Figure 2 - uploaded by Jianwei Qin
Content may be subject to copyright.
Source publication
Raman chemical imaging combines Raman spectroscopy and digital imaging to visualize the composition and structure of a target, and it offers great potential for food safety and quality research. In this study, a laboratory-based Raman chemical imaging platform was designed and developed. The system utilizes a 785 nm spectrum-stabilized laser as an...
Contexts in source publication
Context 1
... containing a stack of single band images is built up as the scanning is performed in the spectral domain. No relative movement between the sample and the detector is required for this method. Imaging systems that use filters (e.g., filter wheels containing fixed bandpass filters and electronically tunable filters) belong to the area‐scanning method. The 3‐D hyperspectral cubes acquired by point‐scanning, line‐scanning, and area‐scanning methods are generally stored in the formats of band interleaved by pixel (BIP), band interleaved by line (BIL), and band sequential (BSQ), respec‐ tively. Different formats have different advantages in terms of image processing operations and interactive analysis. The BIP and BSQ formats offer optimal performance for spectral and spatial access of the hyperspectral data, respectively. The BIL format gives a compromise in performance between spa‐ tial and spectral analysis. The three data storage formats are interconvertible. In this study, we developed a laboratory‐based point‐ scanning Raman chemical imaging system with the capacity to acquire hyperspectral Raman images from large food and biological samples. The objective of this article was to pro‐ vide a detailed description of the Raman chemical imaging system. Specific objectives were to: S Present details of system design, including hardware components, interface software, and operation proce‐ dures. S Perform spectral and spatial calibrations for the Raman chemical imaging system. S Demonstrate performance of the system with an exam‐ ple application, i.e., detection of melamine in dry milk. A schematic diagram of the developed Raman chemical imaging system is illustrated in figure 2. The system was built on an optical breadboard with dimensions of 914 × 762 mm 2 . It is a point‐scanning spectral imaging system that utilizes a 16‐bit high‐performance spectroscopic charge‐coupled de‐ vice (CCD) camera (Newton DU920N‐BR‐DD, Andor Technology, Inc., South Windsor, Conn.). The CCD has an area array of 1024 × 256 pixels, with quantum efficiency (QE) greater than 90% at 800 nm and about 45% at 1000 nm. The CCD is thermoelectrically cooled to ‐70 ° C during spectral data acquisition to minimize the dark current. The cam- era is directly connected to a computer using a USB cable, through which both camera control and data transfer are car‐ ried out. A Raman imaging spectrometer that is specifically designed for 785 nm laser excitation (Raman Explorer 785, Headwall Photonics, Fitchburg, Mass.) is mounted to the camera. The basic structure of the spectrometer includes a narrow input slit, a concave mirror, and a convex reflection grating (Julian et al., 2009). The input slit is 5 mm long × 100Ă m m wide and admits light from the samples. The concave mirror is used to guide the incoming light from the input slit to the reflection grating, where the incident beam is dispersed into different wavelengths in a reflection manner. The con‐ cave mirror reflects the dispersed light to the CCD detector, where continuous spectra are formed. The spectrometer cov‐ ers a Raman shift range of ‐98 to 3998 cm ‐1 (or a wavelength range of 779 to 1144 nm) with a spectral resolution of 3.7Ăcm ‐1 . The reflection grating‐based Raman spectrometer is constructed entirely from reflective optics with high effi‐ ciencies. Thus, it is ideal for low‐light measurements such as Raman and fluorescence imaging. A spectrum‐stabilized laser module with an output of 785Ănm and bandwidth of 0.05 nm (I0785MM0350MF‐NL, Innovative Photonic Solutions, Monmouth Junction, N.J.) serves as the excitation source for the system. The 785 nm la‐ ser has a special advantage over short‐wavelength lasers in that it greatly reduces fluorescence emissions from the sam‐ ples. This module is specifically designed to be used in a fiber‐coupled configuration for high‐resolution Raman spec‐ troscopy applications. The maximum output power of the la‐ ser is 350 mW, and the power can be adjusted by dialing the laser drive current as read out on the LED panel. A fiber optic Raman probe (RPB, InPhotonics, Norwood, Mass.) is used to focus the laser on the samples and acquire Raman scattering signals. The probe utilizes a long‐pass filter assembly to eliminate light at and below the laser wavelength (i.e., Ray‐ leigh and anti‐Stokes scattering). A bifurcated optical fiber bundle is used to connect the laser module, the Raman probe, and the Raman imaging spectrometer. Diameters of the ex‐ citation and collection fibers in the bundle are 105 and 200Ă m m, respectively. The laser light generated from the module is transferred to the probe via the excitation fiber. Af‐ ter laser‐sample interactions, the Raman signals are collected by the probe and delivered to the imaging spectrometer via the collection fiber. A programmable, two‐axis motorized positioning table (MAXY4009W1‐S4, Velmex, Bloomfield, N.Y.) is used to move the samples in two perpendicular direc‐ tions ( x and y , as illustrated in fig. 2) under the Raman probe. The movement of the table is controlled by the computer via a stepping motor controller. The table can travel in a square area of 127 × 127 mm 2 with a displacement resolution (ad‐ vance per step) of 6.35 m m. The parameterization and data‐transfer interface software (fig. 3) for the Raman chemical imaging system was devel‐ oped on a platform of LabVIEW and Vision Development Module (National Instruments, Austin, Tex.) in the Microsoft Windows operating system. Software Development Kits (SDKs) provided by the manufacturers of the camera and the motorized positioning table were used in the LabVIEW pro‐ gramming environment to fulfill various functions such as cameral control, data acquisition, sample movement, and synchronization. A Raman spectrum and a Raman image at user‐selected band are displayed and updated point by point to let users monitor the scan progress in real time. The 3‐D Raman image data are stored in the band interleaved by pixel (BIP) format, which can be analyzed by commercial software packages such as ENVI (ITT Visual Information Solutions, Boulder, Colo.). The Raman probe, positioning table, and sample materials are placed in a closed black box (fig. 2) to avoid influence of ambient light. To ensure the best possible signal‐to‐noise ra‐ tio (SNR) for the Raman spectra, single‐track mode provided by the camera's manufacturer is used for spectral data ac‐ quisition. With this method, a rectangular area on the CCD sensor is defined to only include pixels that are illuminated by the incoming light from the Raman imaging spectrometer. All the rows within this specified area are vertically binned together into the shift register of the CCD and then digitized. As a result, a single Raman spectrum is obtained for each spa‐ tial point. Vertical center position (pixel index) and height (inĂpixels) are specified by the system software to determine the CCD area for acquiring spectral data. The single‐track method is particularly useful in low‐level light applications since the contribution of the dark current in the measured sig‐ nal is minimized by excluding the pixels that are not in the defined area. An example of single‐track mode for spectral data acquisition is shown in figure 4. It can be seen that light is dispersed to a narrow area on the CCD. Under the current system settings, a rectangular area with a height of 27 pixels was selected for useful measurement data (i.e., the area be‐ tween two dashed lines in fig. 4a). The highest intensity (CCD count) of the Raman spectra extracted from any single row within the defined CCD area is less than 1500 (fig. 4b). After vertical binning, the intensity of the final Raman spec‐ trum is at least one order higher than those of the individual spectra (fig. 4c), indicating the effectiveness of the single‐ track spectral data acquisition method. Figure 5 illustrates the point‐scanning pattern followed by the Raman chemical imaging system for acquiring spatial in‐ formation from the target. The process is a row‐wise scan starting from a point at the lower right corner of the scene (P11 in fig. 5). After the first row is completed, the sample is moved to focus on the second row, and the scan continues at the first point in this row (i.e., P21). This process is repeated until the last point (i.e., Pnm) is scanned. The spatial resolu‐ tion of the acquired images is determined by the step sizes for two scan directions (i.e., x and y ). Spatial range of the scene is determined by the combination of the step sizes and the numbers of scans (i.e., m and n ) for the x and y directions. Af‐ ter all the points are scanned, the two‐axis positioning table moves the sample back to the start point, and a 3‐D image data with a dimension of m × n × 1024 (1024 bands) will have been obtained. A Raman spectrum is generally presented as a shift in en‐ ergy from that of the excitation source. Although Raman shift is essentially a relative unit, the spectral dimension i.e., x ‐axis) of a Raman spectrum is traditionally expressed as wavenumber in cm ‐1 instead of D cm ‐1 . Spectral calibration for the Raman chemical imaging system is intended to define the wavenumbers for the pixels along the spectral dimension. Spectrally established light sources (e.g., spectral calibration lamps, lasers, fluorescent lamps, and broadband lamps equipped with interference bandpass filters) are usually used for absolute wavelength calibrations for line‐scanning hyper‐ spectral imaging systems. For spectral calibration of relative Raman shift, it is more convenient to use an excitation source with fixed spectral output and chemicals with known relative wavenumber shifts. A guide for Raman shift standards for spectrometer calibration has been established by ASTM In‐ ternational (ASTM Standards, 2007). This guide covers Ra‐ man shift wavenumbers of eight liquid and solid chemicals measured using Fourier transform or dispersive Raman spectrometers with high ...
Context 2
... measurement of oil and water content in olive pomace (Muik et al., 2004), phosphorus sensing in soils (Bogrekci and Lee, 2005), in‐ spection of plant pathogens (Schmilovitch et al., 2005), and detection of melamine in human foods and animal feed (Lin et al., 2008; Liu et al., 2009; Okazaki et al., 2009). Most existing agricultural applications rely on commercial Raman spectrometer systems, which cannot cover large areas of sample surfaces due to the size limitation of the point mea‐ surement. Spatial information, which is crucial for evaluat‐ ing food safety and quality, thus cannot be obtained by traditional spectrometer measurement. Raman chemical imaging is a novel technique that combines the advantages of Raman spectroscopy and digital imaging, with the aim of visualizing the composition and morphology of the target (Morris et al., 1996; Evans et al., 2005). This technique could open a new avenue for tackling problems in the research field of food safety and quality inspection. Related imaging sys‐ tems have been developed, and some instruments are already commercially available. However, most products in the mar‐ ket perform Raman measurement at microscopic level (onĂmicrometer or nanometer scales). The spatial range cov‐ ered by such systems cannot satisfy the requirements of whole‐surface inspection for individual food items. The optical techniques discussed above (i.e., reflectance, fluorescence, and Raman) can be implemented using either spectroscopy measurement or hyperspectral imaging. Hyper‐ spectral imaging is capable of acquiring both spatial and spectral information, and it has been recently developed as a useful tool for safety and quality evaluation of food and agri‐ cultural products (Lu and Chen, 1998; Kim et al., 2001). There are three major approaches for acquiring 3‐D hyper‐ spectral cubes [hypercubes ( x , y , l )]. These are point‐ scanning, line‐scanning, and area‐scanning methods, as illustrated in figure 1. In the point‐scanning method (also known as the whiskbroom method), a single point is scanned along two spatial dimensions ( x and y ) by moving either the sample or the detector. A spectrophotometer is used to ac‐ quire a spectrum for each pixel in the scene. Hyperspectral image data are accumulated pixel by pixel in an exhaustive manner. Two‐axis motorized positioning tables are usually needed to finish the image acquisition. The line‐scanning method (also known as the pushbroom method) is an exten‐ sion of the point‐scanning method. Instead of scanning one point each time, this method simultaneously acquires a line of spatial information as well as spectral information corre‐ sponding to each point in the line. A special 2‐D image ( y , l ) with one spatial dimension ( y ) and one spectral dimension ( l ) is taken at a time. A complete hypercube is obtained as the line is scanned in the direction of motion ( x ). Hyperspectral systems based on imaging spectrographs work in the line‐ scanning mode. Both point‐scanning and line‐scanning methods are spatial scanning methods. The area‐scanning method (also known as the band sequential method), on the other hand, is a spectral scanning method. This approach ac‐ quires a single band 2‐D grayscale image ( x , y ) with full spa‐ tial information at once. A hypercube containing a stack of single band images is built up as the scanning is performed in the spectral domain. No relative movement between the sample and the detector is required for this method. Imaging systems that use filters (e.g., filter wheels containing fixed bandpass filters and electronically tunable filters) belong to the area‐scanning method. The 3‐D hyperspectral cubes acquired by point‐scanning, line‐scanning, and area‐scanning methods are generally stored in the formats of band interleaved by pixel (BIP), band interleaved by line (BIL), and band sequential (BSQ), respec‐ tively. Different formats have different advantages in terms of image processing operations and interactive analysis. The BIP and BSQ formats offer optimal performance for spectral and spatial access of the hyperspectral data, respectively. The BIL format gives a compromise in performance between spa‐ tial and spectral analysis. The three data storage formats are interconvertible. In this study, we developed a laboratory‐based point‐ scanning Raman chemical imaging system with the capacity to acquire hyperspectral Raman images from large food and biological samples. The objective of this article was to pro‐ vide a detailed description of the Raman chemical imaging system. Specific objectives were to: S Present details of system design, including hardware components, interface software, and operation proce‐ dures. S Perform spectral and spatial calibrations for the Raman chemical imaging system. S Demonstrate performance of the system with an exam‐ ple application, i.e., detection of melamine in dry milk. A schematic diagram of the developed Raman chemical imaging system is illustrated in figure 2. The system was built on an optical breadboard with dimensions of 914 × 762 mm 2 . It is a point‐scanning spectral imaging system that utilizes a 16‐bit high‐performance spectroscopic charge‐coupled de‐ vice (CCD) camera (Newton DU920N‐BR‐DD, Andor Technology, Inc., South Windsor, Conn.). The CCD has an area array of 1024 × 256 pixels, with quantum efficiency (QE) greater than 90% at 800 nm and about 45% at 1000 nm. The CCD is thermoelectrically cooled to ‐70 ° C during spectral data acquisition to minimize the dark current. The cam- era is directly connected to a computer using a USB cable, through which both camera control and data transfer are car‐ ried out. A Raman imaging spectrometer that is specifically designed for 785 nm laser excitation (Raman Explorer 785, Headwall Photonics, Fitchburg, Mass.) is mounted to the camera. The basic structure of the spectrometer includes a narrow input slit, a concave mirror, and a convex reflection grating (Julian et al., 2009). The input slit is 5 mm long × 100Ă m m wide and admits light from the samples. The concave mirror is used to guide the incoming light from the input slit to the reflection grating, where the incident beam is dispersed into different wavelengths in a reflection manner. The con‐ cave mirror reflects the dispersed light to the CCD detector, where continuous spectra are formed. The spectrometer cov‐ ers a Raman shift range of ‐98 to 3998 cm ‐1 (or a wavelength range of 779 to 1144 nm) with a spectral resolution of 3.7Ăcm ‐1 . The reflection grating‐based Raman spectrometer is constructed entirely from reflective optics with high effi‐ ciencies. Thus, it is ideal for low‐light measurements such as Raman and fluorescence imaging. A spectrum‐stabilized laser module with an output of 785Ănm and bandwidth of 0.05 nm (I0785MM0350MF‐NL, Innovative Photonic Solutions, Monmouth Junction, N.J.) serves as the excitation source for the system. The 785 nm la‐ ser has a special advantage over short‐wavelength lasers in that it greatly reduces fluorescence emissions from the sam‐ ples. This module is specifically designed to be used in a fiber‐coupled configuration for high‐resolution Raman spec‐ troscopy applications. The maximum output power of the la‐ ser is 350 mW, and the power can be adjusted by dialing the laser drive current as read out on the LED panel. A fiber optic Raman probe (RPB, InPhotonics, Norwood, Mass.) is used to focus the laser on the samples and acquire Raman scattering signals. The probe utilizes a long‐pass filter assembly to eliminate light at and below the laser wavelength (i.e., Ray‐ leigh and anti‐Stokes scattering). A bifurcated optical fiber bundle is used to connect the laser module, the Raman probe, and the Raman imaging spectrometer. Diameters of the ex‐ citation and collection fibers in the bundle are 105 and 200Ă m m, respectively. The laser light generated from the module is transferred to the probe via the excitation fiber. Af‐ ter laser‐sample interactions, the Raman signals are collected by the probe and delivered to the imaging spectrometer via the collection fiber. A programmable, two‐axis motorized positioning table (MAXY4009W1‐S4, Velmex, Bloomfield, N.Y.) is used to move the samples in two perpendicular direc‐ tions ( x and y , as illustrated in fig. 2) under the Raman probe. The movement of the table is controlled by the computer via a stepping motor controller. The table can travel in a square area of 127 × 127 mm 2 with a displacement resolution (ad‐ vance per step) of 6.35 m m. The parameterization and data‐transfer interface software (fig. 3) for the Raman chemical imaging system was devel‐ oped on a platform of LabVIEW and Vision Development Module (National Instruments, Austin, Tex.) in the Microsoft Windows operating system. Software Development Kits (SDKs) provided by the manufacturers of the camera and the motorized positioning table were used in the LabVIEW pro‐ gramming environment to fulfill various functions such as cameral control, data acquisition, sample movement, and synchronization. A Raman spectrum and a Raman image at user‐selected band are displayed and updated point by point to let users monitor the scan progress in real time. The 3‐D Raman image data are stored in the band interleaved by pixel (BIP) format, which can be analyzed by commercial software packages such as ENVI (ITT Visual Information Solutions, Boulder, Colo.). The Raman probe, positioning table, and sample materials are placed in a closed black box (fig. 2) to avoid influence of ambient light. To ensure the best possible signal‐to‐noise ra‐ tio (SNR) for the Raman spectra, single‐track mode provided by the camera's manufacturer is used for spectral data ac‐ quisition. With this method, a rectangular area on the CCD sensor is defined to only include pixels that are illuminated by the incoming light from the Raman imaging spectrometer. All the rows within this specified area are vertically binned together into the ...
Context 3
... description of the Raman chemical imaging system. Specific objectives were to: S Present details of system design, including hardware components, interface software, and operation proce‐ dures. S Perform spectral and spatial calibrations for the Raman chemical imaging system. S Demonstrate performance of the system with an exam‐ ple application, i.e., detection of melamine in dry milk. A schematic diagram of the developed Raman chemical imaging system is illustrated in figure 2. The system was built on an optical breadboard with dimensions of 914 × 762 mm 2 . It is a point‐scanning spectral imaging system that utilizes a 16‐bit high‐performance spectroscopic charge‐coupled de‐ vice (CCD) camera (Newton DU920N‐BR‐DD, Andor Technology, Inc., South Windsor, Conn.). The CCD has an area array of 1024 × 256 pixels, with quantum efficiency (QE) greater than 90% at 800 nm and about 45% at 1000 nm. The CCD is thermoelectrically cooled to ‐70 ° C during spectral data acquisition to minimize the dark current. The cam- era is directly connected to a computer using a USB cable, through which both camera control and data transfer are car‐ ried out. A Raman imaging spectrometer that is specifically designed for 785 nm laser excitation (Raman Explorer 785, Headwall Photonics, Fitchburg, Mass.) is mounted to the camera. The basic structure of the spectrometer includes a narrow input slit, a concave mirror, and a convex reflection grating (Julian et al., 2009). The input slit is 5 mm long × 100Ă m m wide and admits light from the samples. The concave mirror is used to guide the incoming light from the input slit to the reflection grating, where the incident beam is dispersed into different wavelengths in a reflection manner. The con‐ cave mirror reflects the dispersed light to the CCD detector, where continuous spectra are formed. The spectrometer cov‐ ers a Raman shift range of ‐98 to 3998 cm ‐1 (or a wavelength range of 779 to 1144 nm) with a spectral resolution of 3.7Ăcm ‐1 . The reflection grating‐based Raman spectrometer is constructed entirely from reflective optics with high effi‐ ciencies. Thus, it is ideal for low‐light measurements such as Raman and fluorescence imaging. A spectrum‐stabilized laser module with an output of 785Ănm and bandwidth of 0.05 nm (I0785MM0350MF‐NL, Innovative Photonic Solutions, Monmouth Junction, N.J.) serves as the excitation source for the system. The 785 nm la‐ ser has a special advantage over short‐wavelength lasers in that it greatly reduces fluorescence emissions from the sam‐ ples. This module is specifically designed to be used in a fiber‐coupled configuration for high‐resolution Raman spec‐ troscopy applications. The maximum output power of the la‐ ser is 350 mW, and the power can be adjusted by dialing the laser drive current as read out on the LED panel. A fiber optic Raman probe (RPB, InPhotonics, Norwood, Mass.) is used to focus the laser on the samples and acquire Raman scattering signals. The probe utilizes a long‐pass filter assembly to eliminate light at and below the laser wavelength (i.e., Ray‐ leigh and anti‐Stokes scattering). A bifurcated optical fiber bundle is used to connect the laser module, the Raman probe, and the Raman imaging spectrometer. Diameters of the ex‐ citation and collection fibers in the bundle are 105 and 200Ă m m, respectively. The laser light generated from the module is transferred to the probe via the excitation fiber. Af‐ ter laser‐sample interactions, the Raman signals are collected by the probe and delivered to the imaging spectrometer via the collection fiber. A programmable, two‐axis motorized positioning table (MAXY4009W1‐S4, Velmex, Bloomfield, N.Y.) is used to move the samples in two perpendicular direc‐ tions ( x and y , as illustrated in fig. 2) under the Raman probe. The movement of the table is controlled by the computer via a stepping motor controller. The table can travel in a square area of 127 × 127 mm 2 with a displacement resolution (ad‐ vance per step) of 6.35 m m. The parameterization and data‐transfer interface software (fig. 3) for the Raman chemical imaging system was devel‐ oped on a platform of LabVIEW and Vision Development Module (National Instruments, Austin, Tex.) in the Microsoft Windows operating system. Software Development Kits (SDKs) provided by the manufacturers of the camera and the motorized positioning table were used in the LabVIEW pro‐ gramming environment to fulfill various functions such as cameral control, data acquisition, sample movement, and synchronization. A Raman spectrum and a Raman image at user‐selected band are displayed and updated point by point to let users monitor the scan progress in real time. The 3‐D Raman image data are stored in the band interleaved by pixel (BIP) format, which can be analyzed by commercial software packages such as ENVI (ITT Visual Information Solutions, Boulder, Colo.). The Raman probe, positioning table, and sample materials are placed in a closed black box (fig. 2) to avoid influence of ambient light. To ensure the best possible signal‐to‐noise ra‐ tio (SNR) for the Raman spectra, single‐track mode provided by the camera's manufacturer is used for spectral data ac‐ quisition. With this method, a rectangular area on the CCD sensor is defined to only include pixels that are illuminated by the incoming light from the Raman imaging spectrometer. All the rows within this specified area are vertically binned together into the shift register of the CCD and then digitized. As a result, a single Raman spectrum is obtained for each spa‐ tial point. Vertical center position (pixel index) and height (inĂpixels) are specified by the system software to determine the CCD area for acquiring spectral data. The single‐track method is particularly useful in low‐level light applications since the contribution of the dark current in the measured sig‐ nal is minimized by excluding the pixels that are not in the defined area. An example of single‐track mode for spectral data acquisition is shown in figure 4. It can be seen that light is dispersed to a narrow area on the CCD. Under the current system settings, a rectangular area with a height of 27 pixels was selected for useful measurement data (i.e., the area be‐ tween two dashed lines in fig. 4a). The highest intensity (CCD count) of the Raman spectra extracted from any single row within the defined CCD area is less than 1500 (fig. 4b). After vertical binning, the intensity of the final Raman spec‐ trum is at least one order higher than those of the individual spectra (fig. 4c), indicating the effectiveness of the single‐ track spectral data acquisition method. Figure 5 illustrates the point‐scanning pattern followed by the Raman chemical imaging system for acquiring spatial in‐ formation from the target. The process is a row‐wise scan starting from a point at the lower right corner of the scene (P11 in fig. 5). After the first row is completed, the sample is moved to focus on the second row, and the scan continues at the first point in this row (i.e., P21). This process is repeated until the last point (i.e., Pnm) is scanned. The spatial resolu‐ tion of the acquired images is determined by the step sizes for two scan directions (i.e., x and y ). Spatial range of the scene is determined by the combination of the step sizes and the numbers of scans (i.e., m and n ) for the x and y directions. Af‐ ter all the points are scanned, the two‐axis positioning table moves the sample back to the start point, and a 3‐D image data with a dimension of m × n × 1024 (1024 bands) will have been obtained. A Raman spectrum is generally presented as a shift in en‐ ergy from that of the excitation source. Although Raman shift is essentially a relative unit, the spectral dimension i.e., x ‐axis) of a Raman spectrum is traditionally expressed as wavenumber in cm ‐1 instead of D cm ‐1 . Spectral calibration for the Raman chemical imaging system is intended to define the wavenumbers for the pixels along the spectral dimension. Spectrally established light sources (e.g., spectral calibration lamps, lasers, fluorescent lamps, and broadband lamps equipped with interference bandpass filters) are usually used for absolute wavelength calibrations for line‐scanning hyper‐ spectral imaging systems. For spectral calibration of relative Raman shift, it is more convenient to use an excitation source with fixed spectral output and chemicals with known relative wavenumber shifts. A guide for Raman shift standards for spectrometer calibration has been established by ASTM In‐ ternational (ASTM Standards, 2007). This guide covers Ra‐ man shift wavenumbers of eight liquid and solid chemicals measured using Fourier transform or dispersive Raman spectrometers with high spectral resolutions. The eight mate‐ rials are naphthalene, 1,4‐bis(2‐methylstyryl)benzene, sul‐ fur, 50/50 (v/v) toluene/acetonitrile, 4‐acetamidophenol, benzonitrile, cyclohexane, and polystyrene, and they cover a wide wavenumber range of 85 to 3327 cm ‐1 . In this study, polystyrene and naphthalene (fig. 6) were se‐ lected for spectral calibration of the developed Raman chem‐ ical imaging system. Each of these solid chemicals was placed in an individual plastic plate. A 10 × 10 spatial scan with a step size of 1 mm for both the x and y directions was performed for each sample. Figure 7 illustrates the spectral calibration results based on these two standard materials. Ra‐ man spectra of polystyrene and naphthalene are shown in fig‐ ures 7a and 7b, respectively. Each spectrum is an average of 100 spectra extracted from the 1 cm 2 scanned area. Five and seven peaks were identified in the Raman spectra of polysty‐ rene and naphthalene, respectively, collectively covering a wavenumber range between 513.8 and 1602.3 cm ‐1 . Linear, quadratic, and cubic regression functions were used to estab‐ lish the relationship between the known wavenumbers of the 12 identified Raman peaks and the corresponding pixel ...
Context 4
... fiber. Af- ter laser-sample interactions, the Raman signals are collected by the probe and delivered to the imaging spectrometer via the collection fiber. A programmable, two-axis motorized positioning table (MAXY4009W1-S4, Velmex, Bloomfield, N.Y.) is used to move the samples in two perpendicular direc- tions (x and y, as illustrated in fig. 2) ...
Context 5
... Raman probe, positioning table, and sample materials are placed in a closed black box ( fig. 2) to avoid influence of ambient light. To ensure the best possible signal-to-noise ra- tio (SNR) for the Raman spectra, single-track mode provided by the camera's manufacturer is used for spectral data ac- quisition. With this method, a rectangular area on the CCD sensor is defined to only include pixels that are illuminated by the ...
Citations
... The objective of Raman imaging is to concurrently capture the Raman spectrum and visually depict the spatial presence and distribution of the object. Integrating Raman spectroscopy with imaging into a single system allows for the acquisition of morphological and structural details of samples (Qin et al., 2010;Sacré et al., 2014). There are three primary methods for acquiring Raman images: (1) The sample is moved from one point to another on the sample table, and the Raman spectrum is collected one by one at each spatial location called a point scanning (Figure 4a). ...
Ensuring the highest standards of food quality and safety is crucial for maintaining human health. This requires the creation of sophisticated and highly sensitive techniques for detecting and identifying food pollutants. Raman spectroscopy, known for its rapidity, simplicity, sensitivity, and noninvasive characteristics, has become a prominent method in the field of food safety and quality analysis. This review discusses in detail the fundamental principles, instruments, methodology, and data processing techniques of Raman spectroscopy. It specifically emphasizes the practical applications of Raman spectroscopy in assessing foodborne pathogens, toxic gases in spoiled foods, and monitoring the translocation of pesticide in plants. Finally, we discussed the development trend and challenges of Raman spectroscopy in the field of food safety analysis, aiming to provide an accessible and practical guide. This review article equips readers with essential insights into conducting and understanding Raman spectroscopy experiments, thereby contributing to development in the field of food safety analysis.
... In contrast, Qin et al. (2010) developed a desk-top Raman spot-scanning chemical imaging system to obtain detailed information from large food matrices in a shorter time. The hyperspectral Raman images were obtained with a high spectral resolution of 0.1 mm in a Raman displacement range of 102.2 to 2538.1 cm -1 [101]. ...
... In contrast, Qin et al. (2010) developed a desk-top Raman spot-scanning chemical imaging system to obtain detailed information from large food matrices in a shorter time. The hyperspectral Raman images were obtained with a high spectral resolution of 0.1 mm in a Raman displacement range of 102.2 to 2538.1 cm -1 [101]. ...
Raman spectroscopy has evolved into an important fast, rapid, direct, and non-destructive technique that has recently been applied in different fields.
Objective: The present work aims to study the theoretical bases and the experimental techniques relate to Raman spectroscopy and highlight the performance as well as the different applications of the technique.
Method: Spectroscopy, in general, is the study of the interaction between electromagnetic radiation and matter, which corresponds to the emission or transmission of energy in the form of a wave at a given frequency. Raman spectroscopy is based on the inelastic diffusion of photons on electrons. The change in electron energy level leads to different modes of vibration of a molecule.
These different vibration modes occur at specific frequencies for each molecule.
Results: Raman spectroscopy is used in chemistry as a tool to identify molecules in a sample. Indeed,
each Raman peak is associated with a vibration mode of a molecule; it is considered as a more useful approach to monitor the chemical parameters of samples tested in several fields, especially in food safety
Conclusion: This review covers the current research status and prospects of Raman spectroscopy.
The Raman effect is considered from the time of its discovery as a great gift for chemists because it contributes to a better characterization of the structure of matter.
... Traditional matching algorithms [8,9,10], based on spectral curve characteristic peak matching, dominate the existing food spectrum detection. However, those methods require manual extraction, setting, and adjustment of characteristic peaks based on domain knowledge, without the ability to learn and extract features independently. ...
Spectral detection technology, as a non-invasive method for rapid detection of substances, combined with deep learning algorithms, has been widely used in food detection. However, in real scenarios, acquiring and labeling spectral data is an extremely labor-intensive task, which makes it impossible to provide enough high-quality data for training efficient supervised deep learning models. To better leverage limited samples, we apply pre-training & fine-tuning paradigm to the field of spectral detection for the first time and propose a pre-training method of deep bidirectional transformers for spectral classification of Chinese liquors, abbreviated as Spectrum-BERT. Specifically, first, to retain the model's sensitivity to the characteristic peak position and local information of the spectral curve, we innovatively partition the curve into multiple blocks and obtain the embeddings of different blocks, as the feature input for the next calculation. Second, in the pre-training stage, we elaborately design two pre-training tasks, Next Curve Prediction (NCP) and Masked Curve Model (MCM), so that the model can effectively utilize unlabeled samples to capture the potential knowledge of spectral data, breaking the restrictions of the insufficient labeled samples, and improving the applicability and performance of the model in practical scenarios. Finally, we conduct a large number of experiments on the real liquor spectral dataset. In the comparative experiments, the proposed Spectrum-BERT significantly outperforms the baselines in multiple metrics and this advantage is more significant on the imbalanced dataset. Moreover, in the parameter sensitivity experiment, we also analyze the model performance under different parameter settings, to provide a reference for subsequent research.
... With Raman scattering, it is possible to obtain information on the structure of a molecule, which allows the chemical and structural characterization and identification of complex biological material. The main advantages of this analytical technique include the accuracy of the measurements, the large amount of information obtained (at a relatively low cost), the ability to examine the sample without complex preparation and processing, and non-destructiveness [33,34]. ...
Field studies were conducted from 2016 to 2019 (south-eastern Poland; 49°58′40.6″ N 22°33′11.3″ E) with the aim to identify the chemical composition of winter wheat grain upon foliar application of biostimulants, of which PlanTonic BIO (containing nettle and willow extracts) showed antifungal activity. The main chemical compositions and their spatial distribution in wheat grain were characterized by Raman spectroscopy technique. It was established that applied biostimulants and hydro-thermal conditions changed the chemical composition of the grain during all the studied years. A similar chemical composition of the grain was achieved in plants treated with synthetic preparations, including both intensive and extensive variants. The second group, in terms of an increase in fatty acid content, consists of grains of plants treated with biostimulants PlanTonic BIO, PlanTonic BIO + Natural Crop and PlanTonic BIO + Biofol Plex. The future of using biostimulants in crop production, including those containing salicylic acid and nettle extracts, appears to be a promising alternative to synthetic crop protection products.
... HSI systems collect a hyperspectral cube (x , y , λ) that includes 2D spatial information as well as spectral information λ, and they have been used in a wide range of applications such as on-site environmental remote sensing [3], food safety control [4,5], soil classification [6], and gas detection [7]. Based on the scanning mode, HSI systems are categorized into three types: point scanning [8], plane scanning (snapshot imaging) [9], and line scanning [10]. A point scanning imager acquires spectral information of the measured objects point by point, which can be time-consuming. ...
A four-dimensional hyperspectral imager (FDHI) that combines fluorescence spectral detection and 3D surface morphology measurement is proposed. The FDHI consists of a hyperspectral line-scanner, a line structured light stereo vision system, and a line laser. The line laser is used as both the excitation light for the fluorescence and the scanning light line for the 3D profiling. At each scanning step, the system collects both fluorescent and 3D spatial data of the irradiated line region, which are fused to 4D data points based on a line mapping relationship between the datasets, and by scanning across the measurement object, a complete 4D dataset is obtained. The FDHI shows excellent performance with spatial and spectral resolution of 26.0 µm and 3 nm, respectively. The reported FDHI system and its applications provide a solution for 4D detection and analysis of fluorescent objects in meters measurement range, with advantage of high integration as two imaging modules sharing a same laser source.
... Like near-infrared hyperspectral imaging, Raman imaging combines Raman spectroscopy with digital imaging to uniquely capture pixel-based spectral and spatial information of samples, i.e., mapping composition and morphology [137]. Due to their high specificity, low sensitivity to water, and little sample preparation, Raman techniques have recently found applications in food quality and safety inspection [138,139]. ...
Fruits and vegetables are very important agricultural products in daily life. Evaluating the quality attributes of fresh fruits and vegetables by nondestructive sensing techniques has been an intensive research topic over the past two decades. The research progress on the detection of internal and external quality attributes of fresh fruits and vegetables using various nondestructive spectroscopic and imaging techniques, including visible/near-infrared spectroscopy, time-resolved and space-resolved spectroscopy, machine vision, hyperspectral and multispectral imaging, fluorescence techniques, X-ray imaging, computed tomography scanning, magnetic resonance imaging, and Raman techniques, is presented and discussed in this review. Each kind of fruit or vegetable shows great variability in physical characteristics (including size, shape, color, and temperature) and biological characteristics (including cultivar, season, maturity level, and geographical origin). This physical and biological variability complicates the quality evaluation of fresh fruits and vegetables. To eliminate the influence of variability and improve the inspection accuracy, a lot of attempts, including pre-processing, light intensity transformation, global model, band math, model transfer, etc., have been made in image correction and spectral compensation methods. This review provides a detailed summary of the various methods for solving the problem of physical and biological variability, as well as their advantages and disadvantages. Additionally, the current problems to be solved in spectroscopic and imaging technologies and the research trends of nondestructive measurement of the quality of fresh fruits and vegetables in the future are also revealed.
... HSI collects a three-dimensional hyperspectral data cube that contains not only spectral data but also spatial data. Several prototypes have been described in literature for food fraud testing, e.g., Raman imaging for milk powder authentication [47] or apple contamination [48], short-wave infrared HSI on nut quality [49], or hyperspectral imaging for contamination detection [50][51][52]. Steps to valorise smartphone cameras have been reviewed by Rateni et al., 2017 [11], and McGonigle et al., 2018 [53]. ...
This review summarises miniaturised technologies, commercially available devices, and device applications for food authentication or measurement of features that could potentially be used for authentication. We first focus on the handheld technologies and their generic characteristics: (1) technology types available, (2) their design and mode of operation, and (3) data handling and output systems. Subsequently, applications are reviewed according to commodity type for products of animal and plant origin. The 150 applications of commercial, handheld devices involve a large variety of technologies, such as various types of spectroscopy, imaging, and sensor arrays. The majority of applications, ~60%, aim at food products of plant origin. The technologies are not specifically aimed at certain commodities or product features, and no single technology can be applied for authentication of all commodities. Nevertheless, many useful applications have been developed for many food commodities. However, the use of these applications in practice is still in its infancy. This is largely because for each single application, new spectral databases need to be built and maintained. Therefore, apart from developing applications, a focus on sharing and re-use of data and calibration transfers is pivotal to remove this bottleneck and to increase the implementation of these technologies in practice.
... These types of imaging techniques are non-destructive methods that use radio waves and magnetic fields, typically used for analysing the internal parts of target objects. Qin et al., proposed a Raman chemical imaging system to determine the quantity and spatial spreading of melamine mixed with the milk (Qin, Chao, & Kim, 2010). Gan et al., food inspection application uses non-contact ultrasonic systems to detect physicochemical changes in food products. ...
Economic development of many countries depends mainly on agriculture. The role of agriculture has created a vast impact on various fields’ namely industrial activities, employability and capital formation. Further, to improve the economic growth of a country, technology-based agriculture processes can be deployed in pre-production and post-production stage. This study mainly focuses on the various post-production activities such as grading, sorting, quality evaluation, monitoring, defects detection, recognition and maturity detection. Traditionally, various agricultural post-production stages such as grading, sorting and quality assessment is mostly done as manual process. Such inspections are prone to errors due to their ineffectiveness and unreliability. Thus, advancements in technology can be deployed for enhancing manual agricultural tasks. This study deals with image processing techniques in agricultural post-production stages. With the support of computer vision-based systems (CVS), efficiency in post-production various activities can be improved. Various feature properties like size, colour, shape of the agricultural products like vegetables and fruits can be used in CVS for grading and sorting. The steps involved in Computer Vision-based Grading System (CVGS) are image acquisition, preprocessing, image segmentation, features extraction and classification. Though many intelligent models exist for grading agricultural products, there are still broad challenges that need to be resolved. These challenges can be weighed against the advantages of CVGS and the grading process can be improved a lot. This work is based on the exhaustive survey of CVS and the steps involved in CVGS. The outcome of this survey is presented as a comparative analysis of various technologies for grading systems available in the market.
... In addition, only the spectral signal of the sample can be obtained, and the distribution of the substance inside the sample cannot be known. High-throughput Raman imaging integrates the advantages of Raman spectroscopy and digital imaging, which can obtain Raman spectra and spatial distribution information simultaneously during sample scanning [Qin et al., 2010;Wang et al., 2017a;Zhai et al., 2017]. There are three ways to acquire Raman images: point-scan, line-scan, and plane-scan [Lohumi et al., 2017]. ...
Azodicarbonamide (ADA) additives are limited or prohibited from being added to wheat flour by various countries because they may produce carcinogenic semicarbazide in humid and hot conditions. This study aimed to realize the non-destructive detection of ADA additives in wheat flour using high-throughput Raman imaging and establish a quantitative analysis model. Raman images of pure wheat flour, pure ADA, and wheat flour-ADA mixed samples were collected respectively, and the average Raman spectra of each sample were calculated. A partial least squares (PLS) model was established by using the linear combination spectra of pure wheat flour and pure ADA and the average Raman spectra of mixed samples. The regression coefficients of the PLS model were used to reconstruct the 3D Raman images of mixed samples into 2D grayscale images. Threshold segmentation was used to classify wheat flour pixels and ADA pixels in grayscale images, and a quantitative analysis model was established based on the number of ADA pixels. The results showed that the minimum detectable content of ADA in wheat flour was 100 mg/kg. There was a good linear relationship between the ADA content in the mixed sample and the number of pixels classified as ADA in the grayscale image in the range of 100 – 10,000 mg/kg, and the correlation coefficient was 0.9858. This study indicated that the combination of PLS regression coefficients with threshold segmentation had provided a non-destructive method for quantitative detection of ADA in Raman images of wheat flour-ADA mixed samples.
... At present, our research group is combining the point scanning technique with the DWECRS system to develop a point-scan dual-wavelength excitation Raman spectral imaging system for inspecting samples with large surface areas [52][53][54][55]. We hope that the DWECRS system will be used in more highly fluorescent applications, such as porcine adipose tissue diagnosis [56] and chocolate quality control [57]. ...
As fluorescence is the major limitation in Raman scattering, near-infrared excitation wavelength ( ${\gt}{780}\;{\rm nm}$ > 780 n m ) is preferred for fluorescence suppression in Raman spectroscopy. To reduce the risk of fluorescence interference, we developed a dual-wavelength excitation combined Raman spectroscopy (DWECRS) system at 785 and 830 nm. By a common optical path, each laser beam was focused on the same region of the sample by a single objective lens, and the dual-wavelength excitation Raman spectra were detected by a single CCD detector; in addition, 785 and 830 nm excitation Raman spectra can be directly constructed as combined Raman spectrum in the DWECRS system. The results of pure peanut oil and glycerol indicate that the combined Raman spectrum cannot only reduce fluorescence interference but also keep a high signal-to-noise ratio in the high-wavenumber region. The results of dye-ethanol solutions with different concentrations show that the handheld DWECRS system can be used as a smart method to dodge strong fluorescence. Furthermore, we developed a peak intensity ratio method with the DWECRS system to distinguish different types of edible oils. The peak intensity ratio distribution chart of edible oils showed each oil normalized center was relatively independent and nonoverlapped, which can be used as the basis of edible oil classification analysis. In the future, the DWECRS system has potential to be used as a tool for more complex applications.
