Multifocal hyperspectral Raman system and methods for imaging of materials

Abstract

A hyperspectral Raman imaging system having the ability to focus on excitation laser beam over a relatively wide field of view due to the use of a lens array, in particular a microlens array. Hyperspectral selection is provided in one embodiment through the use of dual-axis controlled dielectric filtration. Methods for analyzing materials with the system are disclosed. The device or system can be used in generally any application where investigation of materials is required.

Claims

1. A Raman imaging system, comprising: an excitation source providing a beam; and a lens array located downstream from the excitation source in a beam path and including a plurality of lenses that split the beam into a plurality of excitation spots that are spread over an area larger than the area of the beam of the excitation source, wherein the excitation spots are adapted to be routed onto a sample, wherein a Raman signal received from the sample is routed to a charge-coupled device located downstream from the lens array.

2. The Raman imaging system according to claim 1, wherein the excitation source is a monochromatic laser having a wavelength from about 250 to 1064 nm.

3. The Raman imaging system according to claim 2, wherein the lens array is a microlens array, wherein a dichroic mirror is located downstream from the microlens array and able to route the excitation spots through a focusing lens that is adapted to focus the plurality of beams onto the sample, wherein the resulting Raman signal is reroutable through the dichroic mirror.

4. The Raman imaging system according to claim 3, further including a tunable band-pass filter that is located downstream from the dichroic mirror and upstream from the charged-coupled device.

5. The Raman imaging system according to claim 4, wherein a flip beam splitter is present downstream from the tunable band-pass filter and can be switched to route the Raman signal to either the charged-coupled device or a spectrometer also located downstream from the flip beam splitter.

6. The Raman imaging system according to claim 5, wherein the tunable band pass filter is rotatable to scan a range of wavelength numbers, also enabling acquisition of a continuous spectrum at each pixel.

7. The Raman imaging system according to claim 6, wherein a laser line filter is located between the excitation source and the microlens array in a beam path of the Raman imaging system.

8. The Raman imaging system according to claim 1, wherein the lens array increases a cross sectional area of the beam to an area that is at least 25% larger than an area of the beam incident on a first lens of the plurality of lenses.

9. The Raman imaging system according to claim 8, wherein the area increase is at least 50%.

10. A method for non-flat surface correction of Raman images using the Raman imaging system according to claim 1, comprising the steps of: obtaining a Raman image with the Raman imaging system of claim 1; obtaining a reference image; dividing the Raman image by the reference image to normalize for relative variations and spacial intensity, which can be expressed as $I_{\mathrm{corr}}=\frac{I_{\mathrm{adjusted}\mathrm{raw}}}{I_{\mathrm{ref},}}$ where I.sub.adjusted raw is an intensity matrix of the raw image which can be adjusted by one or more of smoothing, filtering and subjecting to another form of numerical processing, I.sub.ref is an intensity matrix of the reference image, and I.sub.corr is an intensity variance corrected final image.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The invention will be better understood and other features and advantages will become apparent by reading the detailed description of the invention, taken together with the drawings, wherein:

(2) FIG. 1 is a schematic illustrating A) one example of classical excitation with a single focal point at the center and B) that the introduction of a plurality of lenses, in particular in the form of a microlens array (MLA) in the excitation path of a Raman imaging system that splits a beam received from the excitation source and produces a plurality of excitation spots that are spread over an area larger than the area of the beam of the excitation source, with the larger area preferably covering the full field of view;

(3) FIG. 2 is a schematic view of one embodiment of the components and layout of a wide-field Raman imaging system of the present invention utilizing dual-axis controlled dielectric filtration, wherein LP is a telescope lens pair, LL is a laser line filter, CL is a cylindrical lens, FL is a focusing lens, DM is a dichroic mirror, EF is an edge filter, TBF is a tunable band-pass filter wherein the filter can be rotated to scan a range of wavelength numbers, also enabling the acquisition of a continuous spectrum at each pixel (hyperspectral Raman analysis), FBS is a flip beam splitter and L is a lens;

(4) FIG. 3 is a schematic view of one embodiment of the components and layout of a wide-field Raman imaging system of the present invention utilizing a lens array, wherein MLA is a microlens array, TBF is a tunable bandpass filter, DM is a dichroic mirror, NF is a notch filter, LL is a laser line filter, FL is a focusing lens, L is a lens, CL is a cylindrical lens, LP is a telescope lens pair, FBS is a flip beam splitter, P is a linear polarizer, HWP is a half-wave plate, PA is a polarization analyzer, and PS is a polarization scrambler;

(5) FIG. 4 illustrates one embodiment of a kinematic dielectric wavenumber filtering system of another innovative aspect of the present invention that enables hyperspectral imaging, wherein in one embodiment a motorized wheel can accommodate 1 to 10 band-pass filters which can be switched controllably to cover different wavenumbers, wherein the azimuthal rotation enables wavelength tuning to cover a range near or around a given wavenumber;

(6) FIG. 5 illustrates A) tooth specimen was sectioned transversely to expose enamel-dentin junction at which the mineral content increases spontaneously, B) Raman point map of mineral variation at dentin (blue, low mineralization) and enamel (red, high mineralization) junction that lasted for 2 hours, C) single spot global Raman image of similar region that lasted for 1 minute, wherein however, meaningful information was limited to the center of the field of view. Intensity plots are the vertical red dashed lines;

(7) FIG. 6 illustrates A) global Raman image of an enamel-dentin junction B) Image in A was normalized by a reference image reflecting the intensity variations of excitation over the field of view. The normalization resulted in the recovery of information from a greater fraction of the field of view;

(8) FIG. 7 illustrates another embodiment of a multi-lens array-containing Raman imaging system of the invention including A) an overview and B) the following components 1: Laser, 2: Lens pair or beam expander, 3: laser filter, 4: microlens array, 5: dichroic mirror, 6: edge filter, 7: tunable kinematic band-pass filter, 8: flip mirror, 9: lens, 10: mirror, 11: spectrometer, 12: 2D-CCD camera, 13: objectives/lenses; and

(9) FIG. 8 illustrates wide-field Raman images (cylindrical lens engaged) using 25 mm lens (A) and 50 mm lens (B). The area of coverage (red dashed borders) was quadrupled by using lower magnification lenses. Other lenses can be used, including, but not limited to, concave and convex lenses, as well as other forms.

DETAILED DESCRIPTION OF THE INVENTION

(10) As indicated herein, one embodiment of the present invention discloses a multifocal hyperspectral Raman imaging system. As illustrated in FIGS. 1, 3 and 7, in one embodiment the excitation scheme employs a lens array including a plurality of lenses, preferably a microlens array, to spread the excitation spots over a desired, large or full field of view. Thus, the plurality of lenses, e.g. microlens array, in a beam path increase an area (cross-section that is perpendicular to the beam path) of a beam generated by an excitation source from an initial area to a larger area, wherein the larger area is generally at least 25%, desirably at least 50%, and preferably at least 90% larger than the area of the beam incident on a first lens of the plurality of lenses or on the microlens array. In one embodiment, direct imaging of areas, for example up to 55 mm, an order of magnitude of better coverage than known existing systems is possible with high power lasers. Discretization of excitation reduces the risk of damage to the sample; while enabling the utilization of a high power laser.

(11) A series of dielectric band-pass filters can be used to isolate the wavenumbers of interest on the Raman signal path. It is optionally possible to scan a wavenumber range using the computer controlled kinematic wavenumber filters, enabling the acquisition of hyperspectral images, see FIGS. 2 and 7 for example.

(12) Variations in illumination over the space continuum (due to distortions introduced by optical components and surface curvature of samples) can be corrected by reference images. Software can streamline the registration of the reference and Raman image sequences and process them to generate corrected images.

(13) With respect to hardware, a suitable excitation source is chosen, preferably a monochromatic laser having a wavelength of from about 250 to about 1064 nm. In the examples set forth hereinbelow a 785 nm laser was utilized. A 532 nm wavelength can also be utilized in other embodiments which results in a reduced risk of damage to a sample. Lasers having a higher power, such as 1064 nanometers can also be utilized and are preferred in some embodiments in order to reduce background fluorescence. A negative cylindrical lens, for example f=200 mm, can be used to defocalize the laser spot onto a rectangular profile on the sample. Other lenses can be used so that the field of view can be expanded.

(14) A telescope lens pair is located downstream from the laser. Placed thereafter is a laser line filter and a cylindrical lens. After passing therethrough, the light encounters a dichroic mirror and subsequently is passed or routed through a focusing lens onto the sample. The scattered Raman signal from the sample is then routed through an edge filter, a tunable band-pass filter to a flip beam splitter which can be used to route the signal to the spectrometer and also be flipped off during the image acquisition so that a full Raman signal can be delivered to a charge-coupled device or CCD camera. As illustrated in FIG. 3, a lens array including a plurality of lenses, such as a microlens array is placed in the excitation path to spread the excitation spots over a full or desired field of view.

(15) In one embodiment, a microlens array, such as available from Thorlabs as model MLA150-7AR-M or SUSS MicroOptics of Switzerland as model number 18-00232, can be utilized. The microlens array discretizes the laser power to an array of spots which in turn spreads the laser power over the desired or full field of view. As mentioned above, this results in reduced risk of damage to the sample while enabling the utilization of high power laser.

(16) The embodiment set forth in FIG. 3 also includes a linear polarizer, half-wave plate, notch filter, a polarization analyzer, a polarization scrambler and a tunable bandpass filter. Depending on the mode of data collection, such as Raman spectrometer, wide field at 2D CCD or polarized analysis, some of the optics can be removed or even other optics can be added, as known to those of ordinary skill in the art.

(17) As described hereinabove, in one embodiment the tunable bandpass filter provides this system with hyperspectral imaging. Hyperspectral imaging allows use of the Raman imaging system with a broad range of applications which cannot be met by a set number of wave numbers. The system includes a bandpass filter which allows a 4D imaging modality where two dimensions are the axes of the imaging plane. One dimension is associated with the wavenumber and the temporal acquisition is the fourth dimension. In this option, a given dielectric filter is rotated azimuthally, see FIG. 4, to cover a wavenumber range of 250 cm.sup.1 about the central wavenumber of the dielectric filter. For instance, covering a range of 200-1200 cm.sup.1 requires two dielectric filters which are centered at 400 cm-1 and 950 cm.sup.1. The filter wheel programmed to engage the filter #1, rotate filter #1 azimuthally, engage the filter #2, rotate filter #2 azimuthally, to cover the said range. In this fashion, a range up to 2700 cm.sup.1 is covered.

(18) A spectrometer is used after the dielectric filters on the signal collection path to enable the tuning of kinematic dielectric band-pass filter to the wavenumber of interest; and, to collect full wavenumber range Raman spectrum of the same region to validate the wide-field image. A flip beam splitter is used to switch between the spectrometer and the imaging camera. The splitter can be manually switched by a mechanical knob that is accessible from outside the housing in one embodiment. Alternative to the spectrometer, tuning and calibration of the kinematic band-pass filter can be accomplished by providing light sources at known wavelengths by using tungsten halogen lamp as a source and a set of bandpass filter set at various wavelengths.

(19) The following results illustrate that multilens arrays can be used to excite the desired or full field of view as a series of discrete points and adjustability of the field of view coverage.

(20) There is a stepwise increase in the mineral content at the junction of enamel (highly mineralized) and dentin (partially mineralized) regions of tooth. This predictable increase in mineral content is a good test case to assess performance of chemical imaging methods. The enamel dentin junction was exposed by cutting a section perpendicular to the longer axis of the tooth (FIG. 5A).

(21) Hardware

(22) The excitation source was a 785 nm laser (50 mW on sample) (FIG. 2). A negative cylindrical lens (f=200 mm) was used to defocalize the laser spot into a rectangular profile on the sample. Other lenses were also used to demonstrate that the field of view can be expanded. An edge notch filter at 785 nm suppressed the residual excitation.

(23) A dielectric filter was used to retain the wavenumber of interest corresponding to mineral Raman signal (at 960 cm.sup.1 originating from phosphate groups of mineral). The dielectric filter can be rotated to scan a range of wavenumbers, also enabling the acquisition of a continuous spectrum at each pixel (hyperspectral Raman analysis). For this particular application, acquisition at a fixed wavenumber is sufficient to get the information on mineral content. The global Raman image was captured by a near-infrared CCD camera (Ikon-M 934, loaned by Andor, USA). Signal integration time ranged from 60-120 seconds. The flip mirror was used to route the signal to the spectrometer which is used to confirm that the dielectric filter is rotated to an angle that provides the desired Raman wavenumber range. The mirror was flipped off during the image acquisition so that full Raman signal can be delivered to the CCD camera.

(24) The regions observed by the wide-field system were also mapped by a Raman microscope (Xplora, 785 nm, Horiba Jobin Yvon) to confirm the observations of the global Raman imaging system. Images were point mapped at 4040 points with 25 m increment.

(25) Data Analysis

(26) The dark current noise due to deep cooling was acquired pre hoc and autocorrected for every image. Background fluorescence was photobleached until it was stabilized following which the image was acquired. A second image was captured at the base of the mineral peak at 880 cm.sup.1 to correct for background fluorescence by subtracting the background intensity image from the image collected at 960 cm.sup.1 peak of mineral phosphate symmetric stretch band. Outlier points due to cosmic radiation were removed from the background corrected image.

(27) Results

(28) The Raman point map of the enamel-dentin junction (FIG. 5B) by the research grade Raman microscope (Xplora, Horiba Jobin Yvon, NJ) confirmed the expected stepwise transition of mineral content from low mineralization (dentin) to high mineralization (enamel). A similar transition in mineral content was also observed by the global Raman imaging set up (FIG. 5C); however, the information was limited to the center of the field of view.

(29) By using a cylindrical defocusing lens (200 mm), the excitation profile was shaped as an elongated elliptical region to collect signal from an area of 0.10.25 mm.sup.2 (FIGS. 5C and 8A). The area coverage was quadrupled to 0.20.5 mm.sup.2 by using lower magnification lens (FIG. 8B).

(30) The imaging data of the tooth demonstrated that 2-D Raman imaging can: 1) visualize mineral variations; 2) larger area of image can be obtained with lower magnification collection lens.

(31) The insertion of microlens array into the system illustrated in FIG. 2 distributes the laser intensity of the field of view more effectively. Utilization of a high power laser and employment of image normalization routine, see FIG. 6, enables the system to image the entire field of view at faster acquisition times.

(32) FIG. 7 illustrates another embodiment of a multi-lens array-containing Raman imaging system.

(33) It is desirable that the system illustrated in FIG. 7 acquires Raman images over at least a 55 mm.sup.2 area in a timeframe ranging from 0.1 to 10 seconds, depending on the Raman activity of the sample. The sub-units of the Raman imaging system include the excitation path, signal conditioning and acquisition, software interface and mechanical framework for sample positioning.

(34) As described hereinabove, the multilens or microlens array discretizes the laser power to an array of spots which in turn spreads the laser power over the full field of view. This results in reduced risk of damage to the sample, while enabling the utilization of a high power laser. A high power laser can allow delivering 40-400 mW excitation laser light to each spot within the array image. Various available 2D CCD cameras based on silicon chips all have peeks sensitivity at green visible light range, which will provide greater signal collection than other wavelengths at a relatively low cost. Alternatively, other lasers can be utilized as described herein, such as a 1064 nm laser, which is suitable for investigators working on fluorescent specimens.

(35) Beam Conditioning:

(36) Beam shape is conditioned before being reflected on sample space in two regards: 1) expansion of beam to cover a larger field of view, and, 2) multifocal discretization of excitation. The laser beam is expanded to 6 mm for example by using a telescope lens pair. The expanded beam is delivered through a microlens array. Imaging area are expanded to the full range of the field of view of the imaging lens to reflect excitation arrays of 1515 focused spots (FIG. 1). Each spot forms a new image at size comparable to that of the single image before inserting the microlens array (MLA). Expected performance in the spatial domain is listed in Table 1. The resolution was estimated using Abbe diffraction limit: D=0.61/NA, where d is the image diameter and NA is the numerical aperture of the objective lens.

(37) The resulting beam is diverted to the sample by a dichroic mirror and focused on to the sample by objectives/lenses mounted on a rotating turret. Depending on the lens, the resulting illumination area can thus be tailored to cover regions as large as 5 mm5 mm with spacing between each spot in illumination array from 5 m to 250 m (Table 1). Depending on the magnification of the lens, the spatial resolution will vary between 0.4 m to 1.6 m. Depending on the application; the system can be configured at fixed magnification. Alternatively, by using a set of low to high magnification lenses, the system can also be configured as a multiscale imaging setup covering cm to micrometer size scales in one system when desired.

(38) TABLE-US-00001 TABLE 1 Image Image Field of Spatial diameter diameter Image Spot Lens View Resolution w/o MLA with MLA Distance 100X 400 m 0.4 m 1 m 40 m 5 m 10X 500 m 1.2 m 13 m 500 m 30 m 25 mm 1 m 0.7 m 33 m 1 m 100 m (1 aperture) 100 mm 5 m 1.6 m 130 m 5 m 250 m (1 aperture)

(39) Signal Collection and Conditioning

(40) Photons reflected from the samples are collected by the same lens, and pass through the dichroic mirror on the reflection path. An edge (or notch) filter is used to remove the Rayleigh line. Following this stage, the wavenumber (or wavenumber range) of interest is extracted.

(41) In addition to a hyperspectral version of the system, the bandpass filter can, in one embodiment, offer imaging for a limited number of wavenumbers, for example up to about 500 wavenumbers utilizing a bandpass filter that is adjustable on a single axis and contain at least two different filters and for example from 2 to about 5 filters with.

(42) Data Acquisition

(43) The data can be acquired by an ultrasensitive 2D CCD camera, such as Andor Ikon-M 934, whose quantum efficiency is between 90-95% at 400-800 nm range. The camera also features negligible dark current with deep thermoelectric cooling as low as 100 C. This camera is equipped with high resolution sensors having 10241024 active pixels. The dark background, signal integration time, signal averaging and wavenumber scan range can automatically controlled by software, such as LabView software.

(44) Data Processing

(45) Since the Raman signal is proportional to the amount of excitation, variations in illumination over the space continuum need to be corrected. The variation of illumination over the sample stems from: 1) distortions/aberrations introduced by optical components, 2) surface curvature of the sample (when present) that introduces variations in depth of focus. These effects can be simply and expeditiously accounted for by acquisition of a reference image. Dividing the Raman image with the reference image normalizes for the relative variations in spatial intensity, which can be expressed as

(46) $I_{\mathrm{corr}}=\frac{I_{\mathrm{adjusted}\mathrm{raw}}}{I_{\mathrm{ref},}}$
where I.sub.adjusted raw is the intensity matrix of the raw image which can be adjusted by, for example, but not limited to smoothing, filtering or subjected to another form of numerical processing, I.sub.ref is intensity matrix of the reference image, while I.sub.corr is the intensity variance corrected final image.

(47) This reference image correction method is demonstrated in FIG. 6 by using the mineralization. In this case the mineral information is included at 960 cm.sup.1 whereas a second reading taken at 880 cm.sup.1 is used as the reference background. This reference image was used to divide the fluorescence background corrected image. FIG. 6A shows the image before normalization which is limited to less than 10% of the field of view. Following the normalization (FIG. 6B), the enamel/dentin contrast increased and information was recovered from more than 70% of the imaged area.

(48) While in accordance with the patent statutes the best mode and preferred embodiment have been set forth, the scope of the invention is not limited thereto, but rather by the scope of the attached claims.