Method and Device for Bone Scanning in Meat

专利摘要:
A method and device for detecting bone in meat identifies fragments larger than about 1 mm using spectral optical and ultrasound images. spectral imaging can detect foreign material near the surface and ultrasound can detect material within the sample. The sample is irradiated by light and the reflected light or raman scattered light is measured. The sample is similarly irradiated by ultrasound and the reflected or transmitted sonic waves provide a set of amplitude data points, including time delay. These data points are then processed by statistical methods to derive a set of vectors in n-dimensional space, which are compared with a calibrated set of derived vector data that have distinct identi fi cation loci for each surface type. presence or absence of defects.
公开号:BR112017001407A2
申请号:R112017001407-6
申请日:2015-07-21
公开日:2019-11-12
发明作者:Prystupa David
申请人:7386819 Manitoba Ltd；
IPC主号:

专利说明:

“METHOD AND DEVICE FOR BONE MEASURE SCAN”
Field Of Invention [001] The present invention relates to the detection of small objects, totally or partially, embedded in the soft tissue. Generally, though not essentially, objects are bone fragments or very small bones in the flesh. Large bones are not a problem because they are easily visible. Commercially, most commonly, the meat is chicken breast, since the bone tends to fragment when the breast is boned. The invention can also be applied to poultry, fish and other meats likely to contain very small bone fragments or bones.
Background of the Invention [002] Bone fragments or rigid objects larger than 1 mm, which may be present in food products, represent a risk to human health. Consequently, bone fragments represent both regulatory risk and litigation risk for food processing operations. For a bone detection method to be commercially viable, the method must be able to reliably detect bone fragments at the small end of the band. Superficial defects are more common, inlaid defects less common.
[003] C bone is a composite matrix with a variety of morphologies. The main structural components of bone are Cas hydroxyapatite (PO ^ sOH and type I collagen. Collagen is also the main constituent of cartilage, which is often closely associated with bone. Significant amounts of lipid and hydration water are also associated with bone. in the native state Other biomolecules are present, but not in sufficient quantity to have a significant effect on the types of measurements discussed in this document.The technical problem is to find bone in a meat matrix composed of protein and lipid.
[004] The first approaches to the problem involved variants of the transillumination method (backlighting) for processing fish fillets.
Petition 870170004653, of 23/01/2017, p. 17/73
2/39
In this approach, the meat sample is backlit and variations in the transmitted intensity indicate the presence of an absorbent object (usually bone) in the meat. The main deficiency of this method is that the tissue scatters photons in all discontinuities of the refractive index, effectively on the scale of cell dimensions. Due to scattering on cell surfaces, information about the direction of propagation of a photon is almost completely randomized within approximately 3 mm of travel. Photon diffusion models better describe the spread of photons through tissue. To further complicate matters, an increase in the thickness of the meat has the same mitigating effect as the presence of a bone under the surface. The transillumination method is therefore limited to thin samples with uniform thickness. Although it is possible under laboratory conditions to measure photons that travel through a pulsed laser, without scattering, for up to 10 cm of tissue by time-gating methods, the fraction transmitted directly is in the order of 10-12 of the incident intensity. The cost and sophistication required to increase the range of the transillumination method makes it unsuitable for the food processing applications contemplated by the present invention.
[005] The US7363817 patent discloses a method of transillumination using 500 nm to 600 nm of backlighting with a planar array of LEDs and off-axis ultrasound scattering, added to provide some sensitivity to defects in the mass. The light detector (camera) is aligned with the incident light. The described method measures the attenuation between an acoustic transmitter and a receiver oriented to capture off-axis scattering. Mei's scattering theory applies in the regime where the size of a scattering object is close to the wavelength of the scattered wave. In this regime, spreading can be highly directional, and detection depends on the random presence of a detector at the appropriate spreading angle. Second, the signal for a small defect can be lost within a larger texture signal within the meat matrix.
Petition 870170004653, of 23/01/2017, p. 18/73
3/39 [006] US patent 4631413 discloses an elegant method in which the fluorescence of bone, cartilage and fat is excited by UV radiation. This method has the advantage that the fluorescence of the protein matrix is minimal. High amplitude indicates bone, cartilage or fat, while low amplitude indicates meat.
[007] US patent 7460227 describes a later variant of the UV fluorescence method, which measures fluorescence at two wavelengths to improve discrimination between cartilage and bone. The UV fluorescence method, as well as the transillumination method, is limited to fine samples due to the high photon scattering cross section of the meat. In an industrial environment, there is a need to protect workers from the UV radiation used in this method.
[008] Most of the prior attention to the problem of bone fragment detection has focused on the development of x-ray modalities, which have much smaller scattering cross sections than photons at longer wavelengths and, therefore, can directly form images of defects deeply concealed in the tissue. In addition, the spread of x-rays depends on the electronic density and is therefore more sensitive to heavy elements, such as Ca in bone, than light elements H, C, O and N in the mass matrix. The x-ray method has limited the ability to detect weakly mineralized cartilage and bones or to account for the variation in sample thickness. Historically, this has led X-ray systems from simple direct imaging to sophisticated computed tomography systems. X-rays measure electronic density, which is higher in heavier atoms, especially Ca and P, which are both constituents of bone. Several US patents describe this approach as shown below.
[009] 5585603 object mass [010] 6023497 tuned detector [011] 6299524 [012] 6512812 single transmitter
Petition 870170004653, of 23/01/2017, p. 19/73
4/39 [013] 6546071 single emitter [014] 6563904 single emitter [015] 6600805 2 sources [016] 6370223 2 sources plus laser profile for factoring out thickness [017] 6449334 2 sources, 2 energies [018] 6597759 2 sources , 2 energies [019] 6597761 CT [020] 5182764 CT [021] 6430255 CT [022] 6590956 CT [023] 6018562 CT [024] 7060981 CT increased speed using multiple sources at increased cost [025] CT, computed tomography measures the wave intensity at multiple angles and back calculates an image. Several problems remain even with the most recent CT systems. There is a requirement to protect workers from x-rays and to document radiation exposure daily. X-ray emitters use high voltage and are operated in a humid environment, presenting more risk to workers. The high cost of capital and the high cost of maintenance has limited the adoption of x-ray methods in food processing applications.
Brief Description of the Invention [026] In accordance with a first aspect of the invention, a method is provided for detecting defects in a meat sample on a production line, comprising the steps of:
emission of at least one wavelength of light in an area of said meat sample;
Petition 870170004653, of 23/01/2017, p. 20/73
5/39 receiving reflected light from said area of said meat sample, measuring the amplitude of said reflected light;
and in a data processor, comparison of the amplitudes of said reflected light for each area of said meat sample by multivariate analysis; and determining, based on the referred multivariate analysis, the presence of surface defects in said meat sample.
[027] As an important feature of the invention, preferably, the method includes the additional steps of:
emission of at least one ultrasound frequency in an area of said meat sample receiving ultrasound returned from said meat sample;
measurement of amplitudes and travel times of the returned ultrasound;
comparison of the amplitudes and travel times of the referred ultrasound returned to each area of the meat sample by multivariate analysis;
determination from the referred multivariate analysis of the presence of superficial and internal defects in said meat sample.
[028] According to a second aspect of the invention, a method is provided for the detection of defects in a meat sample on a production line, comprising the steps of:
[029] emission of at least one ultrasound frequency in an area of that meat sample;
[030] receiving ultrasound returned from said meat sample;
[031] measurement of amplitudes and travel times of the returned ultrasound;
[032] comparison of the amplitudes and travel times of the referred ultrasound returned to each area of the meat sample by multivariate analysis;
[033] determination from the referred multivariate analysis of the presence of
Petition 870170004653, of 23/01/2017, p. 21/73
6/39 surface and internal defects in the meat sample.
[034] The methods, as described in greater detail below, can provide one or more of the following characteristics, objects or advantages.
[035] The main object is to provide a robust and economical way to detect small defects both on the surface and inside the meat mass.
[036] A main object is to provide a spectral imaging system and a method for detecting surface defects in a meat sample, replacing meat inspectors on a production line.
[037] A main object is to provide an acoustic ultrasound system and a method for detecting bones on and in a meat sample.
[038] Another main object is to provide a device for detecting defects in a meat sample on a production line, having at least one light emitter and at least one optical detector to register optical signals that provide the signals as data to a processor of data, which processes the data in order to indicate the presence of defects in said meat samples, the data processor has an associated indicator that indicates the presence of a defect in said meat sample.
[039] Another main object is to provide a device that has at least one ultrasound emitter and at least one acoustic detector to register acoustic signals, in which both optical and acoustic detectors provide signals as data to a data processor.
[040] A secondary object is to provide a device in which the light emitter is selected from the group consisting of a broadband white light source, a light source with at least two types of LEDs of different wavelengths , an almost monochromatic laser light source to excite the radiation spread by Raman, an almost monochromatic LED light source filtered through at least one band-pass filter to excite the radiation
Petition 870170004653, of 23/01/2017, p. 22/73
7/39 spread by Raman, a light source with at least two strobe LEDs (strobed) of different wavelengths, a light source in the near infrared and an ultraviolet light source to excite the radiation spread by Raman.
[041] Another secondary object is to provide a device with a light source with at least two types of LEDs with wavelengths between 620 and 640 and 720 and 760 nm.
[042] Another secondary object is to provide a device with a light source with at least three types of LEDs with wavelengths between 540 and 570, 620 and 640 and 720 and 760 nm.
[043] Another secondary object is to provide a device with an ultraviolet light source that emits light with a wavelength between 200 and 220 nm to excite the Raman scattering.
[044] Another secondary object is to provide a device in which an almost monochromatic laser light source emits visible-wavelength light and infrared light selected from the group consisting of 488, 515, 532, 594, 633, 635 , 650, 660, 670, 785, 808, 830, 850, 980 and 1064 nm to excite Raman scattering.
[045] Another secondary object is to provide a device in which an almost monochromatic LED light source emits visible-wavelength light and infrared light selected from the group consisting of 488, 515, 532, 594, 633, 635 , 650, 660, 670 785, 808, 830, 850, 980 and 1064 nm to excite the filtered Raman spread through at least one bandpass filter to excite the Raman spread.
[046] Another secondary object provides a device in which a light source in the near infrared emits light with a wavelength between 900 and 2600 nm.
[047] Another secondary object provides an ultrasound emitter, which is
Petition 870170004653, of 23/01/2017, p. 23/73
8/39 a transverse array of transducers.
[048] Another secondary object is to provide an array of ultrasound transducers, each being controlled separately by a logic processor that drives a switching energy circuit for an energy converter for each of the said transducers.
[049] Another secondary object is to provide an optical detector selected from the group consisting of a transverse linear scanning detector comprising pixels, a focal plane matrix of pixels and said pixels measuring amplitudes of light.
[050] Another secondary object is to provide, when the optical detector is a matrix of focal plane of pixels, an associated wavelength selector.
[051] Another secondary object is to provide an associated wavelength selector selected from the group consisting of a prism, a diffraction network and a band-pass filter, in which the focal plane matrix comprises a plurality of transversal matrices separate pixels, each separate matrix corresponding to a different selected wavelength.
[052] Another secondary object is to provide an associated wavelength selector which is a Fourier transform spectrometer with an optical detector selected from the group consisting of an integral optical detector for said Fourier transform spectrometer and a detector optical connected to said spectrometer with Fourier transform through an auxiliary detector connection.
[053] Another secondary object is to provide an acoustic detector selected from the group consisting of the ultrasound emitter comprising a transverse matrix of transducers and a separate matrix of acoustic transducers isolated acoustically from said ultrasound emitter, in which the detector acoustic measures the acoustic amplitudes and the travel time of each acoustic amplitude.
Petition 870170004653, of 23/01/2017, p. 24/73
9/39 [054] Another secondary object is to provide a data processor to receive a plurality of light amplitudes that correspond to a sample area of that meat sample, and the data processor using multivariate analysis generates n- orthogonal dimensions, by projection into n eigenvectors from a calibration set and compare these data vectors with vectors in a calibration set to determine whether they correspond to bone, cartilage, fat, meat or skin, or contaminant for each sample area of the sample, when bone is identified, a logic signal is sent to trigger a pass-fail gate stopping the sample, otherwise, no logic signal is sent.
[055] Another secondary object is to provide that the data processor additionally identifies the amplitudes of the areas neighboring the said sample area, directly adjacent and diagonally, for each wavelength, the amplitudes of the area and neighboring areas for all wavelengths. are subjected to multivariate analysis, which generates orthogonal n-dimensional data vectors, by projection into n eigenvectors from a calibration set, and compares these data vectors with vectors in a calibration set, which further determines the presence of edges between sample areas, when an edge is identified, a logic signal is sent to trigger a fault door stopping the sample, otherwise, no logic signal is sent.
[056] Another secondary object is to provide that the data processor receives a plurality of acoustic amplitudes and travel times of said amplitudes that correspond to a sample area of said meat sample, said data processor compares said amplitudes with amplitudes standard to determine the presence of bone in said sample, when bone is present, a logic signal is sent to trigger a fault gate stopping the sample, otherwise, no logic signal is sent.
[057] Another secondary object is to provide a data processor that
Petition 870170004653, of 23/01/2017, p. 25/73
10/39 centers on the mean and normalizes the referred amplitudes to the standard deviation for each wavelength.
[058] Another main object is to provide a method for detecting defects in a meat sample on a production line comprising the steps of emitting at least one wavelength of light in an area of said meat sample; receiving reflected light from said area of said meat sample, measuring the amplitude of said reflected light, comparing the amplitudes of said reflected light for each area of said meat sample by multivariate analysis, determining from said multivariate analysis of presence of surface defects in the meat sample.
[059] A secondary object is to provide a method comprising the additional steps of emitting at least one ultrasound frequency in an area of said meat sample; receiving ultrasound returned from said meat sample; measurement of the amplitudes and travel times of the referred ultrasound returned, comparison of the amplitudes and travel times of the referred ultrasound returned for each area of the meat sample by multivariate analysis; determination from the referred multivariate analysis of the presence of superficial and internal defects in said meat sample.
[060] Another secondary object is to provide a method comprising the additional steps of comparing the amplitudes of said reflected light for each area of said meat sample and the amplitudes and travel times of said ultrasound returned to each area of said meat sample by multivariate analysis and determination from said multivariate analysis of the presence of superficial and internal defects in the meat sample.
[061] Another secondary object is to provide a method in which a single wavelength of light is emitted and the reflected light is spread by Raman, with Petition 870170004653, of 23/01/2017, p. 26/73
11/39 comprising the additional step of dispersing said Raman scattered light through a wavelength selector to separate the Raman scattered light into different wavelengths and, further, the additional step of measuring the amplitudes of said wavelengths distinct wave.
[062] Another secondary object is to provide a method in which the said single wavelength to excite the light scattered by Raman is almost monochromatic and selected from ultraviolet in the range of wavelengths from 200 to 220 nm and visible light and light infrared at 488, 515, 532, 594, 633, 635, 650, 690, 670, 785, 808, 830, 850, 980 and 1064 nm.
[063] Another secondary object is to provide a method in which said at least one wavelength of light is broadband white light and comprises the additional step of dispersing said reflected light through a wavelength selector to separate the light reflected at different wavelengths and, still, the additional step of measuring the amplitudes of said different wavelengths. Another secondary object of the invention is to provide a method in which said at least one wavelength of light is wavelength in the near infrared selected from the range of 900 to 2600 nm. Another secondary object of the invention is to provide a method in which said at least one wavelength of light comprises at least two separate wavelengths.
[064] Another secondary object is to provide a method comprising the steps of emitting at least two separate wavelengths at different times and the steps of measuring said amplitudes of reflected light at different times.
[065] Another secondary object is to provide a method in which at least two separate wavelengths comprise between 620 and 640 and 720 and 760
Petition 870170004653, of 23/01/2017, p. 27/73
12/39 nm. Another secondary object of the invention is to provide a method in which at least two separate wavelengths comprise between 540 and 570, 620 and 640 and 720 and 760 nm. These wavelengths can be, and conveniently are, non-coherent light emitted by non-coherent LEDs, typically from wavelength bands 540 to 570, 620 to 640,720 at 760 nm. Wavelengths can be, and conveniently are, non-coherent light emitted by non-coherent LEDs, typically from wavelength bands 540 to 570, 620 to 640, 720 to 760 nm. In the 540 to 570 nm band, the central value can be anywhere from 540 to 570 nm; in the 620 to 640 nm band the central value of 630 nm is optimal, in the 720 to 760 nm band the central value can be anywhere from 720 to 760 nm.
[066] Another object is to provide a device for detecting defects in a meat sample on a production line, which comprises at least one ultrasound emitter and at least one acoustic detector for recording acoustic signals, which provides the signals as data for a data processor. The data processor receives a plurality of acoustic amplitudes and travel times of these amplitudes corresponding to a sample area of said meat sample. The data processor compares these amplitudes with standard amplitudes to determine the presence of bone in the sample, when bone is present, a logic signal is sent to trigger a fault gate stopping the sample, otherwise, no logic signal is sent. Preferably, the device comprises at least one array of ultrasound emitters and at least one array of acoustic detectors for recording acoustic signals, the array of acoustic detectors that provide the signals as data to a data processor. The device may comprise at least one array of ultrasound emitters above said production line and at least one array of acoustic detectors for recording acoustic signals below said production line. Alternatively, the device may comprise at least one array of ultrasound emitters below said production line and
Petition 870170004653, of 23/01/2017, p. 28/73
13/39 at least one acoustic detector array for recording acoustic signals above said production line.
[067] Another object is to provide a method for detecting defects in a meat sample on a production line comprising the steps of emitting at least one ultrasound frequency in an area of that meat sample receiving ultrasound returned from the sample of beef; measurement of amplitudes and travel times of the returned ultrasound; comparison of amplitudes and travel times of the ultrasound returned for each area of the meat sample by multivariate analysis; determination from the referred multivariate analysis of the presence of superficial and internal defects in said meat sample.
Detailed Description of the Invention [068] The arrangement described herein provides methods for detecting foreign material on the surface or mass of food products with a combination of spectral imaging and ultrasound measurements. Very loosely, the spectral image is used to detect foreign material near the surface and ultrasound is used to detect foreign material within the sample mass. The sample is irradiated by light and the reflected light or light scattered by Raman is measured to provide a set of amplitude data points. The sample is similarly irradiated by ultrasound and the reflected sonic waves provide a set of amplitude data points that include time delay. These spectral and acoustic data points are then processed by statistical methods to derive a set of vectors in n-dimensional space. These vectors are indicative of the presence or absence of defects. Typically, the vectors indicate the presence of bone, cartilage, fat, meat (meat (food) or muscle in the strict sense) or skin in the sample and, thus, the presence or absence of defects.
Optical Measurements
Petition 870170004653, of 23/01/2017, p. 29/73
14/39 [069] All optical measurements are performed in approximate backspacing geometry to eliminate possible shading effects due to irregular sample shapes. The lighting is diffuse to limit the effect of specular reflection and is as homogeneous as possible. Diffuse lighting is achieved using an enlarged source composed of one or more Lambertian radiators. A diffuser plate can be used to improve homogeneity. The illumination can optionally be polarized, with a polarizer rotated 90 ^Q in relation to the incident polarization positioned between the sample and the detector to reduce specular reflection. The general lighting direction is greater than 150 ^Q , preferably as close as possible to 180 ^Q , allowing spatial considerations, usually within 5 ^Q. It can be 180 ^Q if a beam splitter is used. The space between the illumination and the sample can be air, but preferably, it is a liquid to reduce changes in the refractive index. In another embodiment, a cylinder composed of a material that transmits in the region of wavelength of interest is placed in contact with the sample. The cylinder is cleaned to prevent the development of a biofilm. In all embodiments, an optical system composed of reflective and / or refractive elements is used to map the scattered or reflected radiation from a small region of the sample surface with magnification in a detector element. The linear dimensions of the small surface region are x / 2 and the corresponding spatial frequency is 2 / x. The Nyquist Theorem requires 2 / x sampling to solve characteristics with 1 / x spatial frequency. In addition, the optical system must transfer 2 / x spectral frequency modulations with high fidelity as determined by analyzing the modulation transfer function. The optical detector element is a photodiode, or a bolometer. A bolometer, which responds to electromagnetic radiation over a wide range of wavelengths, is less sensitive and has a slower response time. A bolometer is sensitive to drafts and
Petition 870170004653, of 23/01/2017, p. 30/73
15/39 is normally enclosed in a vacuum housing with an optical window. The optical characteristics of the window material determine the practical wavelength range of the bolometer.
[070] A photodiode detector is generally a semiconductor that operates on the photoelectric effect and has an effective long cut-off wavelength related to the bandwidth. Photodiodes are more sensitive and have a faster response time, but a limited wavelength range. Detector elements of either type are often grouped into arrays and each logical element in the array is called a pixel. A pixel can consist of a single or multiple detector elements. A pixel with multiple detector elements typically has an optical filter in front of each detector element to select different wavelengths. The Bayer RGB matrix used in color cameras is an example. A wide range of wavelength filters is available and devices with up to eight wavelength filters are commercially available. The transfer optics are placed between the sample and the pixel matrix to form an image of the sample in the pixel matrix. The required magnification of the optical system is the ratio between the pixel size and x / 2. In practice, the small surface area of each sample is approximately 1/2 mm square. Transfer optics can use refractive optics (lenses), reflective optics (mirrors) or diffractive optics (Fresnel lens). Reflective optics are achromatic. Care must be taken to select a refractive system that is corrected for chromatic aberration in the wavelength region of interest. A diffractive system can either focus or act as a wavelength filter. Other devices, as is known to a person skilled in the art, may be used instead. A wavelength selector, which can be a prism, diffraction grid or bandpass filter, may be needed to isolate and concentrate specific wavelengths, typically a range of wavelengths.
Petition 870170004653, of 23/01/2017, p. 31/73
16/39
When a Fourier transform spectrometer is used as a wavelength selector, it usually has an integral optical detector, typically a photodiode, bolometer, or a linear or focal plane scanning matrix. Most Fourier transform spectrometers also have an auxiliary detector connection so that a detector can be located outside the spectrometer.
[071] In calibration, each pixel is illuminated with a standard light source for each wavelength and scale factors are then calculated for each pixel to match the response. The scale factor takes into account the geometric variation in the physical size of the pixel elements, as well as the variation in the spectral response of each pixel element. It should be noted that a pixel's spectral response and sensitivity are temperature dependent and a well-designed system will include a temperature sensor next to the pixel element (s) to provide response to a temperature controller or to correct scale factors for changes in temperature. Cryogenically cooled detectors are generally more sensitive. Detectors and detector arrays equipped with Peltier refrigerators are commercially available. Calibration is simpler for a strobe system because the same physical detector elements are used for each wavelength. The scaling factor corrections for each wavelength are determined by the spectral response curve of the detector elements, which at first approximation is the same for all elements in a matrix. The pixel matrix can be a single transverse matrix if the light emitter is a strobe. It is more convenient for stroboscopy, because of the natural variation in photodiode / pixel sensitivity and therefore easier to calibrate for a more reliable mean range. When a linear scan, which is essentially one-dimensional, is used, three lines of pixels, 3x1024, can be used to check for errors and obtain a more reliable mean amplitude. If a two-dimensional pixel focal plane matrix is used, usually 640x480 or 1024x1024, lines
Petition 870170004653, of 23/01/2017, p. 32/73
17/39 selected pixels are used corresponding to the desired wavelengths. Again, in general, more than one line of pixels is used for each desired wavelength band.
[072] In one embodiment, the illumination is provided by a broadband white light source and the light reflected diffusely from the sample is dispersed over wavelengths by a diffraction grid or prism and position in a array of focal plane of pixels, each of which registers a wavelength range.
[073] In another embodiment, a light source with two or more types of LEDs is used and the light reflected diffusely from the sample is dispersed by wavelengths and position on a focal plane matrix.
[074] In another embodiment, the almost monochromatic lighting (which could be a laser, but normally not) is provided by an LED light source together with one or more band-pass filters and the resulting Raman scattered radiation is dispersed by wavelength and position in a matrix of focal plane of pixels. It should be noted that the preferred light source is depolarized for Raman measurements because the intensity spread by Raman is dependent on polarization. An LED light source generally fulfills this requirement. If a laser is used, an alternator (scrambler) may be needed to randomize the polarization. The LEDs have a spectral FWHM of 25 to 40 nm and the required bandwidth (FWHM) is about 0.2 nm or less. A suitable filter with a 0.15 nm bandwidth can be obtained from Andover Corporation, Salem NH. The central transmitted wavelength of an interference filter can be adjusted by rotating the filter and this principle can be used to build a narrow bandpass filter from two or more wider (and less expensive) bandpass filters used in series.
[075] In another embodiment, a laser provides almost illumination
Petition 870170004653, of 23/01/2017, p. 33/73
18/39 monochromatic and the radiation scattered by Raman is dispersed by wavelengths and position in a focal plane matrix, the laser provides a better spectral resolution.
[076] In another embodiment, lighting is provided by two or more sets of LEDs that are stroboscopes and light reflected in a diffuse way (not by Raman) by the sample is collected as a function of the position by a linear scanning detector, which measures both wavelengths, only one wavelength is measured at a time.
[077] In another embodiment, InGaAs photodiodes / pixels are used to collect near-infrared spectra, in the 900 to 2600 nm wavelength range. Alternatively, a microbolometer matrix can be used. There are several suitable infrared emitters in this range, as is well known to a person skilled in the art. The near infrared has theoretically deeper penetration, but less sensitivity.
[078] Realizations that use almost monochromatic radiation to excite a Raman spectrum produce data points that are more independent than other methods described here, that is, more detailed spectra and, therefore, the method has greater diagnostic value. As an illustrative example, bone can be distinguished from muscle by strong Raman scattering about 960 cm ' ¹ from the symmetrical stretch and a weaker set of bands close to 1050 cm' ¹ from the asymmetric PO + stretch in hydroxyapatite. Lipids can be determined from symmetrical and asymmetric CH stretch bands in the region between 2850 cm ' ¹ and 3050 cm' ¹ . Proteins produce a distinct Raman spectrum, which includes information about the protein's secondary structure. The most important protein characteristic is the Amide I band close to 1650 cm ' ¹ of amino acid residues in peptides. For these measurements, the excitation wavelength should be chosen as the shortest wavelength that does not cause a significant increase in the background
Petition 870170004653, of 23/01/2017, p. 34/73
19/39 (background) fluorescence. The intensity of Raman scattering is proportional to the fourth power of the incident frequency. Fluorescence can be avoided by using infrared light near the cost of lower signal levels. A suitable wavelength is 633 nm, which can be provided by an LED or a HeNe laser, which prevents fluorescence. A lens system is normally used to collect scattered radiation from the sample and transmit said radiation to a wavelength selector. The wavelength selector must prevent radiation at and near the incident wavelength from reaching the detector element (s), since the energy at the incident wavelength is usually a factor of one million greater than the energy at the measurement wavelengths. The incident wavelength can be blocked by an interference filter or by a double (or triple) grid system. Both options are commercially available from many suppliers and there are a number of commercially available laser LEDs that have visible and near infrared wavelengths, which are suitable for Raman excitation, including 488, 515, 532, 594, 635 , 650, 660, 610, 785, 808, 830, 850, 980 and 1064 nm. In practice, operating wavelengths can differ from nominal wavelengths by about 5 nm due to variations in operating conditions. The detector is chosen for sensitivity in the range of photon wavelength spread by Raman. An array of avalanche photodiodes is the preferred detector technology since the sensitivity is in the range of fW to pW, which compares favorably with a Raman signal in the nW range. A photomultiplier tube will also work if the excitation wavelength is less than 600 nm. CCD technology will also work, but longer sampling times (or higher input energy) are required due to lower sensitivity. Cartilage, like muscle, is composed of a sequence of amino acids, but has an atypical distribution of amino acids. In cartilage, approximately 1/3 of the amino acid residues are
Petition 870170004653, of 23/01/2017, p. 35/73
20/39 proline. A spectrum of Raman resonance, selectively sensitive to proline, can be excited with radiation between 200 nm and 220 nm. Fluorescence is a problem with UV excitation. Where fluorescence is unavoidable, it is possible to collect a Raman spectrum with a pulsed light source coupled with time-gated detection to reject fluorescence, which reaches a time delay longer than the Raman signal, typically around 200 ns. The detector is turned off after Raman detection, to allow fluorescence to pass, then it is turned on again for the next Raman detection. The output light is passed through a device, usually a diffraction network (in theory, a prism can be used) and its intensity measured in a pixel matrix, alternatively, a spectrometer with Fourier transform, which can be combined with a linear scanning detector or focal plane matrix.
[079] To determine the most effective wavelengths, chicken samples were tested in a range from 400 to 800 nm, in discrete 10 nm bands, and the reflected amplitude was measured for each band. The amplitude was measured in comparison with the standard deviation. The samples approached 700 by 700 pixels although the camera was 1024 by 1024 pixels. The cartilage, bone, skin, fat and muscle areas were identified and masks (masks) covering only unambiguously determined surfaces were used to provide amplitudes of reflected light to the pixels within the mask for each type of surface, which counted from from at least 1,000 pixels to 20,000 to provide reliable mean amplitudes and standard deviations. The 540 to 570 nm, 620 to 640 nm and 720 to 760 nm ranges were the most effective. All three ranges are required, each with a significant contribution to eigenvectors, which explains the variance in the sample. As noted below, eigenvectors are derived enough to identify the nature of the surface.
[080] For achievements that use reflected light, an example is provided
Petition 870170004653, of 23/01/2017, p. 36/73
Instruction 21/39 describing the application of the invention to the problem of finding defects in birds' breasts. In one embodiment, Si-based photodiodes are used. The spectral responses of a chicken rib and chicken breast muscle are statistically indistinguishable in the region around 630 nm and this property makes 630 nm a good standard for reference. In the spectral region close to 720 nm, the averages of the chicken rib and chicken breast distributions are separated by the sum of their standard deviations. Thus, measurements at 630 nm and 720 nm are sufficient to distinguish between chicken rib and chicken breast muscle. Cartilage is more reflective than bone. At 630 nm and 720 nm, the ratio is about 1.1 while at 570 nm the ratio is about 1.8. Thus, the cartilage is inferred by greater reflexivity at 570 nm and reflexivity similar to 720 nm compared to the reference measurement of 630 nm. At 570 nm, chicken fat is about 3.4 times more reflective than muscle compared to the 630 nm reference. The skin approaches the fat for spectral reflexivity. At 720 nm, fat is less reflective than muscle (0.84) in relation to the 630 nm reference. These wavelengths were determined by experiment to be effective and to form a set of bases sufficient for multivariate analysis. The three amplitudes are determined for each pixel. In practice, the three amplitudes for each pixel are subjected to multivariate analysis to derive projections on eigenvectors in n-dimensional space, which are then used to determine the nature of the sampled surface area.
[081] While 570, 630 and 720 nm are preferably stroboscopes, they need not be. The LEDs heat up quickly in the order of microseconds, but the shutdown is slow in the order of 300 microseconds. Consequently, a delay of about 300 microseconds is required between the time an LED is turned off and the start of the next integration period. Stroboscopy generally requires that each LED or group of LEDs of the same wavelength has its energy converter controlled by a switching circuit, such as a
Petition 870170004653, of 23/01/2017, p. 37/73
22/39 bridge H in combination with a current limiter, LED controller or similar logic processor. A BuckPuck (LED Supply, Randolph VT) is a suitable control device.
[082] Recently, conveyor belts carrying meat samples have accelerated to about 15,000 samples per hour, or about 4 per second, corresponding to a line speed of 1600 mm / s. To make 3 measurements per 0.5 mm translation, the required sampling frequency is 9.6 kHz for a linear arrangement, which is problematic considering the LED turn-off time noted above. Our experiments were carried out on an execution line at 800 mm / s, so that the sample rate required for a linear matrix is 4.8 kHz. A two-dimensional focal plane matrix can be used instead of recording multiple images of the same sample region at different times. All that is needed is that the sample translates to less than 0.5 mm during the integration time (0.625 ms in our case). So, a series of images
2-D overlays are collected for each wavelength. The images must be compensated for the relative movement between images in order to provide a single 2-D image for each sample for each wavelength. This can be done in two ways. The integration time is adjusted according to the desired Nyquist spatial resolution, as previously described. It is possible to measure all wavelengths with a focal plane array detector in sequence. The integration time, which is the time the detector is turned on to receive photons and add its energy, is typically about 1/4 to 3/4 millisecond. The focal plane is, for example, 1360 transverse pixels by 1024 in length, for example, an area of about 640 transverse pixels by 240 in length is used as a frame, essentially a single image. The frames are taken at different times, as there are three separate wavelengths. The period between frames is generally longer than the integration time due to
Petition 870170004653, of 23/01/2017, p. 38/73
23/39 to the time required to transmit sensor data to the data processor. During the period between the frames, the sample will have translated a distance X mm, preferably corresponding to 2X pixels. The X value and the pixel offset are calculated from the sample translation rate. The interval is typically 21/2 milliseconds. As the sample passes under the chamber a series of frames are taken at each wavelength, a cycle takes 71/2 milliseconds, in practice corresponding to about 12 pixels. The amplitudes of a particular transverse line of pixels in a frame are compared to lines 11, 12 or 13 in the next frame of the same wavelength, in general, one of these is identified as the same, which is shown to be identical. If the sample is used, the product scaled between a region of a first image and a subsequent image (with a compensation range) is calculated and normalized by the magnitude of each data vector. Compensation that produces a value closer to 1,000 is used. The frames or, rather, the pixel amplitudes corresponding to a small common sample region, after appropriate compensation, are added together to provide a longer effective integration. While up to 20 frames can be used in the example given, the general method can be extended to an arbitrary number of frames using multiple focal plane arrays with fields of view compensated for known displacements. The combined amplitudes increase with the number of frames and the noise increases as the square root of the number of frames provides an overall improvement in the signal / noise ratio proportional to the square root of the number of frames added. This amplification method is particularly useful for Raman measurements with intrinsically weak signals. The gross pixel amplitudes are normalized at each wavelength by a scale factor to normalize the response to a white reference. These amplitudes produce a three-dimensional vector, which is used to characterize the nature of the sample surface.
Petition 870170004653, of 23/01/2017, p. 39/73
24/39 [083] Coding marks can be included on the sample transport substrate (conveyor belt) for the purpose of calculating pixel compensation. These marks are distinct equidistant markings that can be used to match the images from each frame, the markings will have the same position relation for each sample, which can then be identified. Pixel values for the same sample region are added for each wavelength. It is also possible to measure all wavelengths simultaneously using separate detectors using one or more beam splitters. As three sets of focal planes comprising pixels each for a separate wavelength are used, normalization is more complex, since all pixels at all three wavelengths must be taken into account. In this case, the time between measurements is shortened, but care must be taken to align the detectors to a common field of view. In either case, it is possible to record multiple images of each sample region, increasing the effective integration time and improving the resulting signal / noise ratio. As an illustrative example, a camera with a focal plane matrix of 1280 x 1024 pixels can be used and the sample is translated in the Y direction. The sample translates 256 pixels between measurements at the same wavelength. In this example, each physical region is measured 4 times. Data processing time is a function of the number of bytes and the speed of the processor.
[084] Preferably, the optical system is enclosed in a chamber protected from ambient light, including the 60 Hz fluorescent lighting effect. The modulation of the lighting amplitude and the passage of the modulation signal to a blocking amplifier connected to the outputs detector can eliminate the effect of ambient light.
Acoustic Measurements [085] The invention also includes an array of ultrasound transducers arranged to span the width of a sample carrier device in such a way
Petition 870170004653, of 23/01/2017, p. 40/73
25/39 that each region of the sample zone can be scanned. The walls of the sampling region are lined with a material designed to absorb and dampen ultrasonic vibrations. For example, the matrix can be approximately 210 mm across to match the width of the conveyor system used in the optical example. Other sizes are possible and should be chosen to approximately match the size of a particular conveyor system. Three variants are planned. The first couples acoustic vibrations to the sample through an aqueous medium. In this case, retroreflection geometry is preferred. In the second variant, the samples are positioned on one side of a conveyor belt and at least one transducer is coupled through a liquid to the opposite side of the conveyor belt. The acoustic signal is transmitted through the conveyor belt, through the sample and travels through an air gap before being received by at least one transducer. The positions of the transmitter and receiver can be interchangeable. The third couples acoustic vibrations to the sample through a cylinder. In this case, one or more transducers are mounted on the cylinder. The transducer (s) can (m) rotate with the cylinder, but are preferably positioned stationary near the center and mate with the movable surface of the cylinder through a liquid.
[086] In one embodiment, a line of transducers preferably 6 mm in diameter is used, usually with about 32 transducers, which are sufficient to cover a typical chicken breast. The 6 mm transducer is large enough to produce a well-focused ultrasonic wave, but small enough to maintain the return of a defect as small as 0.3 mm within the detection limits. The noise / signal ratio for this size is calculated theoretically. Transducers can resonate between 1 MHz and 20 MHz, more preferably 5 MHz in aqueous media. In the air, a suitable frequency is 200 KHz. Higher frequency gives better resolution and less depth of penetration. The ultrasonic frequency is chosen in such a way that the ultrasonic wavelength is less than
Petition 870170004653, of 23/01/2017, p. 41/73
26/39 the minimum defect size x, and more preferably less than x / 2. At this limit, structures with dimensions x and larger generate an acoustic dipole field that can be observed in the backscatter geometry. When the acoustic signal is reflected it has a number of lobes, which vary according to the situation, both lobes, frontal and backscattered, are always present. Backscatter geometry is used in the present invention in preference to forward scatter geometry, because the weak scattered signal is not combined with the strong incident wave, as is the case in forward scatter geometry. It is worth noting that particular defect geometries, where the defect is approximately the same size as the ultrasonic wavelength, can produce reflected waves that are a strong function of the scattering angle with strong signals at some angles and no signals at other angles . The backscatter geometry does not produce the strongest possible signal in these cases, but it does produce a consistent signal, which is preferable in view of the possibility of a lost signal. The backscatter geometry allows the same transducer to both send and receive ultrasonic waves, as long as the oscillation of the output pulse generation decreases to insignificant levels before the arrival of scattered waves.
[087] In an alternative embodiment, a separate set of transducers can be positioned in close angular proximity, but isolated acoustically from the first set of transducers to function as receivers. The detectors measure the effective acoustic conductance or impedance of the tested material and, thus, indicate its density, differences indicating bone, cartilage, fat and muscle. In this embodiment, the transducers can all emit at once and measure the acoustic response simultaneously. They can also emit with a phase time delay, which can sweep the sample in microseconds. The width of the sample channel can be divided into N regions. The time required to sample each region is approximately the time required for an ultrasound wave to travel from
Petition 870170004653, of 23/01/2017, p. 42/73
27/39 from the transducer to the bottom of the sample carrier and return. The phased array / transducer array sends a short train of acoustic waves focused separately on each region, starting with region 1 and ending with region N, in sequence, until a complete line through the sample region has been interrogated. The process is repeated indefinitely. During the sampling time, the backscattered waves are sampled at twice the frequency of the incident waveform. For example, the time required for a return trip to a 5 MHz wave train through 20 mm of soft tissue is about 28 microseconds, consequently about 280 data points are needed to characterize the backscattered waveform. In another embodiment, more than one region can be sampled at the same time, as long as the regions are far enough apart to avoid cross-talk. As a result of the phase difference, there is destructive interference, except within a small sample region. Essentially, a response is received from one area of the sample at a time.
[088] In the air embodiment, the transducers can all emit at once and the detectors measure the acoustic response simultaneously with each other. They can also emit with a phase time delay, which can sweep the sample in microseconds. The width of the sample channel can be divided into N regions. The time required to sample each region is approximately the time required for an ultrasound wave to travel from the transducer to the receivers at the bottom of the sample carrier. The phased array / transducer array sends a short train of acoustic waves focused separately on each region, starting with region 1 and ending with region N, in sequence, until a complete line through the sample region has been interrogated. The process is repeated indefinitely. For example, the time required for a 200 KHz wave train to pass through 20 mm of soft tissue is about 14 microseconds, in the example shown. In another achievement, more than one region can
Petition 870170004653, of 23/01/2017, p. 43/73
28/39 be sampled at the same time, as long as the regions are sufficiently far apart to avoid crosstalk. As a result of the phase difference, there is destructive interference, except within a small sample region. Essentially, a response is received from one area of the sample at a time. The transmitters can be above and the receivers below the production line or the transmitters can be below and the receivers above the production line.
[089] A larger number of transducers can be used, typically 64 or 128, in a phase array of the same physical size, this configuration is similar to medical ultrasound applications and has similar resolution and sensitivity.
[090] In another embodiment, the amplitude, phase or frequency of the outgoing wave train can be modulated to encode temporal information. When the array of transducers is delayed in phase time, each transducer has its energy converter controlled by a switching circuit, such as an H bridge or similar logic processor.
Data Processing [091] The signals from spectral measurements and ultrasound measurements are transmitted to a data processing device, which uses conventional statistical models to infer the presence or absence of a defect. The information provided includes the amplitude at specific wavelengths from the detectors (s), the acoustic amplitude (s) along with the travel time.
[092] Optical amplitudes can be used as absolute values, when submitted to multivariate analysis. It is preferable that the optical amplitudes are centered on the mean and normalized to the standard deviation. If the amplitude is below a certain threshold (that is, there is no part of the sample present), it is not processed. The average of the amplitudes for the sample is obtained for five transversal scans; this number may vary in practice, depending on the detector. This average is
Petition 870170004653, of 23/01/2017, p. 44/73
29/39 then subtracted from the current scan amplitudes to give amplitudes centered on the mean. The standard deviation for this scan is then calculated and the amplitude centered on the mean divided by the standard deviation to give a normalized amplitude centered on the mean. This takes into account the difference in height in the sample. The normalized amplitude centered on the mean a 'is given by the expression a' = (a-m) / s, where a is the measured amplitude, m is the mean and s is the standard deviation. The amplitudes of edges are identified by the data processor for the eight adjacent areas, for the tested area, directly adjacent and diagonally. In theory, these are then compared for gradient from central amplitude tested to adjacent peripheral amplitudes to detect the presence of an edge and, therefore, bone, when the gradient is greater than the standard by a noise threshold.
[093] In the aqueous case, while there is a spectral amplitude at each wavelength for each area, there is more than one acoustic amplitude for each area. In practice, the acoustic amplitude is plotted as a function of the travel time, whichever way the transducer is used, there are five possible results. First, the ultrasonic wave can be emitted to an unsampled region and simply reflect with attenuation on the opposite side of the sampling region. Second, the ultrasonic wave can find a sample region with an almost homogeneous acoustic impedance. In this case, there will be a backscattered wave from the upper sample surface, weak scattering over the sample mass, another backscattered wave from the lower sample surface, and finally scattering from the lower surface of the sample channel. The third case is the same as case 2, except that a small particle on the upper surface with greater acoustic impedance than the mass increases the amplitude of the spread wave from the upper surface. Case 4 is the same as case 3, except that the small high impedance particle is on the bottom surface and increases the amplitude.
Petition 870170004653, of 23/01/2017, p. 45/73
30/39 tude this reflection. In cases 3 and 4, the increased spread is used in conjunction with optical data to determine the presence of a defect. Case 5 is the same as case 2, except that a high impedance particle is between the upper surface and the lower surface. In this case, there is an extra scattering signal, in an intermediate time between the reception of the upper and lower surface signals. In the non-aqueous case, the presence of bone changes the arrival time of the transmitted wave since the speed of sound is faster in the bone.
[094] Multivariate analyzes, such as Principal Component Analysis (PCA), Neural Networks (NN), Linear Discriminant Analysis (LDA), Partial Least Squares (PLS) and similar algorithms can be used to infer the probability of a bone fragment being gift. Two general methods are used to infer the presence of a defect in optical measurements. First, it is possible to assign a probability of a defect within an individual pixel based on differences in the received signal as a function of the wavelength. Second, the probability of a defect in a region corresponding to a pixel can be calculated by comparing the pixel in question with surrounding pixels to detect edges. Edges imply the presence of bone. This detection is done with a direct gradient calculation, using a Sobel mask, or another edge detection algorithm, which compares adjacent amplitudes to derive an amplitude rate of change (gradient). A larger gradient corresponds to a larger edge and probability of defect. In practice, the eight neighboring amplitudes for each wavelength are combined with the central amplitude to generate an edge probability amplitude for each wavelength. The edge probability amplitudes are included in the data vector used to calculate eigenvectors for calibration or eigenvector projections for operation. The ultrasonic signal as a function of time relative to a reference point is included in the data vector. The pattern produced by an included bone is different, but difficult to model with a
Petition 870170004653, of 23/01/2017, p. 46/73
31/39 direct physical model. The statistical model calculates the cumulative probability that a defect exists within a small sample volume based on all measurements. Specifically, wavelength dependence, edge probability and acoustic feedback as a function of time relative to a surface reflection are loaded into a common data vector and the projection of this data vector into a set of calibration vectors orthogonal is calculated. Preferably, but not necessarily, the data are normalized and centered on the mean by the standard deviation of each measurement. For illustrative purposes, the general method for implementing a PCA (Principal Component Analysis) is outlined here. In the PCA method, the set of reference vectors are eigenvectors, which describe a n-dimensional Principal Component space. A set of eigenvectors and eigenvalues are generated from a set of calibration of data vectors by a multivariate analysis routine (PCA). The data vectors in the reference set represent a set of samples with bone fragments and a set of samples without bone fragments. The number of samples in each set is chosen so that the natural variability within each population is well represented. The covariance matrix is calculated and the eigenvectors and eigenvalues are obtained by diagonalizing the covariance matrix. If the data vector is of dimension m, there will be m eigenvectors and m eigenvalues. If 3 wavelengths are measured, there are 3 amplitudes plus 24 edges and m = 27, more if acoustic measurements are included. All eigenvectors corresponding to single eigenvalues are orthogonal. Degenerate eigenvalues are possible, in which case any of 2 or more degenerate eigenvalues are used to represent the eigenvalue. The sample variance described by each eigenvector is proportional to the magnitude of the associated eigenvalue. Generally> 99% of the variance is described by major eigenvectors 2 to 6 which are called PC1, PC2, PC3, etc., in order from largest to smallest corresponding eigenvalue. The sample variance can be projected in a dimension vector PC space
Petition 870170004653, of 23/01/2017, p. 47/73
32/39 reduced, taking the scalar product of each data vector with each of the 2 to 6 eigenvectors corresponding to the largest eigenvalues. The scalar product provides the projection of the original data array across each main component eigenvector. The new vector space is n-dimensional (n generally less than 6 and more often about 3) and all vectors are orthogonal. If the original data vector is centered on the mean and normalized by the standard deviation, the eigenvector units are standard deviations and this is convenient (but not necessary) for interpreting the data in the PC space. Calibration vectors corresponding to skin, bone, muscle, fat, cartilage, etc. group in different regions of the PC space. The locus of each tissue type distribution, along with the likelihood of increasing the distance from the locus, is modeled. When the system is presented with an unknown, the data vector is projected in the PC space and compared with the model for each type of tissue to generate a probability for each type of tissue. The diagnosis for the sample region is the type of tissue most likely. The data vectors in the bone fragment calibration set project to a different region of the Principal Component space from data vectors in the calibration set without bone fragments. Although some variations in the data vectors are observed in practice, they fall into very different groups with little ambiguity. The main component graphics are available, but require different colors for clear interpretation.
[095] Standard samples of bone, cartilage, fat, meat and skin are used to calibrate eigenvectors. In general, a contaminant does not match any calibration set and stands out. Standard Bayesian statistical methods are used to calculate the probability that a bone fragment is present for each small region of the Principal Component space. The projection of an arbitrary data vector in the Principal Components space determines the probability that the
Petition 870170004653, of 23/01/2017, p. 48/73
33/39 data vector represents a defect of the bone fragment. If the calculated probability exceeds a threshold, a signal is produced by the logic system that can be used to remove the defective part from the process chain. The defective part can optionally be reworked, using a finishing line, and then inspected again. Other wavelengths and algorithms can reach the same end result.
[096] The advantage of the system is that it detects both the surface and the bone embedded in the chicken breast. Although the surface of a food sample can be quite uneven on a large scale, the surface does not normally vary much on a scale of a few millimeters, so that the illumination and the average angle of reflection are almost constant. In this approximation, changes in the reflected intensity gradient exceeding a threshold are indicative of a change in composition and can be used to detect edges. Edge detection is well known and off-the-shelf processing software is commercially available. Once an edge is detected, the algorithm looks for other edges in the vicinity and calculates a probability of defect based on the magnitude of the gradient, the length of the edge and the mutual geometry of all edges within an analysis region. As an illustrative example, bones often have edges that are almost parallel to a characteristic spacing between edges. The detection of parallel edges of several mm long, approximately 2 mm apart, in chicken meat, would cause the algorithm to generate a high probability of the presence of a chicken rib.
[097] The products to be inspected may be in the air. In this case, a transparent disposable film separates the optics from the sample area. The film can be moved slowly between two cylinders at a rate that maintains a clear field of view between the sample and the detector. It is understood that an inspection device
Petition 870170004653, of 23/01/2017, p. 49/73
34/39 optics can be positioned to face each sample surface. In a preferred embodiment, one set of optical detectors faces the upper surface of a sample and a second set of optical detectors faces the lower surface. Preferably, the sample is immersed in a clear liquid solution, which minimizes or eliminates specular reflectance, during optical scanning and also engages acoustic waves in the sample more effectively than an air interface. The clear liquid solution can be primarily water. In this embodiment, a submerged transparent window separates the optics from the sample. The transparent window is preferably lowered to avoid abrasion and cleaned periodically to prevent the accumulation of a biofilm.
Brief Description Of Figures [098] Figure 1 shows a schematic side view of a first method according to the present invention.
[099] Figure 1A illustrates a schematic side elevation view of a second method according to the present invention similar to that of Figure 1.
[0100] Figure 1B illustrates a schematic side elevation view of a third method according to the present invention similar to that of Figure 1.
[0101] Figure 1C illustrates a schematic side elevation view of a fourth method according to the present invention similar to that of Figure 1.
[0102] Figure 2 illustrates a schematic side elevation view of an additional method according to the present invention.
[0103] Figure 3 shows a diagrammatic side elevation view of another embodiment of the device.
[0104] Figure 3A illustrates a diagrammatic side elevation view of another embodiment of the device.
[0105] Figure 4 illustrates an amplitude graph measured as amplitude / standard deviation from time in milliseconds.
Petition 870170004653, of 23/01/2017, p. 50/73
35/39 [0106] Figure 5 illustrates a reflectivity graph measured in relation to the wavelength.
[0107] Figure 6 illustrates a graph of spectral separation measured in relation to the wavelength.
Description of Preferred Embodiments [0108] In Figure 1 an apparatus 10 is provided, where a sample of meat 20 is transported on a conveyor belt 28, an upper support race from which it is transported on a metal plate 22. An acoustic transducer 26 driven by an electronic control 32 it is rigidly mounted on the metal plate 22 and acoustically coupled with grease (not shown). The metal plate 22 is acoustically coupled to the conveyor belt 28 with a thin layer of an aqueous solution (not shown). The conveyor belt 28 is acoustically coupled with a meat sample 20 transported on the belt with a thin layer of the aqueous solution (not shown). An opening 24A is provided on a plate 24 that allows the transmission of signals emitted by transducer 26 and transmitted through sample 20 to an acoustic transducer 30. Plate 24 prevents indirect acoustic disturbances (echo) from colliding on transducer 30. The received signals by the transducer 30 are transferred to and amplified by the electronic control 32. An enclosure 48 surrounds the system 10 and prevents ambient light from entering the device 10.
[0109] The illumination of sample 20 on conveyor 28 is carried out by LEDs 52, 54 and 56. LED 52 is 570 nm, LED 54 is 630 nm and LED 56 is 720 nm. A diffuser 58 is located in front of the LEDs and provides uniform illumination. LEDs 52, 54 and 56 are stroboscopes and the images reflected at each wavelength are collected by a camera 50 and transmitted to electronic control 32. The acoustic and optical signals are combined into a data vector and analyzed for the presence of bone fragment by electronic control 32.
[0110] A device similar to Figure 1 is illustrated in Figure 1A.
Petition 870170004653, of 23/01/2017, p. 51/73
36/39 realization, aperture 24 is transparent to near-infrared radiation and a broadband near-infrared source 62 illuminates the meat sample 20. A spectral chamber 50A forms a near-infrared radiation image reflected in a foreground containing a slit (not shown) to select a sample region approximately 0.5 mm wide. The near-infrared radiation that passes through the crack is collimated and dispersed by a network or prism (not shown) and an image is formed in a microbolometer or InGaAs matrix. Spectral data is transmitted to electronic control 32. The acoustic and optical signals are combined into a data vector and analyzed for the presence of bone fragments by electronic control 32.
[0111] Another similar embodiment is illustrated in Figure 1B where the meat sample 20 is transported on a conveyor belt 28 supported by the metal plate 22. In this embodiment, a cylinder 66 is mounted on a suspension system (not shown) maintains an outer cylindrical surface 66A of the cylinder in contact with and applies pressure to the meat sample 20. The cylinder 66 is filled with liquid 68 which provides acoustic and optical coupling between the cylinder 66 and a transducer 26A inside the cylinder 66. Also, a light source 52A, beam splitter 34 and chamber 50B are located on cylinder 66 so that the illumination of source 52A is directed through divider 34 and through transparent wall 66A with reflected light passing along the same path to the divider 34 which is angled to direct reflected light to the 50B camera. The acoustic and optical signals are combined into a vector of data and analyzed by the control system 32 for the presence of bone fragments by electronic control 38.
[0112] In Figure 1C another similar embodiment is shown, in cross-sectional view, where a sample of meat 20 rests on the conveyor belt 28. The metal plate 22 has upward facing edges to retain an aqueous solution 22C. A matrix of 30A transducers is mounted on the metal plate
Petition 870170004653, of 23/01/2017, p. 52/73
37/39
22.
[0113] Another similar embodiment is illustrated in Figure 2, where the detection device 10 has a housing 48, a chamber 50 and LEDs 52, 54 and 56. LED 52 is 570 nm, optional LED 54 is 630 nm and LED 56 is 720 nm. The LEDs have an associated diffuser 58 located above a cover plate 64. Air vent 66 and 68 remove the heated air from device 10 within housing 48. Below device 10 in sample space 20S is conveyor belt 70, sensors motion control 72 and 74 and fault port 76. Chicken sample 20 is also illustrated.
[0114] In some cases, only LEDs 570 and 720 nm are used. This system generates reflected amplitude, of very strong reflexivity for bone, cartilage, fat, skin, meat / muscle and membrane. Submersion eliminates specular reflection. Several samples were run to determine the effective reflectance. The presence of the optional 630 nm third LED can provide improved detection. Visual comparison of samples for computerized results from the double LED configuration, in comparison, were not as satisfactory as the comparison with the computerized results from the triple LED configuration. Normalization using 630 nm produced better results.
[0115] In Figure 3 another embodiment of the device 10 is illustrated, in which a laser 90 provides light through the linear scanning generator 92, which transforms a circular laser beam into a transverse linear beam, or a set of transverse linear beams. A steering mirror 94 deflects the beam to a beam divider 100 that sends the beam through a window 98 to the chicken sample 20 immersed in water or aqueous fluid 96. Window 98 is lowered below the water level of fluid 96 to avoid bubbles. The reflected Raman scattered light is passed back through window 98, beam divider 100 and filter 102 to the Fourier transform spectrometer 104 for amplitude measurement. Filter 102 is chosen to reject light
Petition 870170004653, of 23/01/2017, p. 53/73
38/39 at laser wavelength 90. Acoustic transducer 106 both emits and receives ultrasound.
[0116] In Figure 3A another embodiment of the device 10 is illustrated, similar to that of Figure 3, in which the chicken sample 20 is immersed in aqueous fluid 96. The sample is illuminated through the window 98, in sequence, by the LED 52 (570 nm), LED 54 (630 nm) and LED 56 (720 nm). The incident light is homogenized by the diffuser 58 and passes through the window 98. The reflected light is passed back through the window 98 and an image is formed by the chamber 104 for measuring the amplitude. The acoustic transducer 106 both emits and receives ultrasound.
[0117] Figure 4 shows an amplitude graph measured in standard deviations from time in milliseconds. The strong response in about 50 microseconds indicates bone.
[0118] In Figure 5, mean reflectance spectra for regions of a chicken breast identified as bone, muscle, membrane, fat and cartilage are given in the 420 to 720 nm range. The illustrated spectra were obtained by averaging pixels of the same type of tissue and dividing the average at each wavelength by the average at 630 nm. Normalization compensates for variations caused by the uneven surface of the chicken breast. Each type of tissue has a distinct medium spectrum.
[0119] In Figure 6, the spectral difference between bone and muscle is illustrated normalized by the sum of standard deviations in the range of 420 to 720 nm. This graph illustrates the relative diagnostic value of each wavelength to distinguish muscle and bone tissue. A higher ratio in absolute value indicates a greater probability of correctly distinguishing between muscle and bone at the level of an individual pixel. A small standard deviation (low variability) in the pixel population for a tissue type for a given wavelength increases the usefulness of that wavelength for diagnostic purposes. Note that the minimum near 630 nm,
Petition 870170004653, of 23/01/2017, p. 54/73
39/39 where muscle and bone are statistically indistinguishable, it is a useful reference point for normalization.

权利要求:
Claims (24)
[1]
1. Method for the detection of defects in a meat sample on a production line, CHARACTERIZED by the fact that it comprises the steps of:
emission of at least one wavelength of light in an area of said meat sample;
receiving reflected light from said area of said meat sample, measuring the amplitude of said reflected light;
and in a data processor, comparison of the amplitudes of said reflected light for each area of said meat sample by multivariate analysis; and determining, based on the referred multivariate analysis, the presence of surface defects in said meat sample.
[2]
2. Method, according to claim 1, CHARACTERIZED by the fact that a single wavelength of light is emitted and the reflected light is spread by Raman, comprising the additional step of dispersing said light spread by Raman through a selector of wavelengths to separate the light scattered by Raman into different wavelengths and, still, the additional step of measuring the amplitudes of said different wavelengths.
[3]
3. Method according to one of claims 1 to 2, CHARACTERIZED by the fact that the said single wavelength is almost monochromatic and selected from ultraviolet in the wavelength range of 200 to 220 nm and visible light and light infrared at 488, 515, 532, 594, 633, 635, 650, 660, 670, 780, 808, 830, 850, 980 and 1064 nm.
[4]
4. Method, according to claim 3, CHARACTERIZED by the fact that said single ultraviolet wavelength is in the range of wavelengths from 200 to 220 nm.
[5]
5. Method according to one of claims 1 to 4, CHARACTERIZED by the fact that the single wavelength is selected from visible light and
Petition 870170004653, of 23/01/2017, p. 56/73
2/6 infrared light at 488, 515, 532, 594, 633, 635, 650, 660, 670, 780, 808, 830, 850,
980 and 1064 nm.
[6]
6. Method according to one of claims 1 to 5, CHARACTERIZED by the fact that said at least one wavelength of light is broadband white light and comprises the additional step of dispersing said reflected light through a selector wavelengths to separate the reflected light into different wavelengths, and also the additional step of measuring the amplitudes of said different wavelengths.
[7]
Method according to one of claims 1 to 6, CHARACTERIZED by the fact that said at least one wavelength of light is a wavelength in the near infrared selected from the range of 900 to 2600 nm.
[8]
Method according to one of claims 1 to 7, CHARACTERIZED by the fact that said at least one wavelength of light comprises at least two separate wavelengths.
[9]
9. Method, according to claim 8, CHARACTERIZED by the fact that it comprises the steps of emitting said at least two wavelengths separated at different times for each wavelength and the steps of measuring said amplitudes of light reflected in distinct times for each wavelength.
[10]
10. Method according to claim 9, CHARACTERIZED by the fact that said at least two separate wavelengths comprise 620 to 640 and 720 to 760 nm.
[11]
11. Method according to claim 9, CHARACTERIZED by the fact that said at least two separate wavelengths comprise three wavelengths from 540 to 570, 620 to 640 and 720 to 760 nm.
[12]
12. Method according to one of claims 9 to 11,
Petition 870170004653, of 23/01/2017, p. 57/73
3/6
CHARACTERIZED by the fact that the steps of measuring the amplitudes of each separate wavelength are measured by the same array of pixels focal plane.
[13]
13. Method according to one of claims 9 to 11, CHARACTERIZED by the fact that the steps of measuring the amplitudes of each separate wavelength are measured by two separate pixel focal plane arrays, each focal plane array measuring one different wavelength.
[14]
14. Method according to one of claims 1 to 13, CHARACTERIZED by the fact that said data processor receives a plurality of light amplitudes corresponding to a sample area of said meat sample, said data processor generates vectors of n-dimensional data from light amplitudes and compares said data vectors with a calibration set generated by multivariate analysis, to determine whether they correspond to bone, cartilage, fat, meat or skin, or contaminant for each sample area of the sample, when unwanted matter is identified, a logic signal is sent to trigger a pass-fail gate stopping the sample, otherwise, no logic signal is sent.
[15]
15. Method, according to claim 14, CHARACTERIZED by the fact that, in addition, said data processor identifies the amplitudes of the areas neighboring said sample area, directly adjacent and diagonally, for each wavelength; calculates the gradient through said sample area and said adjacent areas for all wavelengths; and generates n-dimensional data vectors from said gradients and amplitudes; and compares said data vectors with a calibration set generated by multivariate analysis, which additionally determines the presence of edges between sample areas, when undesired matter is identified, a logic signal is sent to trigger a failure door by stopping the sample, otherwise, no logic signal is sent.
Petition 870170004653, of 23/01/2017, p. 58/73
[16]
16. Method according to one of claims 1 to 15, CHARACTERIZED by the fact that it comprises the additional steps of:
emission of at least one ultrasound frequency in an area of said meat sample receiving ultrasound returned from said meat sample;
measurement of amplitudes and travel times of the returned ultrasound;
comparison of the amplitudes and travel times of the referred ultrasound returned to each area of the meat sample by multivariate analysis;
determination from the referred multivariate analysis of the presence of superficial and internal defects in said meat sample.
[17]
17. Method, according to claim 16, CHARACTERIZED by the fact that it comprises the additional steps of comparing the amplitudes of said reflected light for each area of said meat sample and the amplitudes and travel times of said ultrasound returned for each area of said meat sample by multivariate analysis and determination from said multivariate analysis of the presence of surface and internal defects in said meat sample.
[18]
18. Method according to one of claims 16 to 17, CHARACTERIZED by the fact that said data processor receives a plurality of acoustic amplitudes and travel times of said amplitude that corresponds to a sample area of said meat sample, said data processor compares said amplitudes with standard amplitudes to determine the presence of bone in said sample, when bone is present, a logic signal is sent to trigger a fault gate stopping the sample, otherwise, no logic signal is sent .
[19]
19. Method, according to one of claims 16 to 18, CHARACTERIZED by the fact that said data processor centralizes in the
Petition 870170004653, of 23/01/2017, p. 59/73
5/6 averages and normalizes those amplitudes to the standard deviation for each wavelength.
[20]
20. Method according to one of claims 1 to 19, CHARACTERIZED by the fact that each type of LED of the same wavelength has an energy converter controlled by a switching circuit, in combination with a logic processor, with which each type of LED of the same wavelength is separately strobed.
[21]
21. Method for the detection of defects in a meat sample on a production line, CHARACTERIZED by the fact that it comprises the steps of:
emission of at least one ultrasound frequency in an area of said meat sample;
receiving ultrasound returned from said meat sample;
measurement of amplitudes and travel times of the returned ultrasound;
comparison of the amplitudes and travel times of the referred ultrasound returned to each area of the meat sample by multivariate analysis;
determination from the referred multivariate analysis of the presence of superficial and internal defects in said meat sample.
[22]
22. Method, according to claim 21, CHARACTERIZED by the fact that it comprises the additional steps of comparing the amplitudes of said reflected light for each area of said meat sample and the amplitudes and travel times of said ultrasound returned for each area of said meat sample by multivariate analysis and determination from said multivariate analysis of the presence of surface and internal defects in said meat sample.
[23]
23. Method according to one of claims 21 to 22, CHARACTERIZED by the fact that said data processor receives a plurality of acoustic amplitudes and travel times of said amplitudes
Petition 870170004653, of 23/01/2017, p. 60/73
6/6 that corresponds to a sample area of said meat sample, said data processor compares said amplitudes with standard amplitudes to determine the presence of bone in said sample, when bone is present, a logical signal is sent to trigger a fault port stopping the sample, otherwise, no logic signal is sent.
[24]
24. Method according to one of claims 21 to 23, CHARACTERIZED by the fact that said data processor centralizes in the mean and normalizes said amplitudes to the standard deviation for each wavelength.

类似技术:

---

公开号 | 公开日 | 专利标题

---

BR112017001407A2|2019-11-12|Method and Device for Bone Scanning in Meat

---

US10945415B2|2021-03-16|System and method for monitoring eggs during incubation

---

JP5235586B2|2013-07-10|Biological information processing apparatus and biological information processing method

---

JP2020173260A5|2021-01-07|

---

US9750413B2|2017-09-05|Massively parallel diffuse optical tomography

---

JP2014209094A|2014-11-06|Information acquisition device and information acquisition method for acquiring information on sample using terahertz wave

---

JP5183381B2|2013-04-17|Measuring apparatus and measuring method

---

WO2016059946A1|2016-04-21|Spectroscopic measurement method and spectroscopic measurement device

---

US20130160557A1|2013-06-27|Acoustic wave acquiring apparatus

---

JP5691687B2|2015-04-01|Inspection device

---

CN102803952A|2012-11-28|Hyperspectral identification of egg fertility and gender

---

CA2196187A1|1996-02-15|Apparatus and method for the optical characterization of the structure and composition of a light scattering sample

---

WO2016080442A1|2016-05-26|Quality evaluation method and quality evaluation device

---

JP2010133934A|2010-06-17|Surface measuring apparatus having two measuring unit

---

US20160058295A1|2016-03-03|Photoacoustic wave measurement apparatus and photoacoustic wave measurement method

---

JP2012189390A|2012-10-04|Hair detector

---

US20200096435A1|2020-03-26|Observation container and microparticle measurement device

---

KR20180010589A|2018-01-31|Antibiotics Suitability Test Apparatus With Chaotic Sensor

---

Qin et al.2006|Measurement of the optical properties of apples using hyperspectral diffuse reflectance imaging

---

JP5575293B2|2014-08-20|Subject information acquisition apparatus and subject information acquisition method

---

KR20190032320A|2019-03-27|Antibiotics Suitability Test Apparatus With Chaotic Sensor

---

US20140125969A1|2014-05-08|Apparatus and methods for performing optical tomography on dosimeters for calibrating radiotherapy equipment

---

CN110456010A|2019-11-15|A kind of method of lossless quick detection egg freshness

---

KR20160065342A|2016-06-09|Apparatus and method for fluorescence imaging of regions of interest in a samll animal

---

RU2005122997A|2007-01-27|METHOD FOR DETERMINING THE OPTICAL CHARACTERISTICS OF THE BRAIN TISSUES AND THE DEVICE FOR ITS IMPLEMENTATION

同族专利:

---

公开号 | 公开日

---

US20200217831A1|2020-07-09|

---

WO2016011548A1|2016-01-28|

---

MX2017001034A|2017-10-31|

---

AU2015292225A1|2017-03-16|

---

RU2017105393A3|2019-02-18|

---

CA2968706C|2021-09-28|

---

JP2017534057A|2017-11-16|

---

JP2020173260A|2020-10-22|

---

AU2015292225B2|2020-06-25|

---

RU2017105393A|2018-08-22|

---

US20170205385A1|2017-07-20|

---

CN107003253A|2017-08-01|

---

CN107003253B|2020-10-16|

---

RU2705389C2|2019-11-07|

---

EP3198262A4|2018-07-25|

---

EP3198262A1|2017-08-02|

---

CA2968706A1|2016-01-28|

引用文献:

---

公开号 | 申请日 | 公开日 | 申请人 | 专利标题

---

---

US3877818A|1974-01-28|1975-04-15|Us Agriculture|Photo-optical method for determining fat content in meat|

---

DE2728717C2|1977-06-25|1983-11-10|Pfister Gmbh, 8900 Augsburg|Method and device for the non-contact determination of quality features of a test object of the meat product category, in particular a carcass or parts thereof|

---

IS1279B6|1983-06-13|1987-07-07|Fmc Corporation|Quality control method for fish, bovine, swine and poultry production|

---

US5206699A|1988-05-06|1993-04-27|Gersan Establishment|Sensing a narrow frequency band of radiation and gemstones|

---

GB2219394B|1988-05-06|1992-09-16|Gersan Ets|Sensing a narrow frequency band of radiation and examining objects or zones|

---

US5239180A|1990-02-02|1993-08-24|Boston Advnaced Technologies, Inc.|Laser systems for food analysis based on reflectance ratio detection|

---

WO1991011136A1|1990-02-02|1991-08-08|Boston Advanced Technologies, Inc.|Systems for material analysis based on reflectance ratio detection|

---

JPH04166063A|1990-10-29|1992-06-11|Nisshin Denshi Kogyo Kk|Method and apparatus for sensing and removing foreign material in ground fish meat|

---

WO1993005359A1|1991-09-06|1993-03-18|Commonwealth Scientific And Industrial Research Organisation|Measurement method and apparatus|

---

FR2780790B1|1998-07-03|2000-08-18|Vitreenne Abattage|METHOD AND DEVICE FOR PREDICTING THE TENDERNESS OF MEAT ON THE PROCESSING SITE USING BIOLOGICAL AND / OR PHYSICO-CHEMICAL INFORMATION AND OPTICAL MEASUREMENTS IN THE FIELD OF VISIBLE AND NEAR INFRARED|

---

US6129625A|1999-08-26|2000-10-10|Townsend Engineering Company|Method of trimming a meat portion by ultrasonic and electronic analysis|

---

AUPR068500A0|2000-10-11|2000-11-02|Australian Food Industry Science Centre|Animal carcass splitting|

---

US6992771B2|2001-11-28|2006-01-31|Battelle Memorial Institute|Systems and techniques for detecting the presence of foreign material|

---

US7460227B1|2004-12-30|2008-12-02|The United States Of America As Represented By The Secretary Of Agriculture|Method to detect bone fragments during the processing of meat or fish|

---

CA2550390C|2005-09-16|2014-10-21|The Regents Of The University Of Michigan|Method and system for measuring sub-surface composition of a sample|

---

EP2088869B1|2006-10-06|2010-05-19|Nordischer Maschinenbau Rud. Baader GmbH + Co. KG|Method and device for processing fish, poultry, or other meat products transported in multitude along a processing line|

---

GB2446822A|2007-02-23|2008-08-27|Enfis Ltd|Quality control of meat products using optical imaging|

---

CN100480680C|2007-05-22|2009-04-22|浙江大学|Multiple spectrum meat freshness artificial intelligence measurement method and system|

---

US8351672B2|2007-09-26|2013-01-08|Industry Vision Automation Corp.|Machine imaging apparatus and method for detecting foreign materials|

---

JP4640492B2|2008-10-27|2011-03-02|パナソニック電工株式会社|Bone density measuring device|

---

BRPI1006738A8|2009-03-30|2019-01-29|3M Innovative Properties Co|Optoelectronic method and device for monitoring an analyte in an atmosphere|

---

BR112012033200A2|2010-06-25|2016-12-06|Cellmark Theranostics Llc|method for analyzing biological specimens by spectral imaging|

---

CN102564964B|2011-12-29|2014-07-30|南京林业大学|Spectral image-based meat quality visual non-contact detection method|JP6425178B2|2016-09-20|2018-11-21|九州電子技研株式会社|Raman scattered light detection device and Raman scattered light detection method|

---

US20190346307A1|2016-11-30|2019-11-14|Corning Incorporated|Spectral imaging systems and methods of generating spectral image data using the same|

---

KR101925502B1|2017-01-12|2019-02-27|서강대학교산학협력단|Method for determining concentration and pressure of respective gas of multi-gas|

---

JP6480979B2|2017-05-10|2019-03-13|ファナック株式会社|Measuring device|

---

WO2018232752A1|2017-06-23|2018-12-27|深圳前海达闼云端智能科技有限公司|Method, apparatus and device for detecting substance|

---

US10713563B2|2017-11-27|2020-07-14|Technische Universiteit Eindhoven|Object recognition using a convolutional neural network trained by principal component analysis and repeated spectral clustering|

---

IT201800003091A1|2018-02-27|2019-08-27|Milan Fish S R L|METHOD AND APPARATUS FOR THE INSPECTION OF FOOD PRODUCTS, IN PARTICULAR FISH PRODUCTS|

---

CN108872140B|2018-05-09|2021-01-19|塔里木大学|Method and device for monitoring quality of red dates outdoors|

---

WO2019234054A1|2018-06-06|2019-12-12|Marel Iceland Ehf.|A method of providing feedback data indicating quality of food processing performed by an operator|

---

CA3108903A1|2018-08-07|2020-02-13|Britescan, Llc|Portable scanning device for ascertaining attributes of sample materials|

---

JP6748168B2|2018-09-21|2020-08-26|株式会社ミヤザワ|Article inspecting apparatus having substantially circular side surfaces, and article reversing apparatus|

---

US11137413B2|2018-11-09|2021-10-05|Embedded Designs, Inc|Method of determining conveyor oven belt speed|

---

GB201908806D0|2019-06-19|2019-07-31|Signature Robot Ltd|Surface recognition|

---

CN110208212B|2019-07-04|2021-06-18|中南林业科技大学|Near infrared spectrum omnibearing nondestructive detection device and control method|

---

CN110320175B|2019-07-04|2021-07-13|中南林业科技大学|Near infrared spectrum detection device and control method|

---

CN110243805B|2019-07-30|2020-05-22|江南大学|Fish bone detection method based on Raman hyperspectral imaging technology|

---

JP6778451B1|2020-01-10|2020-11-04|日本分光株式会社|Foreign matter analysis method, foreign matter analysis program and foreign matter analyzer|

---

CN112505049B|2020-10-14|2021-08-03|上海互觉科技有限公司|Mask inhibition-based method and system for detecting surface defects of precision components|

---

CN113074627A|2021-03-12|2021-07-06|中国科学院生物物理研究所|Imaging method and device of direct electronic detection camera and computer equipment|

法律状态:
2020-02-04| B06U| Preliminary requirement: requests with searches performed by other patent offices: procedure suspended [chapter 6.21 patent gazette]|

---

2021-10-13| B350| Update of information on the portal [chapter 15.35 patent gazette]|

优先权:

---

申请号 | 申请日 | 专利标题

---

US201461999206P| true| 2014-07-21|2014-07-21|

---

PCT/CA2015/050678|WO2016011548A1|2014-07-21|2015-07-21|Method and device for bone scan in meat|

---