Patent Application
20140198315
Spectroscopic imaging combines digital imaging and optical spectroscopy techniques, which can include Raman scattering, fluorescence, photoluminescence, laser induced breakdown, ultraviolet, visible and infrared absorption spectroscopies. When applied to the chemical analysis of materials, spectroscopic imaging is also referred to as hyperspectral imaging or chemical imaging. Instruments for performing spectroscopic (i.e. chemical) imaging typically comprise an illumination source, image gathering optics, focal plane array (FPA) imaging detectors and imaging spectrometers.
In general, the size or accessibility of a sample determines the choice of image gathering optic. For example, a microscope is typically employed for the analysis of sub-micron to millimeter spatial dimension samples. For larger objects, in the range of millimeter to meter dimensions, macro lens optics are appropriate. For samples located within relatively inaccessible environments, flexible fiberscope or rigid borescopes can be employed. For very large scale objects, such as planetary objects, or for objects located at a significant stand-off distance from a sensor, telescopes are appropriate image gathering optics.
Two-dimensional, imaging FPA detectors are typically employed to detect images formed by the various optical systems. The choice of FPA detector is governed by the spectroscopic technique employed to characterize the sample of interest. For example, silicon (Si) charge-coupled device (CCD) detectors or complementary metal-oxide-semiconductor (CMOS) detectors are typically employed with visible wavelength fluorescence and Raman spectroscopic imaging systems, while indium gallium arsenide (InGaAs) FPA detectors are typically employed with near infrared spectroscopic imaging systems.
Conventional spectroscopic devices operate over a limited range of wavelengths due to the operation ranges of the detectors or imaging spectrometers possible. This enables analysis in the ultraviolet (UV), visible (VIS), near infrared (NIR), short wave infrared (SWIR), mid infrared (MIR), and long wave infrared (LWIR) wavelengths, as well as some overlapping ranges. These correspond to wavelengths of about 180-380 nm (UV), about 380-700 nm (VIS), about 700-2500 nm (NIR), about 850-1700 nm (SWIR), about 700-1700 (VIS-NIR), about 2500-5000 nm (MIR), and about 5000-25000 (LWIR).
Spectroscopic imaging of a sample is commonly implemented by one of two methods. First, point-source illumination can be used on a sample to measure the spectra at each point of the illuminated area. Second, spectra can be collected over the entire area encompassing a sample simultaneously using an electronically tunable optical imaging filter such as an acousto-optic tunable filter (AOTF), a liquid crystal tunable filter (LCTF), or a multi-conjugate tunable filter (MCF, which is a type of LCTF). Here, the organic material in such optical filters is actively aligned by applied voltages to produce the desired bandpass and transmission function. In hyperspectral imaging (HSI), the spectra obtained for each pixel of an image forms a complex data set referred to as a hyperspectral image. Hyperspectral images may contain the intensity values at numerous wavelengths or the wavelength dependence of each pixel element in the image. Multivariate routines, such as chemometric techniques, may be used to convert spectra to classifications.
A LCTF uses birefringent retarders to distribute the light energy of an input light signal over a range of polarization states. The polarization state of light emerging at the output of the LCTF is caused to vary as a function of wavelength due to differential retardation of orthogonal components of the light, contributed by the birefringent retarders. The LCTF discriminates for wavelength-specific polarization using a polarizing filter at the output. The polarizing filter passes the light components in the output that are rotationally aligned to the polarizing filter. The LCTF is tuned by adjusting the birefringence of the retarders so that a specific discrimination wavelength emerges in a plane polarized state, aligned to the output polarizing filter. Other wavelengths that emerge in other polarization states and/or alignments are attenuated.
A highly discriminating spectral filter is possible using a sequence of several birefringent retarders. The thicknesses, birefringences, and relative rotation angles of the retarders are chosen to correspond to the discrimination wavelength. More specifically, the input light signal to the filter becomes separated into orthogonal vector components, parallel to the respective ordinary and extraordinary axes of each birefringent retarder when encountered along the light transmission path through the filter. These separated vector components are differentially retarded due to the birefringence; such differential retardation also amounts to a change in their polarization state. For a plane polarized component at the input to the filter, having a specific rotational alignment at the input to the filter and at specific discrimination wavelengths, the light components that have been divided and subdivided all emerge from the filter in the same polarization state and alignment, namely plane polarized and in alignment with the selection polarizer (i.e., the polarizing filter) at the output.
A filter as described is sometimes termed a birefringent interference filter because the components that have been divided and subdivided from the input and interfere positively at the output selection polarizer are the components that are passed. Such filters also are sometimes described with respect to a rotational twist in the plane polarization alignment of the discriminated component between the input and the selection polarizer at the output.
There are several known configurations of spectral filters comprising birefringent retarders, such as the Lyot, Solc and Evans types. Such filters can be constructed with fixed (non-tunable) birefringent crystals for the retarders. A filter with retarders that are tuned in unison permits adjustment of the bandpass wavelength. Tunable retarders can comprise liquid crystals or composite retarder elements each comprising a fixed crystal and an optically aligned liquid crystal.
The thicknesses, birefringences, and rotation angles of the retarders are coordinated such that each retarder contributes part of the necessary change in polarization state to alter the polarization state of the passband wavelength from an input reference angle to an output reference angle. The input reference angle may be, for example, 45° to the ordinary and extraordinary axes of a first retarder in the filter. The output reference angle is the rotational alignment of the polarizing filter (or selection polarizer).
A spectral filter may have a comb-shaped transmission characteristic. Increasing or decreasing the birefringence when tuning to select the discrimination wavelength (or passband), stretches or compresses the comb shape of the transmission characteristic along the wavelength coordinate axis.
If the input light is randomly polarized, the portion that is spectrally filtered is limited to the vector components of the input wavelengths that are parallel to one of the two orthogonal polarization components that are present. Only light at the specific wavelength, and at a given reference polarization alignment at the input, can emerge with a polarization angle aligned to the rotational alignment of the selection polarizer at the output. The light energy that is orthogonal to the reference alignment at the input, including light at the passband wavelength, is substantially blocked.
A LCTF thus passes only one of two orthogonal components of input light. The transmission ratio in the passband is at a maximum for incident light at the input to the LCTF that is aligned to a reference angle of the LCTF. Transmission is at minimum for incident light energy at the input is orthogonal to that reference angle. If the input light in the passband is randomly polarized, the best possible transmission ratio in the passband is fifty percent. It is therefore desirable to devise a system and method wherein both orthogonal components of the input light are allowed to transmit through the tunable filter, thereby effectively doubling the throughput at the filter output.
Even using a dual polarization approach, each LCTF is limited to single bandpass, low throughput operation. Therefore, multiple, discrete bandpass measurements are required for analyte discrimination. The need for multiple measurements increases the overall measurement time.
Multivariate Optical Computing (MOC) is an approach which utilizes a compressive sensing device (e.g. an optical computer) to analyze spectroscopic data as it is collected. Other approaches utilize hard coated optical computing filters such as Multivariate Optical Elements (MOEs). MOEs are application-specific optical thin film filters that are used in transmission and reflectance modes. Thin film filters are interference filters with very thin structured layers of different materials with thicknesses on the order of the wavelengths of visible light (for example, 500 nm). Layers at this scale can have remarkable reflective properties due to the light wave interference and the difference in refractive index between the layers, the air, and the substrate. These effects alter the way the optic reflects and transmits light, an effect known as thin film interference. In manufacturing, thin film layers can be achieved through the deposition of one or more thin layers of material onto a substrate. This can be done using a physical vapor deposition process, such as evaporation or sputter deposition, or a chemical process such as chemical vapor deposition. The radiometric response of a MOE-based instrument is proportional to the intended analyte in an associated matrix.
Compressive sensing is the process in which a fully resolved waveform or image is reconstructed from a small set of sparse measurements. A sparse sample implies a waveform or image data set with coefficients close to or equal to zero. Compressive sensing utilizes the redundancy in information across the sampled signal similar to lossy compression algorithms utilized for digital data storage. A fully expanded data set may be created through the solution of an undetermined linear system, an equation where the compressive measurements collected are smaller than the size of the original waveform or image. Compressive measurements can ultimately lead to expedited HSI data collections while still preserving most of the original spectroscopic and spatial information.
While compressive sensing holds potential for decreasing measurement time, the use of MOEs have limitations. For example, MOEs are fixed and lack flexibility for adapting to different analytes. There exists a need for an adaptable filter that can be used to detect a wide variety of analytes while reducing overall measurement time. It would be beneficial if a plurality of such filters could be arranged in a dual polarization configuration to further increase speed of analysis and also provide for assessing multiple analytes simultaneously.
The system and method of the present disclosure overcome the limitations of the prior art by providing a flexible, adaptable filter that can be configured to assess a wide variety of different analytes. A conformal filter is an adaptable filter with the flexibility of conforming to a specific, broadband spectral feature (e.g. pattern or shape). A conformal filter simultaneously transmits multiple passbands that improve discrimination performance for analytes (e.g., discriminating between a target analyte and background), by increasing the throughput of a tunable filter and by increasing the speed of analysis. Conformal filters based on LCTF technology provide an electro-optical filter, with no mechanical moving parts, that collects all wavelengths of light simultaneously, making it unnecessary to continuously tune. Therefore, they can be rapidly tuned with automatable random accessing under computer control. There is no need for iterative tuning to capture all measurements (all wavelengths). The system of the present disclosure holds potential for dramatically improving the speed, performance, and agility of liquid crystal based HSI systems.
Each configuration of a conformal filter is designed to filter interacted photons conforming to at least one spectral shape associated with an analyte of interest. The conformal filter may be configured to operate in conjunction with at least one look-up table (LUT), providing flexibility for detecting multiple analytes of interest in near real-time. The LUT may comprise at least one voltage associated with each stage of the tunable filter. Each voltage is configured to cause the tunable filter to conform to a spectral shape associated with the analyte when applied to the associated stage.
The spectral shape of the conformal filter may be visualized as a traditional multivariate regression vector (e.g. Partial Least Squares—PLS) but is not limited to this in practice. The optical computation occurs by convolving the transmission profile of the conformal filter with the incident radiation from the target and summing the result onto a broadband optical imaging detector. Conformal filter HSI realizes both a Jacquinot and multiplexing advantage over traditional dispersive spectroscopy and discrete bandpass HSI in addition to a potential order of magnitude decrease in measurement time. It is this encoded, multivariate spectral pattern that provides the conformal filter HSI superior specificity over the conventional operation mode of rapidly tuning through a subset of discrete bands.
The present disclosure provides for a system comprising a first optical element, configured to separate a plurality of interacted photons generated from a sample into at least a first optical component and a second optical component. The system may comprise at least one conformal filter assembly, wherein each conformal filter assembly is configured to filter the first optical component and the second optical component conforming to at least one spectral shape associated with an analyte of interest. Each conformal filter assembly may comprise a first conformal filter configured to filter the first optical component and a second optical component configured to filter the second optical component. At least one detector may be configured to detect at least one of the first filtered optical component and the second filtered optical component and generate at least one of a first data set and a second data set.
The present disclosure also provides for a method for assessing characteristics of analytes in a sample. The method may comprise collecting a plurality of interacted photons from a sample and separating the plurality of interacted photons into at least a first optical component and a second optical component. The first optical component and second optical component may each be passed through at least one conformal filter. Each conformal filter may comprise a tunable filter configured to filter an optical component conforming to at least one spectral shape associated with an analyte of interest. A first data set and a second data set may be generated corresponding to the first optical component and the second optical component. At least one optical computation may be applied to at least one of the first data set and the second data set to assess the sample for at least one characteristic of the analyte.
The present disclosure also provides for a system comprising a processor and a non-transitory processor-readable storage medium in operable communication with the processor, wherein the storage medium contains one or more programming instructions that, when executed, cause the processor to perform the following: collect a plurality of interacted photons from a sample, separate the plurality of interacted photons into at least a first optical component and a second optical component, pass the first optical component through at least one conformal filter, pass the second optical component through at least one conformal filter, generate a first data set corresponding to the first filtered optical component, generate a second data set corresponding to the second filtered optical component, and apply at least one optical computation to the first data set and the second data set to assess the sample for at least one characteristic of an analyte.
A system of the present disclosure may comprise a pixilated conformal filter (PCF). The PCF approach takes the light reflected, scattered, transmitted, and/or emitted from a surface of interest and using a lens array, distributes the reflected image over a plurality of conformal filter assemblies. Using another lens array, the system may focus the filtered images onto one or more detectors, such as FPAs, for capture. The two-dimensional FPAs provide spatial information associated with the sample. Each pair of images produced represents a positive and negative portion of a regression vector that may be processed mathematically. Each conformal filter pair may represent an analyte of interest. Because of the inherent polarized operation of liquid crystal filters, additional “pixels” can be obtained without further complicating the primary lens array by employing a polarization beamsplitter after the primary array and orienting the polarization axes of the LC conformal filters to split polarizations. This configuration holds potential for assessing a plurality of analytes simultaneously.
The accompanying drawings, which are included to provide further understanding of the disclosure and are incorporated in and constitute a part of this specification illustrate embodiments of the disclosure, and together with the description, serve to explain the principles of the disclosure.
In the Drawings:
Reference will now be made in detail to the embodiments of the present disclosure, examples of which are illustrated in the accompanying drawings. Wherever possible, the same reference numbers will be used throughout the specification to refer to the same or like parts.
The present disclosure provides for a system and method for assessing characteristics of analytes in a sample using conformal filters. In one embodiment, the conformal filters may comprise tunable filters, such as LCTFs. LCTFs are optical filters that use electronically-controlled liquid crystal elements to transmit a selectable wavelength, or plurality of wavelengths, and exclude others. Often, the basic working principle of LCTFs is based on the Lyot filter which consists of fixed retardance birefringence elements, fixed wave plates, and polarizers. A key difference between a Lyot filter and a LCTF is that the fixed wave plates are replaced by tunable liquid crystal wave plates in the LCTF. Designs other than the Lyot design may also be used. Examples of tunable filters that may be configured for use as a conformal filter may include at least one of: a LCTF, an AOTF, a Lyot LCTF, an Evans Split-Element LCTF, a Solc LCTF, a Ferroelectric LCTF, and a Fabry Perot LCTF.
In one embodiment, the tunable filter may comprise a MCF. A MCF is an imaging filter with serial stages along an optical signal path in a Solc filter configuration. Angularly distributed retarder elements of equal birefringence are stacked in each stage, with a polarizer between stages. The retarders can include tunable (such as abutted liquid crystals tuned in unison), fixed and/or combined tunable and fixed birefringences. In one embodiment, quartz retarders may be used. Although the retardations are equal within each stage, distinctly different retardations may be used for two or more different stages. This causes some stages to pass narrow bandpass peaks and other stages to have widely spaced bandpass peaks. The transmission functions of the serial stages are superimposed with selected tunable peaks coinciding. The resulting conjugate filter has a high finesse ratio and good out of band rejection. In one embodiment, the MCF may comprise filter technology available from ChemImage Corporation, Pittsburgh, Pa.
In
In another embodiment, the present disclosure provides for a conformal filter comprising a modified MCF. In such an embodiment, a tunable filter may be modified or specifically designed so that selected individual stages of a traditional tunable filter comprise multiple, lower resolution liquid crystal cells. As illustrated by
The present disclosure contemplates that other configurations may be used to modify the MCF and that the present disclosure is not intended to be limited to the design in
The present disclosure also provides for a system comprising two or more conformal filters in a dual polarization configuration, exemplary configurations of which are illustrated in
The embodiment of
In this arrangement, the paths of the filtered beams are not parallel through the conformal filters 242a, 242b, but are directed by appropriate reflectors (e.g., mirrors) 276a, 276b to a second optical element 278 (which may be a beam combiner, a polarizing cube or polarizing beam splitter as illustrated) at which the orthogonal components, which can be at the same or different spectral shapes ω1 and ω2. In one embodiment, the components may be combined and directed to a detector 260 through a lens assembly 250. In another embodiment, the components may be kept separate as they are directed to the detector 260. However, the beam paths from one optical element 272 to the other 278 (via individual conformal filters 242a, 242b) may be made symmetrical to avoid, for example, the need for infinitely-corrected optics.
In
In
In the embodiment of
In one embodiment, the two filtered signals may be detected simultaneously. As discussed herein, simultaneous detection of two different spectral shapes holds potential for real-time detection when displayed in a non-overlapping configuration (side-by-side, top to bottom, etc.). In another embodiment, the two filtered signals may be detected sequentially.
It is noted here that although laser light may be coherent, the light received from the sample 230 (e.g., light emitted, scattered, absorbed, and/or reflected) and fed to the conformal filters 242a, 242b may not be coherent. Therefore, wavefront errors may not be present or may be substantially avoided in the two conformal filters versions in
The present disclosure also provides for a system comprising a PCF configuration. A PCF approach extends the idea of a traditional conformal filter approach, essentially creating an “array” of conformal filters, with each “pixel” in the array corresponding to a particular confirmation. This means that each pixel in the filter has a separately encoded, multivariate spectral pattern, specific to a given analyte. A PCF approach allows for the detection of multiple analytes simultaneously, where two conformal filters are needed for assessing each analyte (one corresponding to a positive regression vector and one corresponding to a negative regression vector). Using a dual polarization configuration, as described in reference to
One embodiment is illustrated by
The embodiment in
The present disclosure contemplates a conformal filter may be incorporated into one or more HSI systems using either the dual polarization or PCF configurations described herein. Examples of HSI modalities which may incorporate the system and method described herein may include, but are not limited to: Raman, UV, fluorescence, VIS, and infrared (including SWIR, MWIR, and LWIR). In one embodiment, at least one HSI modality may be used to scan an area to identify regions of interest for further interrogation using one or more conformal filters.
The present disclosure also provides for a method for assessing characteristics of analytes. In one embodiment, illustrated by
In step 320 the plurality of interacted photons may be separated into a least a first optical component and a second optical component. In one embodiment, one or more lenses may be used to orient the plurality of interacted photons into culminated space. The first optical component may be passed through at least one conformal filter in step 330 and a second optical component may be passed through at least one conformal filter in step 340. Each conformal filter may comprise a tunable filter configured to filter an optical component conforming to at least one spectral shape associated with a target of interest. In one embodiment, a first conformal filter may be configured to generate a positive regression vector associated with the spectral shape, while a second conformal filter may be configured to generate a negative regression vector associated with the spectral shape.
In step 350 a first data set may be generated corresponding to the first filtered optical component. In step 360 a second data set may be generated corresponding to the second filtered optical component. In one embodiment, at least one of the first and second data sets may further comprise at least one of: a spectral intensity of the sample and a spectral image representative of the sample. In one embodiment, the first test data set and the second test data set may be generated by the same detector. In another embodiment, the first data set and the second data set maybe generated by two different detectors. The present disclosure contemplates embodiments where detecting the first and second data set may be simultaneous or sequential and the data sets may be displayed in an overlaid configuration or a configuration other than overlaid.
At least one optical computation may be applied in step 370 to the first and second data set to assess the sample for at least one characteristic of the analyte. The optical computation may be any known in the art including, but not limited to: wavelength addition, wavelength subtraction, wavelength multiplication, and wavelength division and may also be referred to as a mathematical computation. The result of applying the optical computation may be a detection image, indicative of one or more characteristics of each analyte under analysis. Examples of characteristics that may be assessed may include, but are not limited to, at least one of: the presence of the analyte in the sample, the absence of the analyte in the sample, a classification of the analyte, a non-classification of the analyte, and a concentration of the analyte.
In one embodiment, the configuration of at least one conformal filter may be determined by searching at least one LUT corresponding to an analyte of interest, wherein the LUT comprises at least one voltage associated with each stage of the tunable filter, and wherein each voltage is configured to cause the tunable filter to conform to a spectral shape associated with the analyte of interest when applied to the associated stage.
The present disclosure also provides for a method for selecting a conformal filter configuration using an iterative process. This method is referred to herein as Real-time Contrast Enhancement (RtCE) and provides for configurations with high analyte specificity and sensitivity by applying active tunable filter voltage adjustment and feedback from a live measurement scene. Such an approach may be used to calibrate a conformal design for an analyte of interest, refine a previous conformal filter design for an analyte of interest, and/or generate a new conformal filter design for an analyte of interest.
One embodiment of this optimization process is provided in
In another embodiment, at least one reference data set may comprise at least one reference image comprising the sample and the matrix. A first region of interest may be selected corresponding to the sample and a second region of interest may be selected corresponding to the matrix. Spectral data may be extracted from these regions of interest.
In one embodiment, at least one chemometric technique may be applied to the at least one reference data set (e.g. spectral data). Examples of chemometric techniques include, but are not limited to: correlation analysis, principle component analysis, principle component regression, partial least squares, multivariate curve resolution, Mahalanobis distance, Euclidian distance, band target entropy, band target energy minimization, partial least squares discriminant analysis, adaptive subspace detection, and combinations thereof. Chemometric techniques may be used to compare test data to reference data.
One or more optical computations may also be applied to the test data set. In addition to wavelength addition, wavelength subtraction, wavelength multiplication, and wavelength division, this optical computation may comprise at least one of the following: T1, and (T1−T2)/(T1+T2), among others known in the art.
A determination of whether or not a tolerance level is met may be made in step 430. In one embodiment, this determination may comprise applying at least one Figure of Merit (FOM) A FOM is a numerical value that may be used to guide the optimization process. Examples of figures of merit that may be applied include, but are not limited to: Standard error of calibration (SEC), Euclidian Distance, standard error of prediction (SEP), 1-Area Under the Receiver Operator Characteristic Curve (AUROC), optical throughput (% T), and combinations thereof. Other FOMs may be used that incorporate optical throughput, signal to noise ratio (SNR), among others. If a tolerance level is met 440a, then a test data set representative of the sample may be generated in step 450. If a tolerance level is not met 440b, then the process may be repeated for at least one other conformal filter configuration until a tolerance level is met. In one embodiment, the present disclosure contemplates the method of
The present disclosure also provides for a system comprising a processor and a non-transitory processor-readable storage medium in operable communication with the processor, wherein the storage medium may contain one or more programming instructions that, when executed, cause the processor to perform the following: collect a plurality of interacted photons from a sample, separate the plurality of interacted photons into at least a first optical component and a second optical component, pass the first optical component through at least one conformal filter, pass the second optical component through at least one conformal filter, generate a first data set associated with the first filtered optical component and a second data set associated with a second filtered optical component, and apply at least one optical computation to the first data set and the second data set to assess the sample for at least one characteristic of the analyte. The storage medium may further contain programming instructions that cause the processor to select conformal filter configurations by searching a LUT corresponding to an analyte and applying the configuration to the conformal filter.
In another embodiment, the system may further comprise one or more programming instructions that, when executed, cause the processor to iteratively configure the conformal filter until a tolerance level is met. In such an embodiment, the instructions may cause the processor to tune the conformal filter to a first configuration to filter interacted photons conforming to at least one spectral shape associated with the analyte, generate at least one reference data set comprising reference data corresponding to the sample and reference data corresponding to a matrix, and determine whether or not a tolerance level is met. If a tolerance level is met, a test data set may be generated. If a tolerance level is not met, then the steps may be repeated for one or more difference configurations until a tolerance level is met. In one embodiment, whether or not a tolerance level is met may be determined by the processor applying at least one figure of merit. In other embodiments the processor may further analyze the test data set by applying at least one of the following: an optical computation and a chemometric technique.
Three samples were prepared comprising AS, AN, and urea. AN was selected as the analyte of interest, AS was selected as a confusant (background), and urea was selected as an interferent. The samples were analyzed using an experimental set up wherein the illumination source comprised a quartz tungsten halogen lamp, the conformal filter comprised a MCF, and the detector comprised a SWIR camera. A brightfield reflectance image 510 and a SWIR reflectance image (T1) 530 were generated. Spectral data for each substance 505 is also illustrated in
While the disclosure has been described in detail in reference to specific embodiments thereof, it will be apparent to one skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the embodiments. Additionally, while the examples provided herein relate to specific analytes, the present disclosure is not limited to these analytes and may be used to detect a wide variety of analytes of interest. Thus, it is intended that the present disclosure cover the modifications and variations of this disclosure provided they come within the scope of the appended claims and their equivalents.
This Application claims priority under 35 U.S.C. §119(e) to the following pending U.S. Provisional Patent Applications: No. 61/752,817, filed on Jan. 15, 2013, entitled “Pixilated Conformal Hyperspectral Imaging System and Method for Use Thereof,” and No. 61/799,291, filed on Mar. 15, 2013, entitled “System and Method for Dual Polarization Using Conformal Filters.” These applications are hereby incorporated by reference in their entireties.
| Number | Date | Country | |
|---|---|---|---|
| 61752817 | Jan 2013 | US | |
| 61799291 | Mar 2013 | US |
