the concentration of other analytes. Mie theory and Monte Carlo computer simulation of light transport in optically absorbing and scattering media were used to investigate the magnitude of the changes in diffuse reflectance and transmittance from changes in glucose. Similarly, the possible interference in the glucose measurement from the other tissue parameters have been assessed and found to be significant.

We have designed a program using laser induced autofluorescence spectroscopy as a possible way to characterize urothelial tumors of the bladder. The autofluorescence spectra were compared between normal, suspicious and tumor areas of human bladder. Three different pulsed laser wavelengths were used for excitation: 308 nm (excimer), 337 nm (nitrogen) and 480 nm (dye laser). Excitation light was delivered by a specially devised multifiber catheter introduced through the working channel of a regular cystoscope under saline irrigation. The fluorescence light was focused into an optical multichannel analyzer detection system. The data was evaluated in 25 patients immediately before resection of a bladder tumor. Spectroscopic results were compared with histopathology. Upon 337 nm and 480 nm excitations, the overall intensity of the fluorescence spectra from bladder tumors was clearly reduced in comparison with normal urothelium, regardless of the stage and the grade of the tumor. upon 308 nm excitation, the shape of tumor fluorescence spectra, including carcinoma in situ, differed drastically from that of normal tissue. In this case, no absolute intensity measurements are needed and clear diagnosis can be achieved from fluorescence intensity ratio (360/440 nm). This spectroscopic study could be particularly useful for the design of a simplified autofluorescence imaging device for real-time routine detection of occult urothelial neoplastic lesions.