Spectral bio-imaging of the eye

Abstract

A spectral bio-imaging method for enhancing pathologic, physiologic, metabolic and health related spectral signatures of an eye tissue, the method comprising the steps of (a) providing an optical device for eye inspection being optically connected to a spectral imager; (b) illuminating the eye tissue with light via the iris, viewing the eye tissue through the optical device and spectral imager and obtaining a spectrum of light for each pixel of the eye tissue; and (c) attributing each of the pixels a color according to its spectral signature, thereby providing an image enhancing the spectral signatures of the eye tissue.

Description

FIELD AND BACKGROUND OF THE INVENTION

[0002] The present invention relates to spectral imaging in general and, more particularly, to spectral bio-imaging of the eye which can be used for non-invasive early detection and diagnosis of eye diseases and for detection of spatial organization, distribution and quantification of cellular and tissue natural constituents, structures and organelles, tissue vitality, tissue metabolism, tissue viability, etc., using light reflection, scattering and emission, with high spatial and spectral resolutions.

[0003] A spectrometer is an apparatus designed to accept light, to separate (disperse) it into its component wavelengths and measure a spectrum, that is the intensity of the light as a function of its wavelength. An imaging spectrometer (also referred to hereinbelow as a spectral imager) is one which collects incident light from a scene and measures the spectra of each pixel or picture element thereof.

[0004] Spectroscopy is a well known analytical tool which has been used for decades in science and industry to characterize materials and processes based on the spectral signature of chemical constituents. The physical basis of spectroscopy is the interaction of light with matter. Traditionally, spectroscopy is the measurement of the light intensity emitted, transmitted, scattered or reflected from a sample, as a function of wavelength, at high spectral resolution, but without any spatial information.

[0005] Spectral imaging, on the other hand, is a combination of high resolution spectroscopy and high resolution imaging (i.e., spatial information). The closest work so far described with respect to the eye concerns either obtaining high spatial resolution information, yet providing only limited spectral information, for example, when high spatial resolution imaging is performed with one or several discrete band-pass filters [see, for example, Patrick J. Saine and Marshall E. Tyler, Ophthalmic Photography, A textbook of retinal photography, angiography, and electronic imaging, Butterworth-Heinemann, Copyright 1997, ISBN 0-7506-9793-8, p. 72], or alternatively, obtaining high spectral resolution (e.g., a full spectrum), yet limited in spatial resolution to a small number of points of the eye or averaged over the whole sample [See for example, Delori F. C., Pfilbsen K. P., Spectral reflectance of the human ocular fundus, Applied Optics Vol. 28, pp. 1061-1077, 1989].

[0006] Conceptually, a spectral imaging system consists of (i) a measurement system, and (ii) an analysis software. The measurement system includes all of the optics, electronics, illumination source, etc., as well as calibration means best suited for extracting the desired results from the measurement. The analysis software includes all of the software and mathematical algorithms necessary to analyze and display important results in a meaningful way.

[0007] Spectral imaging has been used for decades in the area of remote sensing to provide important insights in the study of Earth and other planets by identifying characteristic spectral absorption features. However, the high cost, size and configuration of remote sensing spectral imaging systems (e.g., Landsat, AVIRIS) has limited their use to air and satellite-born applications [See, Maymon and Neeck (1988) Proceedings of SPIE - Recent Advances in Sensors, Radiometry and Data Processing for Remote Sensing, 924, pp. 10-22; Dozier (1988) Proceedings of SPIE—Recent Advances in Sensors, Radiometry and Data Processing for Remote Sensing, 924, pp. 23-30].

[0008] There are three basic types of spectral dispersion methods that might be considered for a spectral bio-imaging system: (i) spectral grating and/or prism, (ii) spectral filters and (iii) interferometric spectroscopy.

[0009] In a grating/prism (i.e., monochromator) based systems, also known as slit-type imaging spectrometers, such as for example the DILOR system: [see, Valisa et al. (Sep. 1995) presentation at the SPIE Conference European Medical Optics Week, BiOS Europe '95, Barcelona, Spain], only one axis of a CCD (charge coupled device) array detector (the spatial axis) provides real imagery data, while a second (spectral) axis is used for sampling the intensity of the light which is dispersed by the grating as function of wavelength. The system also has a slit in a first focal plane, limiting the field of view at any given time to a line of pixels. Therefore, a full image can only be obtained after scanning the grating or the incoming beam in a direction parallel to the spectral axis of the CCD in a method known in the literature as line scanning. The inability to visualize the two-dimensional image before the whole measurement is completed makes it impossible to choose, prior to making a measurement, a desired region of interest from within the field of view and/or to optimize the system focus, exposure time, etc. Grating based spectral imagers are popular in use for remote sensing applications, because an airplane (or satellite) flying over the surface of the Earth provides the system with a natural line scanning mechanism.

[0010] It should be further noted that slit-type imaging spectrometers have a major disadvantage since most of the pixels of one frame are not measured at any given time, even though the fore- optics of the instrument actually collects incident light from all of them simultaneously. The result is that either a relatively large measurement time is required to obtain the necessary information with a given signal-to-noise ratio, or the signal-to-noise ratio (sensitivity) is substantially reduced for a given measurement time. Furthermore, slit-type spectral imagers require line scanning to collect the necessary information for the whole scene, which may introduce inaccuracies to the results thus obtained.

[0011] Filter based spectral dispersion methods can be further categorized into discrete filters and tunable filters. In these types of imaging spectrometers the spectral image is built by filtering the radiation for all the pixels of the scene simultaneously at a different wavelength at a time by inserting in succession narrow band filters in the optical path, or by electronically scanning the bands using AOTF or LCTF (see below).

[0012] Similarly to the slit type imaging spectrometers equipped with a grating, as described above, while using filter based spectral dispersion methods, most of the radiation is rejected at any given time. In fact, the measurement of the whole image at a specific wavelength is possible because all the photons outside the instantaneous wavelength measured are rejected and do not reach the CCD.

[0013] Tunable filters, such as acousto-optic tunable filters (AOTFs) and liquid- crystal tunable filter (LCTFs) have no moving parts and can be tuned to any particular wavelength in the spectral range of the device in which they are implemented. One advantage of using tunable filters as a dispersion method for spectral imaging is their random wavelength access; i.e., the ability to measure the intensity of an image at a number of wavelengths, in any desired sequence without the use of a mechanical filter wheel. However, AOTFs and LCTFs have the disadvantages of (i) limited spectral range (typically, λmax=2λmin) while all other radiation that falls outside of this spectral range must be blocked, (ii) temperature sensitivity, (iii) poor transmission, (iv) polarization sensitivity, and (v) in the case of AOTFs an effect of shifting the image during wavelength scanning.

[0014] All these types of filter and tunable filter based systems have not been used successfully and extensively over the years in spectral imaging for any application, because of their limitations in spectral resolution, low sensitivity, and lack of easy-to-use and sophisticated software algorithms for interpretation and display of the data.

[0015] No literature has been found by the inventors of the present invention describing high resolution spectroscopy combined with high resolution imaging applied to the eye.

[0016] A method and apparatus for spectral analysis of images which have advantages in the above respects was disclosed in U.S. Pat. No. 5,539,517 to Cabib et al., which is incorporated by reference as if fully set forth herein, with the objective to provide a method and apparatus for spectral analysis of images which better utilizes all the information available from the collected incident light of the image to substantially decrease the required frame time and/or to substantially increase the signal-to-noise ratio, as compared to the conventional slit- or filter type imaging spectrometer and does not involve line scanning.

[0017] According to this invention, there is provided a method of analyzing an optical image of a scene to determine the spectral intensity of each pixel thereof by collecting incident light from the scene; passing the light through an interferometer which outputs modulated light corresponding to a predetermined set of linear combinations of the spectral intensity of the light emitted from each pixel; focusing the light outputted from the interferometer on a detector array, scanning the optical path difference (OPD) generated in the interferometer for all pixels independently and simultaneously and processing the outputs of the detector array (the interferograms of all pixels separately) to determine the spectral intensity of each pixel thereof.

[0018] This method may be practiced by utilizing various types of interferometers wherein the OPD is varied to build the interferograms by moving the entire interferometer, an element within the interferometer, or the angle of incidence of the incoming radiation. In all of these cases, when the scanner completes one scan of the interferometer, the interferograms for all pixels of the scene are completed.

[0019] Apparatuses in accordance with the above features differ from the conventional slit- and filter type imaging spectrometers by utilizing an interferometer as described above, therefore not limiting the collected energy with an aperture or slit or limiting the incoming wavelength with narrow band interference or tunable filters, thereby substantially increasing the total throughput of the system.

[0020] Thus, interferometer based apparatuses better utilize all the information available from the incident light of the scene to be analyzed, thereby substantially decreasing the measuring time and/or substantially increasing the signal-to-noise ratio (i.e., sensitivity).

[0021] Consider, for example, the “whisk broom” design described in John B. Wellman (1987) Imaging Spectrometers for Terrestrial and Planetary Remote Sensing, SPIE Proceedings, Vol. 750, p. 140. Let n be the number of detectors in the linear array, m×m the number of pixels in a frame and T the frame time. The total time spent on each pixel in one frame summed over all the detectors of the array is nT/m2. By using the same size array and the same frame rate in a method according to the invention described in U.S. Pat. No. 5,539,517, the total time spent summed over all the detectors on a particular pixel is the same, nT/m2.

[0022] However, whereas in the conventional grating method the energy seen by every detector at any given time is of the order of 1/n of the total, because the wavelength resolution is 1/n of the range, in a method according to the invention described in U.S. Pat. No. 5,539,517 the energy is of the order of unity because the modulating function is an oscillating function (e.g., sinusoidal (Michelson) or a similar periodic function, such as the low finesse Airy function with Fabry-Perot) whose average over a large OPD range is 50%. Based on the standard treatment of the Fellgett advantage (or multiplex advantage) described in interferometry textbooks [for example, see, Chamberlain (1979) The principles of interferometric spectroscopy, John Wiley and Sons, pp. 16-18 and p. 263], it is possible to show that devices according to this invention have measurement signal-to-noise ratios which are improved by a factor of n0.5 in the cases of noise limitations in which the noise level is independent of signal (system or background noise limited situations) and by the square root of the ratio of the signal at a particular wavelength to the average signal in the spectral range, at wavelengths of a narrow peak in the cases the limitation is due to signal photon noise.

[0023] Thus, according to the invention described in U.S. Pat. No. 5,539,517, all the required OPDs are scanned simultaneously for all the pixels of the scene in order to obtain all the information required to reconstruct the spectrum, so that the spectral information is collected simultaneously with the imaging information.

[0024] Spectral bio-imaging systems are potentially useful in all applications in which subtle spectral differences exist between chemical constituents whose spatial distribution and organization within an image are of interest. The measurement can be carried out using virtually any optical system attached to the system described in U.S. Pat. No. 5,539,517, for example, an upright or inverted microscope, a fluorescence microscope, a macro lens, an endoscope or a fundus camera. Furthermore, any standard experimental method can be used, including light transmission (bright field and dark field), autofluorescence and fluorescence of administered probes, light transmission, scattering and reflection.

[0025] Fluorescence measurements can be made with any standard filter cube (consisting of a barrier filter, excitation filter and a dichroic mirror), or any customized filter cube for special applications, provided that the emission spectra fall within the spectral range of the system sensitivity.

[0026] Spectral bio-imaging can also be used in conjunction with any standard spatial filtering method such as dark field and phase contrast, and even with polarized light microscopy. The effects on spectral information when using such methods must, of course, be understood to correctly interpret the measured spectral images.

[0027] Reflection of visible light from the ocular fundus has been used for many years for research and for routine eye inspection by ophthalmologists. It is also the basis for recording the eye status of a patient for disease and treatment follow up, both as pictures on a camera film and as digital images in the computer memory.

[0028] In contrast, the spectral dependence of the light reflection from different regions of the fundus has been relegated only to research work. The reasons for these facts are (i) images are a very direct means of presenting information to a human being, because they are easily interpreted, compared and remembered by the human brain; (ii) spectral data are much less direct, are not immediately understandable, and to be useful they must usually undergo several layers of mathematical processing before they are related to the bio-physiological properties of the tissue in question; and (iii) there has been so far no affordable instrumentation available to collect and analyze spectral data from the fundus, which is easy to use, fast, and reliable for a research or clinical setting.

[0029] As a result, at present, the use of spectral information in many fields, and in particular in ophthalmology, is lagging enormously behind the imaging techniques.

[0030] Recently, Applied Spectral Imaging Ltd. of Migdal Haemek, Israel, has developed the SPECTRACUBE technology. The SPECTRACUBE technology is based on an interferometer based spectral imager and as such it combines spectroscopy and imaging to use the advantages of both. It collects spectral data from all the pixels of an image simultaneously so that, after appropriate processing, the important chemical composition of the studied object (related to its bio- physiological properties) can be mapped and visualized.

[0031] The SPECTRACUBE technology was employed for spectral (color) karyotyping which simplifies and improves the detection capability of chromosomal aberrations using fluorescence emission [see, Multicolor spectral karyotyping of human chromosomes. E. Schroeck et al., Science, 273, 494-497, 1996; Multicolor spectral karyotyping of mouse chromosomes. Marek Liyanage et al. Nature Genetics p. 312-315, 1996; Spectral Karyotyping. Yuval Garini, et al. Bioimaging 4, p. 65-72, 1996; Hidden chromosome abnormalities in haemotological malignancies detected by multicolor spectral Karyotyping. Tim Veldman, Christine Vignon, Evelin Schrock, Janet D. Rowley & Thomas Ried. Nature Genetics, April, 1997: 406-410.; Spectral Karyotyping: Chromosomes in Color. Turid Knutsen, Tim Veldman, Hesed Padilla-Nash, Evelin Schrock, Morek Liyanage, Thomas Ried. Applied Cytogenetics, 23(2) 1997, pp. 26-32.; and Early Experiences with SKY: A Primer for the Practicing Cytogenetic Technologist. Michele Shuster, Ulrike Bockmuhl, Susanne M. Gollin. Applied Cytogenetics, 23(2) 1997, pp. 33-37].

[0032] Diabetic retinopathy is a potentially visually devastating condition that, in most cases, can be controlled with timely laser treatment [Ferris (1993) (commentary) JAMA 269:1290-1291]. The American Academy of Ophthalmology has suggested screening schedules to detect when patients develop clinical conditions which should be treated [Diabetic Retinopathy: American Academy of Ophthalmology Preferred Practice Patterns. San Francisco, Cal.: American Academy of Ophthalmology Quality of Care Committee Retinal Pane, American Academy of Ophthalmology, 1989].

[0033] However the suggested screening schedule is expensive, and for some individuals even the current expensive screening is not sufficient because patients occasionally develop severe retinopathy between scheduled examinations. In spite of this, it has been shown that this screening is cost effective [Javitt et al. (1989) Ophthalmology 96:255-64]. This work shows that a large amount of money could be saved in health care follow up, if high and low risk patients could be more effectively identified. Therefore, any method that could increase the accuracy and reduce the cost of screening for diabetic retinopathy would be of high clinical value.

[0034] Currently, the recommended screening evaluation for diabetic retinopathy includes a detailed retinal evaluation and, in selected cases, color retinal photography [Diabetic Retinopathy: American Academy of Ophthalmology Preferred Practice Patterns. San Francisco, Cal.: American Academy of Ophthalmology Quality of Care Committee Retinal Pane, American Academy of Ophthalmology, 1989]. Fluorescein angiography of the retina is routinely performed today, but it is invasive, unpleasant, and causes occasional deaths. Furthermore, the additional information obtained by fluorescein angiography does not help categorize patients into those who may benefit from immediate laser treatment and those who will not [Ferris (1993) (commentary) JAMA 269:1290-1].

[0035] The oxygen supply of the retina is provided by both the choroidal and retinal circulation. The choroid serves as the oxygen source for the photoreceptors in the avascular outer retina, whereas the retinal circulation plays a crucial role in maintaining the oxygen supply to the neural elements and nerve fibers in the inner retina. Because of the high oxygen needs of the retina, any alteration in circulation such as seen in diabetic retinopathy, hypertension, sickle cell disease, and vascular occlusive diseases results in functional impairment and extensive retinal tissue.

[0036] Noninvasive measurements of the oxygen saturation of blood in retinal vessels was first proposed by Hickham et al. [Hickham et al. (1963) Circulation 27:375] using a two-wavelength photographic technique (560 and 640 nm) for retinal vessels crossing the optic disk (the region where the optic nerve connects to the retina). A more advanced approach based on the three wavelength method of Pittman and Duling is presented in Delori (1988) Applied Optics 27:1113-1125.

[0037] The present invention is the first step towards showing the usefulness of spectral imaging in general and the SPECTRACUBE technology in particular, as a new tool for the analysis of the physiological state of various structures of the human ocular fundus and enhance the accuracy of diagnosis and prognosis of certain diseases which affect the eye.

[0038] The ability to collect data of physiological importance in a spatially organized way, to store them for later retrieval and to display them in an enhanced image mode for easy interpretation provides a new horizon in ophthalmology.

[0039] There is thus a widely recognized need for, and it would be highly advantageous to have methods of spectral bio-imaging of the eye which can be used for non-invasive early detection and diagnosis of eye diseases.

SUMMARY OF THE INVENTION

[0040] According to the present invention there is provided a method for spectral imaging of an eye tissue, which can be used for non-invasive early detection and diagnosis of eye associated diseases and for detection of spatial organization, distribution and quantification of cellular and tissue natural constituents, structures and organelles, tissue vitality, tissue metabolism, tissue viability, etc., using light reflection, scattering and emission, with high spatial and spectral resolutions.

[0041] According to further features in preferred embodiments of the invention described below, provided is a spectral bio-imaging method for enhancing spectral signatures of an eye tissue, the method comprising the steps of (a) providing an optical device for eye inspection being optically connected to a spectral imager; (b) illuminating the eye tissue with light via the iris, viewing the eye tissue through the optical device and spectral imager and obtaining a spectrum of light for each pixel of the eye tissue; and (c) attributing each of the pixels a color according to its spectral signature, thereby providing an image enhancing the spectral signatures of the eye tissue.

[0042] According to still further features in the described preferred embodiments the spectral imager is selected from the group consisting of a filters based spectral imager, a monochromator based spectral imager and an interferometer based spectral imager.

[0043] According to still further features in the described preferred embodiments step (b) includes (i) collecting incident light simultaneously from all pixels of the eye using collimating optics; (ii) passing the incident collimated light through an interferometer system having a number of elements, so that the light is first split into two coherent beams which travel in different directions inside the interferometer and then the two coherent beams recombine to interfere with each other to form an exiting light beam; (iii) passing the exiting light beam through a focusing optical system which focuses the exiting light beam on a detector having a two-dimensional array of detector elements; (iv) rotating or translating one or more of the elements of the interferometer system, so that an optical path difference between the two coherent beams generated by the interferometer system is scanned simultaneously for all the pixels; and (v) recording signals of each of the detector elements as function of time using a recording device to form a spectral cube of data.

[0044] According to still further features in the described preferred embodiments the optical device is selected from the group consisting of a fundus camera and a funduscope.

[0045] According to still further features in the described preferred embodiments the spectrum of light represents light selected from the group consisting of, light reflected from the eye tissue, light scattered from the eye tissue and light emitted from the eye tissue.

[0046] According to still further features in the described preferred embodiments the light emitted from the eye tissue is selected from the group consisting of administered probe fluorescence, administered probe induced fluorescence and auto-fluorescence.

[0047] According to still further features in the described preferred embodiments the light used for illuminating the eye tissue is selected from the group consisting of coherent light, white light, filtered light, ultraviolet light and a light having a small wavelength range.

[0048] According to still further features in the described preferred embodiments the two-dimensional array is selected from the group consisting of a video rate CCD, a cooled high dynamic range CCD, an intensified CCD and a time gated intensified CCD.

[0049] According to still further features in the described preferred embodiments the eye tissue is selected from the group consisting of eye retina, a retinal blood vessel an optic disk, an optic cup, eye macula, cornea, iris and choroidal layer, or any combination thereof.

[0050] According to still further features in the described preferred embodiments the eye tissue includes a blood vessel the method is for detecting and mapping the oxygenation level of hemoglobin along the blood vessel.

[0051] According to still further features in the described preferred embodiments step (c) is effected using a mathematical algorithm which computes a Red-Green- Blue color image using predefined wavelength ranges.

[0052] According to still further features in the described preferred embodiments the spectral signature and, as a result, the color is affected by a substance selected from the group consisting of hemoglobin, cytochromes, flavins, nicotinamide adenine dinucleotide, nicotinamide adenine dinucleotide phosphate, collagen, elastin and melanin.

[0053] According to still further features in the described preferred embodiments enhancing the spectral signatures of the eye tissue includes an enhancement selected from the group consisting of enhancing arteries, enhancing veins, enhancing hemoglobin concentration and enhancing hemoglobin oxygen saturation level.

[0054] According to still further features in the described preferred embodiments the method further comprising the step of correcting spatial and spectral information for movements of the eye tissue via a spatial registration and spectral correction procedures.

[0055] According to still further features in the described preferred embodiments provided is a method of evaluating a medical condition of a patient comprising the step of enhancing spectral signatures of an eye tissue of the patient by (a) providing an optical device for eye inspection being optically connected to a spectral imager; (b) illuminating the eye tissue of the patient with light via the iris, viewing the eye tissue through the optical device and spectral imager and obtaining a light spectrum for each pixel of the eye tissue; (c) attributing each of the pixels a color according to its spectral signature, thereby providing an image enhancing the spectral signatures of the eye tissue; and (d) using the image to evaluate the medical condition of the patient.

[0056] According to still further features in the described preferred embodiments the medical condition is selected from the group consisting of diabetic retinopathy, ischemia of the eye, glaucoma, macular degeneration, CMV eye infection, retinitis, choroidal ischemia, acute sectorial choroidal ischemia, ischemic optic neuropathy, and corneal and iris problems.

[0057] According to still further features in the described preferred embodiments provided is a display comprising an image presenting an eye tissue, wherein each pixel in the image has a color according to a spectral signature of a tissue element from which it is derived, thereby enhancing the spectral signatures of the eye tissue.

[0058] According to still further features in the described preferred embodiments provided is a spectral bio-imaging method for obtaining a spectrum of a region (corresponding to a pixel or few pixels in the image) of an eye tissue, the method comprising the steps of (a) providing an optical device for eye inspection being optically connected to a spectral imager; (b) illuminating the eye tissue with light via the iris, viewing the eye tissue through the optical device and spectral imager and obtaining a spectrum of light for each pixel of the eye tissue; and (c) displaying a spectrum (a spectrum of a single pixel or an average spectrum of several pixels) associated with the region of interest. Spectra of specific regions in the eye are known in the art, however using the above method enables a practitioner to precisely select a region of interest, such that the spectrum obtained is the spectrum of interest.

[0059] The present invention successfully addresses the shortcomings of the presently known configurations by providing an image of the eye which enhances spectral signatures of constituents thereof, characterized by high spatial and spectral resolutions.

BRIEF DESCRIPTION OF THE DRAWINGS

[0060] The invention herein described, by way of example only, with reference to the accompanying drawings, wherein:

[0061] FIG. 1 is a block diagram illustrating the main components of an imaging spectrometer constructed in accordance with U.S. Pat. No. 5,539,517 (prior art).

[0062] FIG. 2 illustrates a Sagnac interferometer, as used in an imaging spectrometer in accordance with U.S. Pat. No. 5,539,517 (prior art).

[0063] FIG. 3 shows a definition of pseudo-RGB (Red, Green and Blue) colors for emphasizing chosen spectral ranges. The intensity for each pseudo-color is calculated by integrating the area under the curve, after multiplying it by one of the curves.

[0064] FIG. 4 a is a spectral image of a human right eye acquired using the SPECTRACUBE system.

[0065] FIG. 4 b is a spectral image of the human right eye of FIG. 4a after spatial registration and spectral correction.

[0066] FIG. 5 a presents a portion of an interferogram function of a given pixel derived from the spectral image of FIG. 4a.

[0067] FIG. 5 b presents a portion of an interferogram function of the same pixel of FIG. 5a, which pixel is derived from the spectral image of FIG. 4b.

[0068] FIG. 6 a presents spectra of five adjacent pixels derived from the spectral image of FIG. 4a, the position of each pixel is indicated.

[0069] FIG. 6 b presents spectra of five adjacent pixels derived from the spectral image of FIG. 4b, the position of each pixel is indicated.

[0070] FIGS. 7 a -f present the operation of a fringes suppression algorithm.

[0071] FIGS. 8 a and 8b presents a spectral image of a healthy retina. Spectrally distinct regions are designated in FIG. 8b.

[0072] FIG. 9 presents plots of hemoglobin extinction coefficients from the literature.

[0073] FIG. 10 presents plots of inverted log of reflectivity spectra of a vein and an artery.

[0074] FIG. 11 presents spectra of pixels from the disk, the cup, the retina, and a retinal vessel, as measured according to the present invention.

[0075] FIG. 12 is a schematic cross section of the retina, demonstrating the reflection of different wavelengths from different retinal depths.

[0076] FIGS. 13 a -c compares plots of spectra extracted from several eye regions reported in the prior art (13a) with spectra measured according to the present invention of the same regions (13b) and of other regions (13c).

[0077] FIGS. 14 a -e present an RGB image, an enhanced RGB image, a 610 and 564 nm images and a hemoglobin oxygenation image of portion of a retina including retinal blood vessels of a healthy individual.

[0078] FIG. 15 presents plots of spectra derived from a hemorrhage and healthy retinal regions, according to the present invention.

[0079] FIG. 16 presents plots of inverted log reflectivity spectra of normal, intermediate and degenerate macular tissue of a single patient suffering macular degeneration, as measured according to the method of the present invention.

[0080] FIG. 17 presents an RGB image of a region in the macula of the patient of FIG. 16.

[0081] FIGS. 18 a -d present an RGB image, a 610 and 564 nm images and a hemoglobin concentration image of an optic disk of a healthy individual.

[0082] FIGS. 19 a -e present an RGB image, a 610 and 564 nm images, a hemoglobin concentration image and a key image of an optic disk of a glaucoma patient.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0083] The present invention is of a method for spectral bio-imaging of the eye which can be used for non-invasive early detection and diagnosis of eye diseases. Specifically the present invention can be used for detection of spatial organization, distribution and quantification of cellular and tissue natural constituents, structures and organelles, tissue vitality, tissue metabolism, tissue viability, etc., using light reflection, scattering and emission, with high spatial and spectral resolutions.

[0084] The principles and operation of a method according to the present invention may be better understood with reference to the drawings and accompanying descriptions.

[0085] The present invention is of a spectral bio-imaging method for enhancing spectral signatures of an eye tissue (e.g., ocular fundus tissue). The method includes the following steps.

[0086] First, an optical device for eye inspection, such as, but not limited to a funduscope or a fundus camera, which is optically connected to a spectral imager is provided.

[0087] Second, the eye tissue is illuminated with light via the iris, the eye tissue is viewed through the optical device and spectral imager and a light spectrum for each pixel of the eye tissue is obtained.

[0088] Third, each of the pixels is attributed a color according to its spectral signature, thereby an image enhancing the spectral signatures of the eye tissue is provided.

[0089] Any spectral imager may be used to perform the method of the present invention. A suitable spectral imager is, for example, a filters based spectral imager, a monochromator (grating/prism) based spectral imager and or an interferometer based spectral imager. A description concerning the operation and construction advantages and disadvantages of each of these spectral imagers is provided in the Background section above and the Examples section that follows.

[0090] According to a preferred embodiment of the invention the spectral imager includes an interferometer. In this case, step (b) above includes the following (i) collecting incident light simultaneously from all pixels of the eye using collimating optics; (ii) passing the incident collimated light through an interferometer system having a number of elements, so that the light is first split into two coherent beams which travel in different directions inside the interferometer and then the two coherent beams recombine to interfere with each other to form an exiting light beam; (iii) passing the exiting light beam through a focusing optical system which focuses the exiting light beam on a detector having a two-dimensional array of detector elements; (iv) rotating or translating one or more of the elements of the interferometer system, so that an optical path difference between the two coherent beams generated by the interferometer system is scanned simultaneously for all the pixels; and (v) recording signals of each of the detector elements as function of time using a recording device to form a spectral cube of data.

[0091] According to a preferred embodiment of the invention the two-dimensional array is selected from the group consisting of a video rate CCD, a cooled high dynamic range CCD, an intensified CCD and a time gated intensified CCD.

[0092] Be it an interferometer based spectral imager or any other spectral imager, the light analyzed to derive a spectrum of each of the pixels of the eye tissue may be light reflected from the eye tissue, light scattered from the eye tissue and/or light emitted from the eye tissue. The light emitted from the eye tissue may be due to administered probe fluorescence, administered probe induced fluorescence and/or auto-fluorescence of the eye tissue.

[0093] Depending on the specific application, the light used for illuminating the eye tissue is, for example, coherent light (e.g., laser), white light, filtered light, ultraviolet light and a light having a small wavelength range.

[0094] Any eye tissue is suitable for examination using the method of the present invention, including, but not limited to, eye retina, a retinal blood vessel, an optic disk, an optic cup, eye macula, cornea and choroidal layer. In many cases the eye tissue includes blood vessels and the method serves for detecting and mapping the oxygenation level and/or concentration of hemoglobin along any of the blood vessel, veins and/or arteries.

[0095] Effecting step (c) above may be accomplished in many ways, for example, using any of the algorithms described under Example 2 below. However, in a preferred embodiment step (c) is effected using a mathematical algorithm which computes a Red-Green-Blue color image using predefined wavelength ranges, all as further described in the Examples section.

[0096] In a prefered embodiment of the invention, the spectral signature of the eye tissue and, as a result, the color of each pixel is affected by a substance such as hemoglobin, cytochromes, flavins, nicotinamide adenine dinucleotide, nicotinamide adenine dinucleotide phosphate, collagen, elastin and/or melanin. The color of each pixel represents the content or concentration of any one or more of these materials or, except for collagen, elastin and melanin, the ratio between their oxidized (e.g., oxygenated, dehydrogenated) and reduced (e.g., hydrogenated, deoxygenated) forms.

[0097] According to the present invention enhancing the spectral signatures of the eye tissue may includes enhancement of physiological structures such as arteries and veins and/or levels of biological substances such as hemoglobin concentration and oxygen saturation level, which is indicative to the level of metabolism and/or vitality of the tissue.

[0098] According to another prefered embodiment of the invention the spectral imager employed includes an interferometer and a procedure for correcting spatial and spectral information for movements of the eye tissue via a spatial registration and spectral correction procedures is employed. Mechanically and/or chemically fixating the analyzed eye obviates this procedure.

[0099] As is evident from the examples below, the method according to the present invention can be used for evaluating a medical condition of a patient. The medical evaluation method includes steps (a)-(c), substantially as described above and further includes a medical evaluation procedure using the image obtained. The medical condition may be any condition that affects the eye, including, but not limited to, diabetic retinopathy, ischemia of the eye, glaucoma, macular degeneration, CMV eye infection (cytomegalovirus eye infection of AIDS patients) retinitis, choroidal ischemia, acute sectorial choroidal ischemia, ischemic optic neuropathy, and corneal and iris problems.

[0100] Further according to the present invention provided is a display which includes an image presenting an eye tissue, wherein each pixel in the image has a color according to a spectral signature of a tissue element (part of a tissue which is equivalent to a pixel in the image, depending on spatial resolution) from which it is derived, thereby enhancing the spectral signatures of the eye tissue. The term “display” as used herein refers to any visual presentation such as, but not limited to, a photograph, a print, screen display or a monitor display.

[0101] Still further according to the present invention provided is a spectral bio- imaging method for obtaining a spectrum of a region (corresponding to a pixel or few pixels in the image) of an eye tissue. The method includes the following steps.

[0102] First, an optical device for eye inspection, such as, but not limited to a funduscope or a fundus camera, which is optically connected to a spectral imager is provided.

[0103] Second, the eye tissue is illuminated with light via the iris, the eye tissue is viewed through the optical device and spectral imager and a light spectrum for each pixel of the eye tissue is obtained. 21

[0104] And third, a spectrum (a spectrum of a single pixel or an average spectrum of several pixels) associated with the region of interest is displayed.

[0105] Reference in now made to the following examples, which together with the above descriptions, illustrate the invention.

EXAMPLE 1

The Measurement Apparatus

[0106] FIG. 1 is a block diagram illustrating the main components of a prior art imaging spectrometer disclosed in U.S. Pat. No. 5,539,517. This imaging spectrometer is constructed highly suitable to implement the method of the present invention as it has high spectral (Ca. 4-14 nm depending on wavelength) and spatial (Ca. 30/M μm where M is the effective microscope or fore optics magnification) resolutions.

[0107] Thus, the prior art imaging spectrometer of FIG. 1 includes: a collection optical system, generally designated 20; a one-dimensional scanner, as indicated by block 22; an optical path difference (OPD) generator or interferometer, as indicated by block 24; a one-dimensional or two-dimensional detector array, as indicated by block 26; and a signal processor and display, as indicated by block 28.

[0108] A critical element in system 20 is the OPD generator or interferometer 24, which outputs modulated light corresponding to a predetermined set of linear combinations of the spectral intensity of the light emitted from each pixel of the scene to be analyzed. The output of the interferometer is focused onto the detector array 26.

[0109] Thus, all the required optical phase differences are scanned simultaneously for all the pixels of the field of view, in order to obtain all the information required to reconstruct the spectrum. The spectra of all the pixels in the scene are thus collected simultaneously with the imaging information, thereby permitting analysis of the image in a real-time manner.

[0110] The apparatus according to U.S. Pat. No. 5,539,517 may be practiced in a large variety of configurations. Specifically, the interferometer used may be combined with other mirrors as described in the relevant Figures of U.S. Pat. No. 5,539,517.

[0111] Thus, according to U.S. Pat. No. 5,539,517 alternative types of interferometers may be employed. These include (i) a moving type interferometer in which the OPD is varied to modulate the light, namely, a Fabry-Perot interferometer with scanned thickness; (ii) a Michelson type interferometer which includes a beamsplitter receiving the beam from an optical collection system and a scanner, and splitting the beam into two paths; (iii) a Sagnac interferometer optionally combined with other optical means in which interferometer the OPD varies with the angle of incidence of the incoming radiation, such as the four-mirror plus beamsplitter interferometer as further described in the cited U.S. patent (see FIG. 14 there).

[0112] FIG. 2 illustrates an imaging spectrometer constructed in accordance with U.S. Pat. No. 5,539,517, utilizing an interferometer in which the OPD varies with the angle of incidence of the incoming radiation. A beam entering the interferometer at a small angle to the optical axis undergoes an OPD which varies substantially linearly with this angle.

[0113] In the interferometer of FIG. 2, all the radiation from source 30 in all the pixels, after being collimated by an optical collection system 31, is scanned by a mechanical scanner 32. The light is then passed through a beamsplitter 33 to a first reflector 34 and then to a second reflector 35, which reflects the light back through the beamsplitter 33 and then through a focusing lens 36 to an array of detectors 37 (e.g., a CCD). This beam interferes with the beam which is reflected by 33, then by second reflector 35, and finally by first reflector 34.

[0114] At the end of one scan, every pixel has been measured through all the OPD's, and therefore the spectrum of each pixel of the scene can be reconstructed by Fourier transformation. A beam parallel to the optical axis is compensated, and a beam at an angle (θ) to the optical axis undergoes an OPD which is a function of the thickness of the beamsplitter 33, its index of refraction, and the angle θ. The OPD is proportional to θ for small angles. By applying the appropriate inversion, and by careful bookkeeping, the spectrum of every pixel is calculated.

[0115] In the configuration of FIG. 2 the ray which is incident on the beamsplitter at an angle β (β=45° in FIG. 2) goes through the interferometer with an OPD=0, whereas a ray which is incident at a general angle β-θ undergoes an OPD given by the following:

OPD (β,θ,t,n)=t[(n2−sin2(β+θ))0.5−(n2−sin2(β−θ))0.5+2 sin β sin θ  (1)

[0116] where β is the angle of incidence of the ray on the beamsplitter; θ is the angular distance of a ray from the optical axis or interferometer rotation angle with respect to the central position; t is the thickness of the beamsplitter; and n is the index of refraction of the beamsplitter.

[0117] It follows from Equation 1 that by scanning both positive and negative angles with respect to the central position, one gets a double-sided interferogram for every pixel, which helps eliminate phase errors giving more accurate results in the Fourier transform calculation. The scanning amplitude determines the maximum OPD reached, which is related to the spectral resolution of the measurement. The size of the angular steps determines the OPD step which is, in turn, dictated by the shortest wavelength to which the system is sensitive. In fact, according to the sampling theorem [see, Chamberlain (1979) The principles of interferometric spectroscopy, John Wiley and Sons, pp. 53-55], this OPD step must be smaller than half the shortest wavelength to which the system is sensitive.

[0118] Another parameter which should be taken into account is the finite size of a detector element in the matrix. Through the focusing optics, the element subtends a finite OPD in the interferometer which has the effect of convolving the interferogram with a rectangular function. This brings about, as a consequence, a reduction of system sensitivity at short wavelengths, which drops to zero for wavelengths equal to or below the OPD subtended by the element. For this reason, one must ensure that the modulation transfer function (MTF) condition is satisfied, i.e., that the OPD subtended by a detector element in the interferometer must be smaller than the shortest wavelength at which the instrument is sensitive.

[0119] Thus, imaging spectrometers constructed in accordance with the invention disclosed in U.S. Pat. No. 5,539,517 do not merely measure the intensity of light coming from every pixel in the field of view, but also measure the spectrum of each pixel in a predefined wavelength range. They also better utilize all the radiation emitted by each pixel in the field of view at any given time, and therefore permit, as explained above, a significant decrease in the frame time and/or a significant increase in the sensitivity of the spectrometer. Such imaging spectrometers may include various types of interferometers and optical collection and focusing systems, and may therefore be used in a wide variety of applications, including medical diagnostic and therapy and biological research applications, as well as remote sensing for geological and agricultural investigations, and the like.

[0120] As mentioned above, an imaging spectrometer in accordance with the invention disclosed in U.S. Pat. No. 5,539,517 was developed by Applied Spectral Imaging Ltd., Industrial Park, Migdal Haemek, Israel and is referred herein as SPECTRACUBE.

[0121] The SPECTRACUBE system optically connected to a microscope is used to implement the method for chromosome classification of the present invention. The SPECTRACUBE system has the following or better characteristics, listed hereinbelow in Table 1 below.

[0122] The prior art SPECTRACUBE system was used, in accordance with the present invention, to acquire spatially organized spectral data from the eye. However, it will be appreciated that any spectral imager, i.e., an instrument that measures and stores in memory for later retrieval and analysis the spectrum of light emitted by every point of an object which is placed in its field of view, including filter (e.g., acousto-optic tunable filters (AOTF) or liquid-crystal tunable filter (LCTF)) and dispersive element (e.g., grating or prism) based spectral imagers, or other spectral data or multi-band collection devices (e.g., a device in accordance with the disclosure in Speicher R. M., Ballard S. G. and Ward C. D. (1996) Karyotyping human chromosomes by combinatorial multi-flour FISH. Nature genetics, 12:368-375) can be used to acquire the required spectral data. Therefore, it is intended not to limit the scope of the present invention for use of any specific type of spectral data collection devices, nor any specific type of spectral imager.

TABLE 1
Parameter            Performance
Spatial resolution:  30/M μm (M = effective microscope or fore
                     optics magnification)
Field of View:       15/M millimeters
Sensitivity:         20 milliLux (for 100 msec integration time,
                     increases for longer integration times linearly
                     with {square root}T)
Spectral range:      400-1000 nm
Spectral resolution: 4 nm at 400 nm (16 nm at 800 nm)
Acquisition time:    5-50 sec, typical 25 sec
FFT processing time: 20-180 sec, typical 60 sec

[0123] As mentioned above, the SPECTRACUBE system easily attaches to any microscope or macro lens with, for example, C-mount or F-mount connectors, and can stand in any orientation during the measurement. The system may as well be connected to other magnification means and to various types of endoscopes and cameras including funduscopes and fundus cameras. Therefore, spectral images of the eye tissue in various magnification and lighting may be obtained.

[0124] To conduct the present study the SPECTRACUBE system was mounted on the CCD port of a fundus camera (Zeiss Model RC-3 10) and the combined system was situated such that the optical path was substantially horizontal. This facilitates eye inspection, wherein the patient is seated. A white light source was used for illumination of the eye and reflected light was collected and analyzed.

EXAMPLE 2

Display and Analysis of Spectral Images

[0125] a. General

[0126] As mentioned above, a spectral image is a three dimensional array of data, I(x,y,λ), that combines spectral information with spatial organization of the image.

[0127] As such, a spectral image is a set of data called a spectral cube, due to its dimensionality, which enables the extraction of features and the evaluation of quantities that are difficult, and in some cases even impossible, to obtain otherwise.

[0128] Since both spectroscopy and digital image analysis are well known fields that are covered by an enormous amount of literature [see, for example, Jain (1989) Fundamentals of Digital Image Processing, Prentice-Hall International], the following discussion will focus primarily on the benefit of combining spectroscopic and imaging information in a single data set i.e., a spectral cube.

[0129] One possible type of analysis of a spectral cube is to use spectral and spatial data separately, i.e., to apply spectral algorithms to the spectral data and two- dimensional image processing algorithms to the spatial data.

[0130] As an example for a spectral algorithm consider an algorithm computing the similarity between a reference spectrum and the spectra of all pixels (i.e., similarity mapping) resulting in a gray (or other color) scale image (i.e., a similarity map) in which the intensity at each pixel is proportional to the degree of ‘similarity’.

[0131] This gray scale image can then be further analyzed using image processing and computer vision techniques (e.g., image enhancement, pattern recognition, etc.) to extract the desired features and parameters.

[0132] In other words, similarity mapping involves computing the integral of the absolute value of the difference between the spectrum of each pixel of the spectral image with respect to a reference spectrum (either previously memorized in a library, or belonging to a pixel of the same or other spectral image), and displaying a gray level or pseudocolor (black and white or color) image, in which the bright pixels correspond to a small spectral difference, and dark pixels correspond to a large spectral difference, or vice versa.

[0133] Similarly, classification mapping perform the same calculation as described for similarity mapping, yet takes several spectra as reference spectra, and paints each pixel of the displayed image with a different predetermined pseudocolor, according to its classification as being most similar to one of the several reference spectra.

[0134] It is also possible to apply spectral image algorithms based on non-separable operations; i.e., algorithms that include both local spectral information and spatial correlation between adjacent pixels (one of these algorithms is, as will be seen below, a principal component analysis).

[0135] One of the basic needs that arise naturally when dealing with any three-dimensional (3D) data structure such as a spectral cube (i.e., I(xy,λ)), is visualizing that data structure in a meaningful way. Unlike other types of 3D data such as topographic data, D(x,y,z), obtained, for example, by a confocal microscope, where each point represents, in general, the intensity at a different locations (x,y,z) in tree-dimensional space, a spectral image is a sequence of images representing the intensity of the same two-dimensional plane (i.e., the sample) at different wavelengths. For this reason, the two most intuitive ways to view a spectral cube of data is to either view the image plane (spatial data) or the intensity of one pixel or a set of pixels as function of wavelength in a three-dimensional mountain-valley display. In general, the image plane can be used for displaying either the intensity measured at any single wavelength or the gray scale image that results after applying a spectral analysis algorithm, over a desired spectral region, at every image pixel. The spectral axis can, in general, be used to present the resultant spectrum of some spatial operation performed in the vicinity of any desired pixel (e.g., averaging the spectrum).

[0136] It is possible, for example, to display the spectral image as a gray scale image, similar to the image that might be obtained from a simple monochrome camera, or as a multicolor image utilizing one or several artificial colors to highlight and map important features. Since such a camera simply integrates the optical signal over the spectral range (e.g., 400 nm to 760 nm) of the CCD array, the ‘equivalent’ monochrome CCD camera image can be computed from the 3D spectral image data base by integrating along the spectral axis, as follows: 1$\begin{array}{cc}\mathrm{gray\_scale}\left(x,y\right)={\int }_{\lambda 2}^{\mathrm{\lambda 1}}w\left(\lambda \right)·I\left(x,y,\lambda \right)d\lambda & \left(2\right)\end{array}$

[0137] In equation 2, w(λ) is a general weighting response function that provides maximum flexibility in computing a variety of gray scale images, all based on the integration of an appropriately weighted spectral image over some spectral range. For example, by evaluating equation (2) with three different weighting functions, {wr(λ), wg(λ), wb(λ)}, corresponding to the tristimulus response functions for red (R), green (G) and blue (B), respectively, it is possible to display a conventional RGB color image. It is also possible to display meaningful non-conventional color images, wherein the weighting functions differ from RGB. FIG. 3 presents an example of the power of this simple algorithm. Consider choosing {wr, wg, wb} to be Gaussian functions distributed “inside” a spectrum of interest, the resulting pseudo-color image that is displayed in this case emphasizes only data in the spectral regions corresponding to the weighting functions, enabling spectral differences in these three regions to be detected more clearly.

[0138] b. Point Operations

[0139] Point operations are defined as those that are performed on single pixels, (i.e., do not involve more than one pixel at a time). For example, in a gray scale image, a point operation can be one that maps the intensity of each pixel (intensity function) into another intensity according to a predetermined transformation function. A particular case of this type of transformation is the multiplication of the intensity of each pixel by a constant.

[0140] The concept of point operations can also be extended to spectral images: here each pixel has its own intensity function (spectrum), i.e., an n-dimensional vector V1(λ); λε[λ1, λn]. A point operation applied to a spectral image can be defined as one that maps the spectrum of each pixel into a scalar (i.e., an intensity value) according to a transformation function:

v 2 =g (V1(λ)); λε[λ1, λn]  (3)

[0141] Building a gray scale image according to Equation 3 is an example of this type of point operation. In the more general case, a point operation maps the spectrum (vector) of each pixel into another vector according to a transformation function:

V 2(l)=g(V1(λ)); lε[1, N], λε[λ1, λn]  (4), where N≦n.

[0142] In this case a spectral image is transformed into another spectral image.

[0143] One can now extend the definition of point operations to include operations between corresponding pixels of different spectral images. An important example of this type of algorithm is optical density analysis. Optical density is employed to highlight and graphically represent regions of an object being studied spectroscopically with higher dynamic range than the transmission spectrum. The optical density is related to transmission by a logarithmic operation and is therefore always a positive function. The relation between the optical density and the measured spectra is given by Lambert Beer law: 2$\begin{array}{cc}\mathrm{OD}\left(\lambda \right)=-{\log }_{10}\frac{I\left(\lambda \right)}{I_0\left(\lambda \right)}=-{\log }_{10}\tau \left(\lambda \right)& \left(5\right)\end{array}$

[0144] where OD(λ) is the optical density as a function of wavelength, I(λ) is the measured spectrum, IO(λ) is a measured reference spectrum, and τ(λ) is the spectral transmitance of the sample. Equation 5 is calculated for every pixel for every wavelength where IO(λ) is selected from (i) a pixel in the same spectral cube for which OD is calculated; (ii) a corresponding pixel in a second cube; and (iii) a spectrum from a library.

[0145] Note that the optical density does not depend on either the spectral response of the measuring system or the non-uniformity of the CCD detector. This algorithm is useful to map the relative concentration, and in some cases the absolute concentration of absorbers in a sample, when their absorption coefficients and the sample thickness are known. It should thus be noted that the term ‘level’ as used hereinbelow in the claims section also refers to the terms ‘amount’, ‘relative amount’, ‘absolute concentration’ and ‘relative concentration’.

[0146] Additional examples include various linear combination analyses, such as for example: (i) applying a given spectrum to the spectrum of each of the pixels in a spectral image by an arithmetical function such as addition, subtraction, multiplication, division and combinations thereof to yield a new spectral cube, in which the resulting spectrum of each pixel is the sum, difference, product ratio or combination between each spectrum of the first cube and the selected spectrum; and (ii) applying a given scalar to the spectra of each of the pixels of the spectral image by an arithmetical function as described above.

[0147] Such linear combinations may be used, for example, for background subtraction in which a spectrum of a pixel or, preferably, the average spectrum of some or all of the pixels located in the background region is subtracted from the spectrum of each of the other (non-background) pixels; and for a calibration procedure in which a spectrum measured prior to sample analysis is used to divide the spectrum of each of the pixels in the spectral image.

[0148] Another example includes a ratio image computation and display as a gray level image. This algorithm computes the ratio between the intensities at two different wavelengths for every pixel of the spectral image and paints each of the pixels in a lighter or darker artificial color accordingly. For example, it paints the pixel bright for high ratio, and dark for low ratio (or the opposite), to display distributions of spectrally sensitive materials.

[0149] c. Spatial-spectral Combined Operations

[0150] In all of the spectral image analysis methods mentioned above, algorithms are applied solely to the spectral data. The importance of displaying the spectrally processed data as an image is mostly qualitative, providing the user with a useful image. It is also possible, however, depending on the application, to use the available imaging data in even more meaningful ways by applying algorithms that utilize the spatial-spectral correlation that is inherent in a spectral image. Spatial- spectral operations represent the most powerful types of spectral image analysis algorithms. As an example, consider the following situation:

[0151] A sample contains k cell types stained with k different fluorophores (the term ‘cell’ here is used both for a biological cell, and also as ‘a region in the field of view of the instrument’). Each fluorophore has a distinct fluorescence emission spectrum and binds to only one of the k cell types. It is important to find the average fluorescence intensity per cell for each one of the k cell types. To achieve this task the following procedure can be used: (i) each pixel in the image is classified as belonging to one of k+1 classes (k cell types plus a background) according to its spectrum; (ii) the image is segmented into the various cell types and the number of cells from each type is counted; and (iii) the fluorescence energy contributed by each class is summed and divided by the total number of cells from the corresponding class.

[0152] This procedure makes use of both spectral and spatial data. The relevant spectral data takes the form of characteristic cell spectra (i.e., spectral “signatures”), while the spatial data consists of data about various types of cells (i.e., cell blobs) many of which appear similar to the eye. The ideal type of measurement for this type of situation is a spectral image. In the above situation, cells can be differentiated by their characteristic spectral signature. Hence, a suitable point operation will be performed to generate a synthetic image in which each pixel is assigned one of k+1] values. Assuming that the fluorescence emission spectra of the different cell types are known to be si(λ); i1, 2, . . . , k, λε[λ1, λn], and the measured spectrum at each pixel (x, y) is sx,y(λ), λε[λ1, λn], then the following algorithm is a possible method of classification (step 1 above):

[0153] Let e2i be the deviation of the measured spectrum from the known spectrum of the fluorophore attached to cell type i. Then, adopting a least-squares “distance” definition, one can write: 3$\begin{array}{cc}{e}_i^2=\sum _{\lambda \in R_{\lambda }}{\left(s\left(\lambda \right)-s_i\left(\lambda \right)\right)}^2& \left(6\right)\end{array}$

[0154] where Rλ is the spectral region of interest. Each point [pixel (x, y)] in the image can then be classified into one of the k+1 classes using the following criterion: 4$\begin{array}{cc}\mathrm{point}\left(x,y\right)\in \mathrm{class}k+1\mathrm{if}{e}_i^2>\mathrm{threshold}\mathrm{for}\mathrm{all}i\in \left[1,k\right],\mathrm{whereas}\mathrm{point}\left(x,y\right)\in \mathrm{class}\rho \mathrm{if}{e}_i^2<\mathrm{threshold},\mathrm{and}\rho \mathrm{is}\mathrm{such}\mathrm{that}\min \left[e_i^2\right]=e_p^2& \left(7\right)\end{array}$

[0155] Steps ii and iii above (image segmentation and calculation of average fluorescence intensity) are now straightforward using standard computer vision operations on the synthetic image created in accordance with the algorithm described in equations 6 and 7.

[0156] Another approach is to express the measured spectrum sx,y(λ) at each pixel as a linear combination of the k known fluorescence spectra si(A); i=1, 2, . . . , k. In this case one would find the coefficient vector C=[c1, c2, . . . , ck] that solves: 5$\begin{array}{cc}F=\min \sum _{\lambda \in R_{\lambda }}{\left(s\left(\lambda \right)-\hat{s}\left(\lambda \right)\right)}^2\mathrm{where}\hat{s}\left(\lambda \right)=\sum _{i=1}^k{c}_i·s_i\left(\lambda \right),& \left(8\right)\end{array}$

[0157] Solving for 6$\frac{dF}{d{c}_i}=0;\mathrm{for}$

[0158] i=1,2, . . . , k (i.e., find values of ci which minimize F) yields the matrix equation C=A−1B (9), where A is a square matrix of dimension k with elements 7$\begin{array}{cc}{a}_{m,n}=\left[\sum _{\lambda \in R_{\lambda }}{s}_m\left(\lambda \right)·s_n\left(\lambda \right)\right],& \left(10\right)\end{array}$

[0159] and B is a vector defined as 8$\begin{array}{cc}{b}_m=\left[\sum _{\lambda \in R_{\lambda }}{s}_m\left(\lambda \right)·s\left(\lambda \right)\right],m,n=1,2\dots ,k.& \left(11\right)\end{array}$

[0160] Arithmetic operations may similarly be applied to two or more spectral cubes and/or spectra of given pixels or from a library. For example consider applying an arithmetic operations between corresponding wavelengths of corresponding pairs of pixels belonging to a first spectral cube of data and a second spectral cube of data to obtain a resulting third spectral cube of data for the purpose of, for example, averaging two spectral cubes of data, time changes follow-up, spectral normalization, etc.

[0161] In many cases objects present in a spectral image differ from one another in chemical constituents and/or structure to some degree. Using a principal component analysis by producing covariance or correlation matrices enhances these small differences.

[0162] A brief description of the principal component analysis using a covariance matrix is given below. For further details regarding the principal component analysis, the reader is referred to Martens and Naes (1989) Multivariate Calibration, John Wiley & Sons, Great Britain; and to Esbensen et al., Eds. (1994) Multi variance analysis - in practice. Computer-aided modeling as CAMO, and the Unscrambler's User's guide, Trondheim, Norway.

[0163] Thus, the intensities of the pixels of the image at wavelength λi (i=1, . . . N) are now considered a vector whose length is equal to the number of pixels q. Since there are N of these vectors, one for every wavelength of the measurement, these vectors can be arranged in a matrix B′ with q rows, and N columns: 9$\begin{array}{cc}{B}^{\prime }=\mathrm{No}.\mathrm{of}\mathrm{pixels}\begin{array}{c}\mathrm{No}.\mathrm{of}\mathrm{wavelengths}\\ \begin{array}{ccc}{B}_{11}^{\prime }& \cdots & B_{1N}^{\prime }\\ ⋮& & ⋮\\ B_{\mathrm{q1}}^{\prime }& \cdots & B_{\mathrm{qN}}^{\prime }\end{array}\end{array}& \left(12\right)\end{array}$

[0164] For each of the columns of matrix B′ defined is an average: 10$\begin{array}{cc}{M}_i=\frac{1}{q}\sum _{i=1}^q{B}_{\mathrm{ji}}^{\prime };i=1\dots N& \left(13\right)\end{array}$

[0165] and a second normalized matrix B defined as: 11$\begin{array}{cc}B=\mathrm{No}.\mathrm{of}\mathrm{pixels}\begin{array}{c}\mathrm{No}.\mathrm{of}\mathrm{wavelengths}\\ \begin{array}{ccc}\frac{B_{11}^{\prime }}{M_1}& \cdots & \frac{B_{1N}^{\prime }}{M_N}\\ ⋮& & ⋮\\ \frac{B_{\mathrm{q1}}^{\prime }}{M_1}& \cdots & \frac{B_{\mathrm{qN}}^{\prime }}{M_N}\end{array}\end{array}& \left(14\right)\end{array}$

[0166] A covariance matrix C is defined for the matrix B: C=BT·B of dimensions N×N. C is diagonalyzed, and eigenvectors and eigenvalues related by: C·Vi=μi·Vi where Vi are N orthogonal unit vectors and μi are the eigenvalues representing the variance in the direction of the i-th unit vector Vi. In general, the lowest components represent the highest variability as a function of pixels.

[0167] The products BVi(i=1, . . . N) are the projections of the spectral image onto the elements of the orthogonal basis, and can be displayed separately as black and white images. These images may reveal features not obvious from a regular black and white image filtered at a certain wavelength.

EXAMPLE 4

Spectral Imaging of Moving Objects

[0168] According to the present invention provided are spectral images of the eye collected preferably by an interferometer based spectral imager.

[0169] Since, in order to perform a measurement, an interferometer based spectral imager must collect several frames of an examined object in a period of time that varies from ca. 5 to 60 seconds, a considerably longer period of time as compared with a camera or video camera snapshot, spectral imaging of moving objects, like the eye results in blurring of the image of the object and in disrupting the algorithm used to calculate the spectrum of each pixel thereof.

[0170] Indeed, while using the apparatus disclosed in U.S. Pat. No. 5,539,517 one This is indeed the case in many applications, such as when spectral imaging is used for color karyotyping and color banding of chromosomes as disclosed in Schroeck et al. (1996) Multicolor spectral karyotyping of human chromosomes. Science 273:494-497. However, in other applications spectral imaging of a moving object is required. This is the case for example when the examined object is an organ of a living creature (e.g., a human eye or a specific region or tissue thereof).

[0171] Any attempt to measure a spectral image of a living organ, which organ is not motionless, will result in artifacts and a distorted or particularly noisy spectral image data. If such an image is acquired using filter or grating based spectral imagers, a spatial image registration procedure will be required for best results. Nevertheless, these spectral imagers suffer limitations as described in the background section and are therefore less prefered.

[0172] On the other hand, should such an image be acquired by an interferometer based spectral imager which have numerous advantages over other spectral imaging systems, not only spatial registration but also spectral correction is required.

[0173] PCT/US97/08153, filed May 12, 1997, which is incorporated by reference as if fully set forth herein, teaches spatial registration and spectral correction for interferometer based spectral imaging.

[0174] U.S. Pat. No. 5,539,517 and other publications [e.g., (i) Schroeck et al. (1996) Multicolor spectral karyotyping of human chromosomes. Science 273:494-497; (ii) Malik et al. (1996) Fourier transform multipixel spectroscopy for quantitative cytology. J. of Microscopy 182:133-140; (iii) Malik and Dishi (1995) ALA mediated PDT of melanoma tumors: light-sensitizer interactions determined by a novel spectral imaging system. Proceedings of optical methods for tumor treatment and detection: Mechanisms and techniques in photodynamic therapy IV, Feb. 4-5, 1995, San Jose, Calif., SPIE Vol. 2392, pp. 152-158; (iii) Malik et al. (1994) A novel spectral imaging system combining spectroscopy with imaging - application for biology. Proceedings of optical and imaging techniques in biomedicine, Sep. 8-9, 1994, Lille, France, SPIE Vol. 2329, pp. 180-184; (iv) Malik et al. (1996) Fourier transform multiplex spectroscopy and spectral imaging of photoporphyrin in single melanoma cells. Photochemistry and photobiology 63:608-614; and (v) Soenksen et al. (1996) Use of novel bio-imaging system as an imaging oximeter in intact rat brain. Proceedings of advances in laser and light spectroscopy to diagnose cancer and other diseases III, Jan. 29-30, 1996, San Jose Calif., SPIE Vol. 2679, pp. 182-189] teach spectral imaging devices and methods, in which the light from a surface of an examined object is collected by an optical aperture or field lens, passed through an interferometer, in which it is split into two coherent rays, and then it is focused by focusing optics onto a two-dimensional detector array device (e.g. a CCD in the UV to visible range) having a surface of detector elements, such that the detector's surface represents a real image of the object's surface.

[0175] The signals from each and all detector elements of the detector array, as obtained from many successive frames of the detector array, are recorded, while the interferometer is scanned in synchronization with the detector frames.

[0176] Since at each position of the interferometer, the optical path difference (OPD) between the two split beams through which a detector element sees its corresponding picture element (pixel) varies in a known way, at the end of the scan, the signals collected for each pixel form a function called interferogram, which is the intensity of light as function of the optical path difference (OPD) for that particular pixel. Because the interferometer speed is constant, the CCD frame time is constant and the OPD is proportional to the interferometer angular position, the OPD samples are equally spaced.

[0177] According to the well known teachings of the Fourier transform spectroscopy, the mathematical Fourier transform operation applied to this interferogram function yields a spectrum, i.e., the intensity of light in every wavelength emitted by the pixel in question.

[0178] Since interferogram functions are known for every pixel of the object's surface, spectra can be calculated and known for every pixel thereof, by applying the Fourier transformation to all of the interferograms thus collected.

[0179] U.S. Pat. No. 5,539,517 teach several embodiments of spectral imaging devices and methods, each is capable of measuring a spectral image of an object, which devices differ from one another in the type of interferometer used therein.

[0180] It is well known that, in general, no matter what interferometer is used, at any one position of the interferometer scan, the OPD is different for an on-axis and an off-axis ray, and as a consequence, the OPD differs from pixel to pixel in the same frame.

[0181] For example, as explained in “The principles of interferometric spectroscopy” by John Chamberlain, John Wiley & Sons, 1979, page 220, Equations 8.3 and 8.4b, in a Michelson interferometer the OPD varies according to the following Equation: 12$\begin{array}{cc}\mathrm{OPD}=\frac{\lambda /2}{\left(1-\cos \alpha \right)}& \left(15\right)\end{array}$

[0182] where λ is the wavelength of light, and α is the angle between the on-axis and the off-axis rays.

[0183] According to Equation 15, the OPD dependence on the specific pixel is relatively low. In fact, in Equation 2, a is a small angle, and therefore the term (1-cosa) varies slowly as a2.

[0184] However, in a triangular interferometer such as that shown in FIG. 2, the OPD varies faster, i.e., linearly with the projection of the angle of incidence of the ray in the horizontal direction (equivalent to the projection of the distance of the corresponding pixel from the center of the image in the horizontal direction) as shown in Equation 31 in column 13 of U.S. Pat. No. 5,539,517.

[0185] This fact has two important consequences for an interferometer based spectral imager.

[0186] First, one has to keep track of the OPD for every pixel and every detector frame, so that at the end of the scan, it is possible to reconstruct the spectrum through the Fourier Transform algorithm. This is done by knowing (i) the

[0187] Second, should the examined object move during the measurement, the spatial registration of the various frames is lost, and the actual OPD of each pixel in each frame is different than it would have been should the object be still. Thus, if a spectral image of that object is calculated while neglecting its movements during the measurement, and the object is displayed using the collected data, for example via a Red-Green-Blue (RGB) function defined over some or all the spectral range, then (i) the image will look blurred due to loss of spatial registration during the measurement, and (ii) the calculated spectra will not represent the actual spectra, these spectra will look very noisy and not consistent due to the use of incorrect (i.e., non-registered) OPDs in the Fourier transformation.

[0188] Before turning to the description of the method for spatial registration and spectral correction for interferometer based spectral imaging which can be used to obtain spectral images of moving objects, the prior art method for measurement of a stationary object will be described.

[0189] Thus, a measurement of a stationary objects include the following steps.

[0190] First, the spectral imaging device is aligned and focused with respect to the examined object.

[0191] Second, the interferometer is scanned in equally spaced OPD steps, while acquiring and storing successive frames of the object by the CCD.

[0192] Third, the data is ordered (e.g., by a software) into an interferogram function for every pixel of the object's image.

[0193] Fourth, preferably some well known preprocessing steps called windowing or apodization (see, Chamberlain (1979) The principles of interferometric spectroscopy, John Wiley and Sons, pp. 131 and following pages) are performed, in order to regularize the data such that the data of the measurement, which is a discrete and finite set of data, can be used instead of a theoretical continuous interferogram function.

[0194] Fifth, “zero filling” procedure is typically performed, such that the number of data for each interferogram is completed to a number of points which equals a power of two of the original number of data, in order to fill-in the spectrum with more interpolated points and to use fast Fourier transform algorithms (see, Chamberlain (1979) The principles of interferometric spectroscopy, John Wiley and Sons, pp. 311 and following pages).

[0195] Sixth, the complex (real and imaginary parts) Fourier transforms are calculated by applying the fast Fourier transform algorithm on each of the interferograms. Alternatively, yet less preferably, a straight Fourier transform algorithm is applied. In the latter case “zero filling” is not required.

[0196] Seventh, the spectrum of every pixel is calculated as the module (length) of the complex function so obtained, a function defined on discrete values of a conjugate parameter to the OPD, the wavenumber σ, which in turn is the reciprocal of the wavelength: σ=1λ.

[0197] The fast Fourier transform algorithm reduces very considerably the calculation time but it can be used only when the OPD's are equally spaced and when the number of points in which the interferogram is defined equals to a power of two. For this reason the straightforward Fourier transform algorithm is generally not used.

[0198] The method for spatial registration and spectral correction for interferometer based spectral imaging which can be used to obtain spectral images of moving objects is described hereinafter.

[0199] The following description concerns an object that moves rigidly and linearly on a plane substantially perpendicular to the line of sight of the imager in a random or non-random movement. In other words the object moves in such a way that all of its parts keep their shape and size, and their relative distances, as seen through the spectral imager.

[0200] Thus, in the case of a rigidly moving object in random directions without changing plane, (i.e., without getting closer or farther from the instrument, so that the object remains in focus), the measurement steps according to the method of the present invention are as follows.

[0201] First, the spectral imaging device is aligned and focused with respect to the examined object.

[0202] Second, the interferometer is scanned in synchronization with the CCD frames and constant speed, while acquiring and storing successive frames of the object by the CCD. However, contrary to the above prior art description, due to the object's movements, the resulting OPD steps are inherently not equally spaced as described above. The difference between successive OPD's is now random: it is the result of the combined motion of the interferometer and of the object, it can increase or decrease depending on the instantaneous position and velocity of the object with respect to the position and velocity of the interferometer, it can even be negative (meaning decreasing OPD from a data point to the next) and, if the movement is larger than the field of view, or the movement is a sudden displacement larger than the field of view with immediate return to the previous position, a data point can be missing altogether. In some regions of the OPD axes the data points will be dense, in other they will be sparse.

[0203] Third, the data is ordered (e.g., by a software) into an interferogram function for every pixel of the image. However, now the book-keeping is more complicated. In order to accomplish this step one must first find the spatial translation vector of all the frames measured, with respect to a frame taken as reference. This way the actual OPD for every pixel in each frame can be found. Since this is a crucial step of the method according to the present invention it is described in more detail hereinbelow.

[0204] Fourth, preferably some well known preprocessing steps called windowing or apodization are performed, in order to regularize the data such that the data of the measurement which is a discrete data can be used instead of a theoretical continuous interferogram function.

[0205] Fifth, here the method splits into two alternative branches. According to the first, the measured interferogram of each pixel is not further interpolated and will be used with a straightforward Fourier transform algorithm to calculate its corresponding Fourier transform, whereas, according to the second, the measured interferogram of each pixel is interpolated to achieve OPD values which are equally spaced, and will be used with a fast Fourier transform algorithm to calculate its Fourier transform. Each alternative has advantages and disadvantages. Speed is higher in the latter but, as interpolation introduces errors, reliability of the data is higher in the former.

[0206] Sixth, a complex (real and imaginary) Fourier transform for each pixel is calculated by applying the straightforward or fast Fourier transform algorithms to each of the interferograms, depending on alternative choice made under the fifth step above.

[0207] Seventh, the spectrum of every pixel is calculated as the module (length) of the complex function so obtained, a function defined on discrete values of the conjugate parameter to the OPD, the wavenumber σ.

[0208] For further details about the theory of Fourier transformation and the mathematical steps for computing the mathematical spectrum as an approximation to the real physical spectrum, starting from the measured interferogram, the reader is referred to textbooks such as Chamberlain (1979) The principles of interferometric spectroscopy, John Wiley and Sons, which is incorporated by reference as if fully set forth herein.

[0209] In some highlights of chapters 2, 4, 5 and 6 of Chamberlain (1979) the following basics of the Fourier transform manipulation and relevant considerations are described. The Fourier integral relation between a function f(k) and its Fourier transform F(x) is shown on page 31. In principle, f(k) and F(x) are continuous functions of their variable, but in practice they are always known for discrete values, so that the Fourier integral is approximated by an infinite sum as shown on page 55. The infinite sum is in turn substituted with a finite sum as shown on page 57. The perfect and practical interference functions in the case of electromagnetic radiation are derived as shown on pages 96 and 104. The relation between the physical spectrum and the mathematical spectrum is shown on page 127, and the theory of sampling and correction of phase errors are shown in sections 6.7 to 6.11. Finally, the Fast Fourier Transform algorithm is detailed in chapter 10, and is shown to operate only when the discrete intervals are all equal, yet this operation is faster than the straight Fourier summation.

[0210] It will be appreciated by one ordinarily skilled in the art that the third step of the method described hereinabove can be accomplished in many alternative ways. One of these alternatives is as follows.

[0211] First, one of the frames is defined as a reference frame. In principle it is not important which frame is selected as the reference. In practice, however, it is better to select a frame which is roughly centered with respect to the set of translation vectors, so that the overall spatial overlap between the selected frame and all the other frames is maximized. Thus selecting the reference frame assists in finding the translation vectors for each of the frames measured.

[0212] Second, a subtraction image which is the difference in intensity between a first frame and the reference frame is displayed.

[0213] Third, the first frame is moved in small steps to the right-left and up-down directions while always displaying the intensity difference, until a position in which the displayed subtraction image is substantially zero everywhere, or has substantially no features, is found. In the ideal case, in which the movement is completely rigid, the subtraction image equals zero at all pixels of overlap. In practice, however, there will always be a slight pattern, and then the best position is the one in which this pattern is minimized in intensity. Experience proved that it is quite easy to find the substantially zero position by eye, because a slight lack of spatial registration emphasizes the differences between two frames, which are therefore easy to detect. This procedure can be automated using various known algorithms, see for example Anil K. Jain (1989) Fundamentals of digital image processing. Prentice-Hall International and system sciences science, pp. 400-402. However, due to the presence of fringes superimposed on the frames, it is preferred that a fringe suppression algorithm is employed prior to automatic spatial registration of the frames.

[0214] Fourth, the translation vector for the first frame is recorded.

[0215] Fifth, the procedure is repeated for all additional frames of the measurement.

[0216] And finally, knowing the OPD dependence on position (it is a specific dependence for every interferometer), the OPD vector for every pixel in every frame is calculated and stored.

[0217] Problems which may arise during measurement and which affect the final results are mostly associated with the amplitude of the object's movement. For the method to be useful, the amplitude of the movement is preferably not too large. In particular, a number of possible problems may arise.

[0218] First, entire regions of the interferogram maybe missing, making it very difficult to interpolate (in the case of interpolation).

[0219] Second, if the central portion of the interferogram is completely missing, the Fourier transform cannot be calculated.

[0220] And finally, if due to the movement of the examined object the actual OPD steps (after correction for the movement) are larger than the Nyquist condition (one half the lowest wavelength of sensitivity of the instrument), spurious results may be introduced.

[0221] Nevertheless, corrective actions may be undertaken, some of which are listed hereinbelow.

[0222] First, in a case where the central portion is present, it is usually easy to find the center of the interferogram. In this case, if the interferogram is symmetric, data points on one side of it can be reflected to the other side, filling holes in the process.

[0223] Second, take the smallest OPD steps compatible with the needed spectral resolution and measurement time. This again will have the tendency of not allowing large holes in the interferogram.

[0224] Third, repeat the interferometer scan two or three times and then join the data as if measured in one measurement. Thus, if an OPD is missing in one of the scans, chances are (for random movement) that it will not be missing in another.

[0225] Fourth, in an interferometer in which the OPD in a frame varies in one direction only (e.g., horizontal), and if the movement of the object is in one direction only (e.g., the human eye displays preferentially involuntary horizontal movements), make sure the instrument is rotated around the optical axis, so that the OPD gradient is perpendicular to the direction of the object movements. This way the movements affect only the spatial registration of the frames and the interferograms stay almost unaffected, reducing significantly one source for errors.

[0226] Fifth, in the case of a featureless object, it is expected that a movement will not affect the results significantly, since all pixels are equivalent in any case.

[0227] And finally, the following distinctions should be made: (i) an object that moves rigidly and linearly on a plane, i.e., the object moves such that all of its parts keep their shape and size, and their relative distances remain constant as seen through the spectral imager, and (ii) an object that moves linearly on a plane, such that all of its parts keep their shape and size, but the relative distances of the parts may vary in time. Obviously, the former case is simpler than the latter. In addition, once an acceptable solution is found for the former, the latter can in general be solved by segmenting the object into individual areas which may move with respect to one another, although each one separately moves rigidly, and then applying to the individual areas the solution of the former case.

[0228] It should be noted that the considerations described hereinabove are valid for certain type of movements, in particular rigid linear movements (both random or not). However, it will be evident that some of the considerations described hereinabove can be generalized to other types of movements, e.g., non-rigid and/or non-linear movements. In any case, a rotation of the object around an axis which is perpendicular to the line of sight of the instrument cannot be addressed in principle, because parts of the object will change shape and disappear from view during the measurement, and obviously then the measurement will be incomplete.

[0229] As is appreciated by one ordinarily skilled in the art, solving the problem of the moving object is equivalent to calculating the Fourier transform of an interferogram defined for values which are inherently not equally spaced. This problem is known in radio astronomy (see, Synthesis Imaging (1986) Perley, Schwab and Bridle, Report of Summer School of the National Radio Astronomy Observatory, p. 72, Greenbank W. Virginia), where there is clumping of data in the low OPD range, and this introduces large undulations in image intensity which make it difficult to detect weak point sources.

[0230] Obviously, in the cases of a living tissue, it is generally difficult, if not impossible, to keep the analyzed tissue completely motionless. This is due to respiration, heart beat, involuntary movements of the patient, etc. Even when the tissue itself is forced to be stationary by external mechanical means (for example special holders to keep an eye stationary during corneal surgery), the mere fact that blood is circulating in the vessels induces a small movement in the examined tissue. In this case, especially when the object is magnified through a microscope, the movement of the analyzed area is also magnified.

[0231] An interferometer based spectral imager as taught in U.S. Pat. No. 5,539,517, combined with the described for spatial registration and spectral correction, or in other words compensating both spatially and spectrally for movements of the examined object, based on the spectral information that it provides, not only may enable noninvasive evaluation of the oxygen saturation level of hemoglobin in retinal blood vessels and hemoglobin concentration thereat, but also, because of the imaging information that it provides, it may be used for the detection and mapping of retinal ischemia. Joined to advanced spectral analysis algorithms such as but not limited to principal component or neural network algorithms, it may also prove useful for classification of the different retinopathy stages, and treatment categorization of for example diabetic patients.

[0232] Many chemicals in the living tissue are related to the functioning of the vessels and to metabolism. Therefore, even though the primary element for retinal ischemia is oxygen, which can be measured through the concentration of hemoglobin in the oxy- and deoxy forms, important information can be obtained also by measuring the concentration of other constituents, such as NAD+, NADH, flavin, cytochromes, etc.

[0233] Considering the large amount of prior art that has been described for spectral detection of such chemical constituents of tissue, correlating the absorption peaks in reflectance, and the fluorescence peaks in UV or blue light, single or multiple wavelengths excitation, to their concentrations, it is conceived that an interferometer based spectral imager as taught by U.S. Pat. No. 5,539,517, combined with the described method may be used to map concentrations of one or more of such constituents simultaneously in living non-motionless organs/tissues. The particular hardware configuration in which the imager will be operated, will dictate the type and amount of information obtained.

[0234] For example, the simplest and most straightforward configuration is when the imager is attached to the CCD port of a fundus camera, so that the retina is imaged, and the same wide band white light source of the fundus camera is used to measure the reflected light from the retina. In this case oxygen concentrations can be measured using Delori's algorithm [Delori (1995) Appl. Optics 27:1113-1188, and Appl Optics, Vol. 28, 1061; and, Delori et al. (1980) Vision Research 20:1099], or similar, extended to all pixels of the imaged retina. More complicated systems based on interferometer based spectral imagers are: (i) auto-fluorescence spectral imaging; (ii) spectral imaging using UV or blue light fluorescence excitation; (iii) spectral imaging using laser excited fluorescence, singly, simultaneously, or in succession, at the following wavelengths: 650, 442, 378, 337, 325, 400, 448, 308, 378, 370, 355, or any other equivalent wavelengths which give similar information.

[0235] These configurations can be built in several ways, either separately or combined in any number of combinations in the same instrument: the instrument is made of the light source(s), the fundus camera and the spectral imager, including a computer and software to interpret the data and display it in a useful way for the ophthalmologist.

[0236] In all cases of white light reflection, auto-fluorescence, single wavelength continuous wave laser excitation fluorescence, or multiple wavelength laser excitation fluorescence, the sample is illuminated and a spectral image is measured.

[0237] In the case of pulsed laser illumination, the method of work of the spectral imager is slightly modified and requires some hardware changes which are not basic and substantial, but important for the instrument to operate. These changes are the following.

[0238] For single pulsed laser excited fluorescence spectral imaging, the laser pulses and the frame grabbing of the CCD of the imager are synchronized with the scanning angle of the interferometer, so that at each pulse the interferometer performs a step, and a new frame is collected by the computer (several pulses can also be used in general for each frame, as long as this number does not change from frame to frame). In this way, at each OPD value, the interferogram value corresponds to the same number (but different) of pulses of the laser. This is necessary to ensure that each frame is taken with the same total illumination intensity, otherwise, each frame measures the fluorescence resulting from a different number of laser pulses and the interferogram will be distorted.

[0239] For several pulsed lasers induced fluorescence spectral imaging, the method of work can be in two ways: (i) collect a whole spectral cube for each laser separately as above, in succession; this means that during a measurement only one laser is activated, and at the end there is one spectral cube measured for each laser wavelength; and, (ii) pulse each laser in succession in synchronization with the interferometer and the frame grabbing, so that all the lasers are switched in succession before the next step of the interferometer and the next frame is taken; in this case, at the end, only one spectral cube is measured.

[0240] All the information must be analyzed and interpreted. The most important algorithms are going to be of a type that compares the resulting intensities between different wavelengths and between different pixels of the image. These algorithms should consider variations of intensities, and ratios between different regions in the tissue and between different wavelengths. The method will be very sensitive, and may replace slit lamp imaging (white light or filtered light), because it will provide a large quantitative information.

[0241] Other applications will be apparent to one ordinarily skilled in the art. These include visual loss due to choroidal ischemia, acute sectorial choroidal ischemia, ischemic optic neuropathy, corneal and iris problems, etc., and many others which are analyzed today by imaging techniques, either using white light or fluorescence of different origins.

[0242] Since the spectral imagers according to U.S. Pat. No. 5,539,517 can be attached to any imaging optics including endoscopes and laparoscopes, it may be used as an aid to the surgeon before, during or after surgery to accurately define the diseased tissue to be removed, to aid in the decision where to start cutting, where to stop, and to judge whether all diseased tissue has been removed during an operation procedure. These spectral imagers are intrinsically suitable to analyze the nature of the tissue through the chemical composition, related in turn to its spectral characteristics, and to provide a visual map (usually enhanced), for a user to grasp, take decisions and act.

[0243] In the case of cancerous tissues detection in vivo, both the hardware configurations and the types of analysis and display algorithms involved are very similar to the above described ophthalmologic examples. The differences are in the collecting optics (endoscopes of different types instead of for example a fundus camera), in the types of some basic molecular components involved in the detection: some of these are probably common, such as oxygen concentration, additional others are collagen and elastin, genetic material in the cell nuclei, such as DNA chromatin, etc. The illumination and synchronization requirements in the case of multiple wavelengths or pulsed excitation are similar as well [Pitris et al., Paper presented at European Biomedical Optics Week by SPIE, Sep. 12-16, 1995, Barcelona Spain].

[0244] In all these examples spatial registration and spectral correction are required and are provided by the described method.

[0245] The power of the spatial registration and of the spectral correction method is further described and exemplified hereinbelow.

[0246] Spatial Registration and Spectral Correction—the Effect on the Image:

[0247] FIG. 4 a presents a spectral image of the optic disk of the retina of a right eye of a healthy individual using the SPECTRACUBE system, while not employing spatial registration and spectral correction procedures as described in accordance with the method of the present invention. FIG. 4b, on the other hand, presents the very same image after spatial registration and spectral correction procedures according to the present invention.

[0248] In both images the optic disk appears lighter in the middle portion of the image along with blood vessels nourishing the optical nerve with oxygen and other nutrients (arterioles) and removing waste and carbon dioxide generated during metabolism (veins). However, as is clearly evident comparing the two images, due to movements of the eye during measurement, the image of FIG. 4a is highly blurred. Corrective action according to the method of the present invention, in which spatial registration and spectral correction were applied, resulted in a much clearer image as shown in FIG. 4b.

[0249] Furthermore, the images presented in FIGS. 4a and 4b show not only the spatial organization of the tissue, as they also present spectral information, although not in a direct fashion. In other words, the different colors present in the images result from the application of an RGB algorithm to the spectrum of each pixel of the image such that each pixel, according to its spectrum and according to the preselected RGB function is presented by RGB colors in corresponding intensities. It is clear that as a result of the distorted spectra associated with pixels of the image shown in FIG. 4a, as is further demonstrated hereinbelow, the RGB function yields different results when applied to either image.

[0250] This Example emphasizes the importance of spatial registration and spectral correction to obtain clear and informative image of the examined moving object, the eye in the present case.

[0251] The following examples demonstrate specifically the importance of spectral correction to obtain meaningful spectral information from selected regions of the examined object, which information may be employed to achieve information about the metabolic condition of an examined tissue, etc.

[0252] Spectral Correction—the Effect on the Interferogram:

[0253] FIG. 5 a presents a portion of an interferogram calculated for a single pixel (x=112, y=151) of the image presented in FIG. 4a, i.e., while not employing spatial registration and spectral correction procedures as described in accordance with the method of the present invention. FIG. 4b, on the other hand, presents the corresponding portion of an interferogram of the very same pixel after spatial registration and spectral correction procedures according to the present invention.

[0254] Examining the interferogram of FIG. 4a reveals that the left and central parts of the function (measured in equal intervals of time) resembles a typical interferogram, whereas the right portion of the function, is totally atypical. The local maximum indicated by an arrow is due to sudden motion of the examined object (e.g., a saccadic motion of the eye). The uncharacteristic increase of signal is due to the fact that a different point in the object being measured suddenly appeared not in its original place, giving a different value for the interferogram function, as compared with a situation wherein the object remains stationary.

[0255] Nevertheless after spatial registration and spectral correction procedures according to the present invention are applied, as shown in FIG. 5b, the interferogram function of the same pixel appears typical.

[0256] As can be seen in Figure Sb, the corrected interferogram is well behaved. It does not have spurious discontinuities or uncharacteristic portions characterizing the non-corrected interferogram of FIG. 5a.

[0257] However, the corrected interferogram of FIG. 5b is now defined in nonuniform intervals. For example one notices that around frame number 107 the density of data is low, meaning that the eye moved in a direction opposite to the scanning direction of the interferometer, increasing the OPD intervals around it, whereas around frame number 109.5, which is an artificial frame number formed due to the magnitude of movement of the eye in the same direction as the scanning direction of the interferometer, the density of data is higher, decreasing the OPD intervals around it.

[0258] There are therefore, several routes by which one can perform the Fourier integral to approximate the physical spectrum of the specific pixel. According to one route one can interpolate between the given OPD values and then define a new interferogram having equally spaced OPD values, thus allowing use of a fast Fourier transform algorithm to approximate the physical spectrum of that pixel. According to another route, one can calculate the Fourier integral as the sum of the interferogram values weighted according to their own intervals using Equation (16):

f (σ)=1/K·ΣF(xi)Δie(iσxi)  (16)

[0259] where K is a constant, f(σ) is the value of the spectrum at wavelength λ=1/σ, and Δi is the difference between the OPD at xi and the OPD at xi+1. It will be apparent to one ordinarily skilled in the art, that there may be additional ways to approximate the physical spectrum, such as methods as described in Synthesis Imaging (1986) Perley, Schwab and Bridle, Report of Summer School of the National Radio Astronomy Observatory, p. 72, Greenbank W. Virginia.

[0260] Spectral Correction—the effect on the spectrum:

[0261] FIG. 6 a presents spectra of five adjacent pixels derived from the image of FIG. 4a, while not employing spatial registration and spectral correction procedures as described in accordance with the method of the present invention. Four of these pixels are centered around the fifth which is the pixel whose interferogram is shown in FIG. 5a. FIG. 6b, on the other hand, presents spectra of the same five pixels after application of the spatial registration and spectral correction procedures according to the present invention. The dip around 575 nm is characteristic of oxyhemoglobin absorption.

[0262] Comparing the spectra of FIGS. 6a and 6b, one notices two phenomena. First, corresponding spectra are much noisier in FIG. 6a as compared with FIG. 6b. Second, when implementing the method of the present invention, as shown in FIG. 6b, from pixel to pixel the spectra change in a uniform pattern presenting an expected smooth behavior over the entire spectral range, whereas none such behavior can be seen in the spectra of FIG. 6a.

[0263] Thus, these examples emphasize the importance of spectral correction to obtain meaningful interferograms and spectra derived from an examined moving object.

[0264] Spatial Registration of Frames Assisted by Fringe Suppression:

[0265] The raw data of a randomly moving object as measured by an interferometric spectral imager, should be preprocessed before the Fourier Transform is calculated on the pixels interferograms to obtain the best final spectral cube.

[0266] This is due to the fact that in a spectral imager based on a Sagnac or similar type interferometer as herein described, the instantaneous Optical Path Difference (OPD) corresponding to an interferogram data point depends not only on the specific CCD frame but also on the specific pixel to which that data point refers.

[0267] As a result, if the object moves during the measurement, the pixel occupied by a point on the object is different than if the object is stationary, and if no correction is used, the Fourier Transform algorithm uses the wrong OPD for that data point. The resultant spectral image cube can be significantly corrected if by some means the algorithm is made to use the appropriate OPD for each data point instead of the inappropriate one. Finding the appropriate OPD for each interferogram data point requires (i) spatial registration of each acquired frame and recording of its registration vector; and (ii) calculation of the actual OPD for each data point, based on the registration vectors and on the OPD dependence on position.

[0268] However, there is one physical phenomenon, the appearance of fringes, which makes the frame registration more difficult, yet not impossible, when performing this registration automatically. As shown in FIG. 7a, fringes are straight line stripes of intensity modulation superimposed on the frame, which slightly change position, with respect to the frame on which they appear, depending on the scanning position of the interferometer. The origin of the stripes is due to constructive (light stripes) and destructive (dark stripes) interference of the light rays while passing through the interferometer, and their shape (vertical or horizontal straight lines, depending on optical alignment) is due to the fact that all the pixels on a vertical line (or horizontal, respectively) go through the same OPD for every scanned frame, so that they undergo the same amount of interference (for the same wavelength of light). The change in position from frame to frame is due to the fact that the constructive or destructive level of interference for a certain pixel changes according to the interferometer position while scanning. The stripes are not very bothersome when registering the scanned frames by eye one on top of the other, because despite the fringes, the other features (e.g., patterns of blood vessels in the eye) are well visible in each frame, and the appearance of the stripes does not prevent an observer, when superimposing one frame over the other, from deciding the best spatial registration.

[0269] However, when using an automatic algorithm, these stripes may introduce a difficulty, because they represent a nonuniform light intensity change, superimposed on the features of the frame. As already mentioned, the fringes are vertical (or horizontal) stripes which travel in position from frame to frame in a direction perpendicular to the stripes, in unison with the interferometer mirror(s) rotation.

[0270] The input of the fringe suppression algorithm is the cube of interferogram frames with fringes and the output a cube of frames without fringes, as further described hereinbelow.

[0271] Few assumptions are made concerning the operation of the fringes suppression algorithm.

[0272] One assumption is that the fringe “frequency” is approximately known. In other words, it is assumed that the distance in pixels between adjacent fringes is approximately known. One may gain this knowledge from previous experience on a certain type of sample, from each of the frames of the interferogram cubes themselves, or from a calibration process.

[0273] As seen in FIG. 7a, the fringe information is very compactly located in the frequency domain. The center frequency of the fringe can be easily found and the width of the fringe information in the frequency domain is assumed to be constant or nearly constant for all of the scanned frames.

[0274] The fringes suppression algorithm therefore suppresses the fringes by artificially zeroing or interpolating out the signal in the frequency range of the spatial frequency domain where the fringes information lies, for each scanned frame.

[0275] Since the fringes are almost parallel to one of the axes (say x-axis), one can divide the frame into vectors along the axis that is perpendicular to the fringes (say y-axis). FIG. 7b shows the intensity of 200 pixels of such a vector, wherein the fringes are clearly evident between the 100th pixel and the 150th pixel. As shown in FIG. 7c, each vector is thereafter transformed to the frequency domain using, for example, the fast Fourier transform algorithm (FFT), the peak ranging from ca. 0.15 to ca. 0.35 pixel−1 contains the fringe information. As shown in FIG. 7d, for each vector the frequency region where the fringe information is located is zeroed, and, as shown in FIG. 7e, transform back to the spatial domain using, for example, inverse fast Fourier transform algorithm (IFFT). This procedure is performed for each of the vectors of every frame grabbed by the spectral imager while scanning the interferometer and results in a fringes suppressed frame, as shown in FIG. 7f.

[0276] Should for some reason the fringes be offset angularly, i.e., not arranged in exact vertical or horizontal direction, a small decrease in the frequency of the fringe information band will result. This problem can be solved by increasing the width of the region of zeroing or interpolation of the signal in the spatial frequency domain where the fringe information lies.

[0277] As is evident from FIG. 7c, most of the frame's energy is located in the lower bands in the frequency domain. Using a band-stop filter not only preserves the information in each scanned frame but also doesn't blur the frame since the energy in the higher bands is not attenuated, and the edge information is preserved.

[0278] It will be appreciated by one ordinarily skilled in the art that using the Hough transform [Paul V. C. Hough, “Methods and means for recognizing complex patterns”; and U.S. Pat No. 3,069,654, both are incorporated by reference as if fully set forth herein], one can extract the frequency position of the fringe information and use it for the fringe suppression algorithm. The Hough transform can also find the orientation of those fringes and make the necessary adjustments.

[0279] To keep the signal real after the IFFT, the zeroing procedure is performed preferably symmetrically relative to the origin of the spatial frequency axis (Even though not shown in the Figure, the signal in the frequency domain is defined for both positive and negative values of the frequency ƒ, and it is an even or symmetric function of ƒ). The signal after the IFFT, as shown in FIG. 7e, has a very small imaginary residual part that is eliminated using the absolute (or the real) part of the result.

[0280] Returning to FIGS. 7b and 7e, instead of executing the FFP, zeroing and IFFT procedures as hereinabove described, one can simply interpolate the plot of FIG. 7b through the fringes region, intersecting each of the fringes intensity peaks substantially at their central relative intensity to obtain an interpolated plot, as indicated by I (for intersecting) in FIG. 7b, which is otherwise very similar to that shown in FIG. 7e.

[0281] An additional option is, for the region where the fringes information lies in the spatial frequency domain, instead of zeroing the peak (as shown in FIG. 7e), draw a straight line interpolation between the edge points of the peak.

[0282] The preferred fringe suppression algorithm according to the present invention is hereinbelow described in mathematical terms.

[0283] Let X(x,y) be the input frame (as, for example, shown in FIG. 7a), Y(x,y) the corresponding output frame (as, for example, shown in FIG. 7a), x and y are the discrete coordinates of a pixel in the frame, ƒCF the center frequency of the fringe information, ƒLF the low frequency of the fringe information, ƒHF the high frequency of the fringe information, Δf the width of the fringe suppression band and u(f) a step function.

[0284] By definition:

ƒLF=ƒCF−0.5Δƒ  (17)

ƒHF=ƒCF+0.5Δƒ  (18)

[0285] A “zeroing band” function is defined as:

W (ƒ)={1−[u(ƒ−ƒLF)−u(ƒ−ƒHF)]−[u(ƒ+ƒLF)−u(ƒ+ƒHF)]}  (19)

[0286] W(ƒ) (the “zeroing band” function) is defined as a function of the frequency ƒ such that, when multiplied by another function of the frequency ƒ, it leaves it unaltered for values of ƒ lower than ƒLF and higher than ƒHF, and changes it to zero for values of ƒ higher than ƒLF and lower than ƒHF.

[0287] The output frame without fringes can then be expressed as:

Y (x,y)=Re{IFFT{W(ƒ)*FFT{X(:,y)}}}  (20)

[0288] Using the fringe suppressed frames will assist in automatic registration procedures, which otherwise may face difficulties due to the repetitive pattern of the fringes superimposed on the frames.

EXAMPLE 5

Spectral Imaging of Selected Eye Tissues

[0289] The method described and exemplified under Example 4 above was employed to obtain spectral images of eye tissue of healthy and diseased patients as described in the following examples. It should, however, be noted that mechanical and chemical methods for eye fixation are well known in the art and are extensively employed during, for example, invasive eye procedures, such as laser operations. Such methods may be also be employed for spectrally imaging eye tissues according to the present invention. Furthermore, as mentioned above, should the spectral imager of choice be a non-interferometer based imager (e.g., a filters based spectral imager), only conventional spatial registration is required for analyzing the eye. In addition eye tracking methods may be employed. Such methods are used in laser operations to track eye movement.

[0290] In the following, reflection of visible light illuminated via the iris was employed for spectrally analyzing eye tissues of healthy and diseased patients. Illumination was with white light and all the spectral data collected was between 500 and 620 nm. This is the region of hemoglobin absorption, which is the most prominent spectral feature of the retina. This spectral region, however, contains also the long wavelength tails of other absorbing structures: the eye media, the macular pigment and the melanin [Van Norren and L. F. Tiemeijer, Spectral reflectance of the human eye, Vision Res., Vol. 26 No. 2, pp. 313-320, 1986].

[0291] The spectral variation of the specific extinction coefficients of oxygenated (HbO2) and deoxygenated (Hb) hemoglobin are shown in FIG. 9 [see also Delori F. C., Noninvasive technique for oximetry of blood in retinal vessels, Applied Optics Vol. 27, pp. 1113-1125, 1988, taken from O.W. van Assendelft, Spectrophotometry of Hemoglobin Derivatives, C.C. Thomas, Springfield, Ill., 1970]. HbO2 presents two peaks, at 540 and 578 nm, while Hb presents only one peak, at 558 nm. As is well known from the prior art [see for example, Delori F. C., Pfilbsen K.P., Spectral reflectance of the human ocular fundus, Applied Optics Vol. 28, pp. 1061-1077, 1989], peaks in the extinction coefficient mean dips in the reflectance spectrum. Delori [Delori F. C., Noninvasive technique for oximetry of blood in retinal vessels, Applied Optics Vol. 27, pp. 1113-1125, 1988] has shown how reflectance measurements of retinal vessels can be used to measure oxygen saturation (O2Sat) and vessel diameter. Delori's work pioneered this field, but it was only related to vessels of the retina, and it had no imaging capability. As a result Delori did not present spatial maps of O2Sat over the whole retina, and did not map the different regions of the ocular fundus, such as macula, disk, etc. on the basis of spectral information.

[0292] In this work the model presented by Shonat [Ross D. Shonat, Elliot S. Wachman, Wen-hua Niu, Alan P. Koretsky and Daniel Farkas, Simultaneous hemoglobin saturation and oxygen tension maps in mouse brain using an AOTF microscope, Biophysical Journal (1997), in press] was extended to fit the spectra of all retinal regions so that the data of a whole spectral image can be used to map various features, such as, but not limited to, vessels, O2Sat, optical density of total hemoglobin, spectral differentiation between healthy, intermediate and degenerate macular tissue, spectral differentiation between healthy and glaucomatous optic disks and cups, etc.

[0293] a. Modeling the Retina

[0294] FIGS. 8 a -b show spectral images of a retina obtained using the SPECTRACUBE spectral imager. The color presented by each pixel in the images is determined by an RGB algorithm as described under Example 2 above. The weighting functions were wr(570-620 nm), wg(530-570 nm) and wb(500-530 nm) mimicking full transmission in the selected ranges and no transmission outside the selected ranges.

[0295] FIG. 10 shows the inverted log of reflectivity spectra (proportional to extinction coefficient), as measured by the SPECTRACUBE system, of one pixel of a vein and one of an artery. It is seen that the peaks in the vein are less pronounced than in the artery, as expected from the known oxygenated and deoxygenated hemoglobin extinction spectra shown in FIG. 9.

[0296] FIG. 11 shows spectra of pixels from the disk, the cup, the retina, and from a retinal blood vessel. The spectral resolution of this measurement is low, approximately 20 nm, and this is the reason for the shallowness of the dips seen. It is well known in the literature [for example, Patrick J. Saine and Marshall E. Tyler, Ophthalmic Photography, A textbook of retinal photography, angiography, and electronic imaging, Butterworth-Heinemann, Copyright 1997, ISBN 0-7506-9793-8, p. 72] that blue light is mostly reflected by the outer layers of the retinal tissue, while as the wavelength increases, the light is reflected by deeper and deeper layers.

[0297] FIG. 12 is a schematic diagram of the reflection of different wavelengths from different retinal depths. This means that monochromatic images show features characteristic of different depths.

[0298] Different models, such as, for example, the well known O2Sat model used by Delori for retinal vessels, and by Shonat et al., on the surface of rat brain, [Delori F. C., Noninvasive technique for oximetry of blood in retinal vessels, Applied Optics Vol. 27, pp. 1113-1125, 1988, and Ross D. Shonat, Elliot S. Wachman, Wen-hua Niu, Alan P. Koretsky and Daniel Farkas, Simultaneous hemoglobin saturation and oxygen tension maps in mouse brain using an AOTF microscope, Biophysical Journal (1997), in press, both are incorporated by reference as if fully set forth herein], and modifications thereof, might be used to fit the spectral data in each pixel of the image separately to explain the results.

[0299] Such models, if successful, might predict the presence, absence or amount of physiologically important metabolites, such as, but not limited to, hemoglobin, cytochromes, NAD, NADH and flavins, pixel by pixel, and, once displayed in a spatially organized way, may be the basis for highlighting regions of impaired tissue “vitality” or “viability”.

[0300] FIGS. 13 a -c shows spectra extracted from several pixels of a spectral image measured with the SPECTRACUBE system, belonging to different anatomic regions of the retina (FIGS. 13b-c) as compared to spectra measured and published by Delori (FIG. 13a). FIG. 13a presents spectra described by Delori derived from the retina, perifovea and the fovea using point spectroscopy. FIGS. 13b-c presents spectra measured using the SPECTRACUBE system of the same tissues (FIG. 13b) and of a retinal artery, retinal vein and a choroidal blood vessel (FIG. 13c). Comparing FIGS. 13a and 13b, the similarity of the results is evident, although there are also some differences, which may be due to patient variability.

[0301] FIG. 13 c shows the spectra of a retinal artery and a retinal vein and of a choroidal blood vessel. The peak at 560 nm is more pronounced in the artery and the choroidal vessel than in the vein, as expected from higher oxygenation of hemoglobin thereat.

[0302] FIGS. 14 a -e show a portion of retina including retinal blood vessels from a healthy individual. FIG. 14a shows an RGB image of the retina, wherein wr(570-620 nm), wg(530-570 nm) and wb(500-530 nm).

[0303] FIG. 14 b shows an enhanced RGB image of the retina. The intensities of the three colors, red, green and blue, i.e., RGB, are related algebraically to the normalized intensities in the three spectral bands B1, B2 and B3, defined as follows B1=525-590 nm, B2=600-620 nm and B3=500-650 nm. In each of these bands the integral intensity was calculated for each pixel. The intensity was then scaled so that the minimum value over the whole image was zero and the maximum value was one. The red intensity was then given by: R=B2/(1+B1); the green intensity was given by: G=B3; whereas the blue intensity was given by B=(1+B1−B2)/(1+B1). The latter RGB algorithm was employed to specifically enhance the spectral difference between retinal veins (dark red) and arteries (light red). Thus, using the present invention it is possible to strongly enhance metabolic characteristics of the retina and retinal blood vessels.

[0304] FIGS. 14 c and 14d are gray level images wherein for each pixel light of the specified wavelengths (610 and 564 nm, respectively) is given a gray level according to its intensity. Please note that only the vein is highlighted at 610 nm, whereas both arteries and the vein are highlighted at 564 nm. Thus, images at different wavelengths are shown to highlight different aspects of the retinal physiology.

[0305] FIG. 14 e is a hemoglobin oxygenation map for the retinal blood vessels. The map of FIG. 14e was calculated using, for each pixel, the algorithm for O2Sat developed by Shonat [Ross D. Shonat, Elliot S. Wachman, Wen-hua Niu, Alan P. Koretsky and Daniel Farkas, Simultaneous hemoglobin saturation and oxygen tension maps in mouse brain using an AOTF microscope, Biophysical Journal (1997), in press].

[0306] Deoxygenated blood has a higher extinction coefficient in the red wavelengths than oxygenated blood (FIG. 9), and therefore veins look slightly darker and with a slightly different color than arteries, because they carry blood at different levels of oxygenation (see FIG. 14a). However, the color difference is very small and in a conventional color image of the fundus, it can be hard to distinguish between them, except, in some cases, for the largest vessels. Oxygenation mapping or simple enhancing artificial RGB mapping based on spectral features may be a tool that significantly enhances the distinction between the different type of vessels.

[0307] FIG. 15 shows spectra derived from a hemorrhage and healthy retinal regions of a patient suffering from diabetic retinopathy. Please note that the spectra of the affected retinal region is much flatter, probably due to lower levels of oxygenated hemoglobin than the one present in healthy retina.

[0308] b. Modeling the Macula

[0309] Inverted log spectra of the macula have been shown by Brindley at al. [G. S. Brindley and E. N. Willmer, The reflexion of light from the macular and peripheral fundus oculi in man, Journal of Physiology Vol. 116, pp. 350-356, 1952]. Spectral reflectance of the fovea and the log reflectance difference of the normal periphery and fovea are shown by Van Norren et al. [Van Norren and L. F. Tiemeijer, Spectral reflectance of the human eye, Vision Res., Vol. 26 No. 2, pp. 313-320, 1986].

[0310] FIG. 16 shows the inverted log reflectivity spectra of normal, intermediate and degenerate macular tissue of a single patient suffering macular degeneration. The spectra for the macular tissue represents average of twenty five pixels per region. The spectrum of the degenerated macula was divided by a factor of four as indicated by “x4”. It is clearly evident that the spectral signature of the normal, intermediate and degenerated macular tissue are definitely different from one another. Please note that a gradual spectral change toward degeneration spectral signature is apparent in the intermediate tissue. The gradual spectral change from normal to diseased tissue may be used for early detection of the disease and for following disease progression.

[0311] FIG. 17 shows a region in the macula of the above patient ranging from normal (dark) to degenerate (light). The algorithm employed to enhance the spectral signatures of the macular regions was an RGB algorithm where wr(570-620 nm), wg(530-570 nm) and wb(500-530 nm) weighting functions were selected. Please note that while normal macular tissue absorbs most of the illuminated light (i.e., reflects a small fraction thereof) and therefore appears dark, degenerated macular tissue reflects most of the light (i.e., absorbs a small fraction thereof) and therefore appears light. This result is in good agreement with the presence of cones and rods in normal macula and their absent from degenerated macula, since the rods and cones are, by nature, excellent light absorbers.

[0312] c. Modeling the Optic Disk

[0313] Because of the great importance of the optic disk physiology in the diagnosis of glaucoma, results of the analysis of the spectral images as they pertain to this region of the ocular fundus are presented.

[0314] FIGS. 18 a -d show the optic disk of a healthy individual. FIG. 14a shows an RGB image of the disk, wherein wr(570-620 nm), wg(530-570 nm) and wb(500-530 nm). FIGS. 18b and 18c are gray level images wherein for each pixel light of the specified wavelengths (610 and 564 mn, respectively) is given a gray level according to its intensity. FIG. 18d is a hemoglobin concentration map of the disk blood vessels. The map of FIG. 18d was calculated using, for each pixel, an algorithm similar to that used for Hb concentration by Shonat [Ross D. Shonat, Elliot S. Wachman, Wen-hua Niu, Alan P. Koretsky and Daniel Farkas, Simultaneous hemoglobin saturation and oxygen tension maps in mouse brain using an AOTF microscope, Biophysical Journal (1997), in press].

[0315] FIGS. 19 a -e show the optic disk of a glaucoma suspect. FIG. 19e is an image key, schematically presenting the location of the optic disk and cup in FIGS. 19a-d. FIG. 19a shows an RGB image of the disk, wherein wr(570-620 nm), wg(530-570 nm) and wb(500-530 nm). FIGS. 19b and 19c are gray level images wherein for each pixel light of the specified wavelengths (610 and 564 nm, respectively) is given a gray level according to its intensity. FIG. 19d is a hemoglobin concentration map of the disk blood vessels. The map of FIG. 19d was calculated using, for each pixel, an algorithm for Hb concentration similar to that used by Shonat [Ross D. Shonat, Elliot S. Wachman, Wen-hua Niu, Alan P. Koretsky and Daniel Farkas, Simultaneous hemoglobin saturation and oxygen tension maps in mouse brain using an AOTF microscope, Biophysical Journal (1997), in press]. The plots under and to the right of FIG. 19d represent the hemoglobin concentration along the horizontal and vertical lines crossing FIG. 19d.

[0316] Please note that a striking difference in hemoglobin concentration is clearly evident comparing the images of the healthy individual (FIG. 18d) and the glaucoma patient (FIG. 19d).

[0317] While the invention has been described with respect to a limited number of embodiments, it will be appreciated that many variations, modifications and other applications of the invention may be made.

Claims

1. A method of evaluating a medical condition of a patient comprising the step of enhancing spectral signatures of an eye tissue of the patient by: (a) providing an optical device for eye inspection being optically connected to a spectral imager; (b) illuminating the eye tissue of the patient with light via the iris, viewing the eye tissue through said optical device and spectral imager and obtaining a light spectrum for each pixel of the eye tissue; (c) attributing each of said pixels a color according to its spectral signature, thereby providing an image enhancing the spectral signatures of the eye tissue; and (d) using said image to evaluate the medical condition of the patient.

2. The method of claim 1, wherein said medical condition is selected from the group consisting of diabetic retinopathy, ischemia of the eye, glaucoma, macular degeneration, CMV eye infection, retinitis, choroidal ischemia, acute sectorial choroidal ischemia, ischemic optic neuropathy, and corneal and iris problems.

3. A display comprising an image presenting an eye tissue, wherein each pixel in said image is assigned a color according to a spectral signature of a tissue element from which it is derived, thereby enhancing the spectral signature of the eye tissue.

4. A spectral bio-imaging method for obtaining a spectrum of a region of an eye tissue, the method comprising the steps of: (a) providing an optical device for eye inspection being optically connected to a spectral imager; (b) illuminating the eye tissue with light via the iris, viewing the eye tissue through said optical device and spectral imager and obtaining a spectrum of light for each pixel of the eye tissue; and (c) displaying a spectrum associated with said region.