DE112014004052T5 - NMR Sample Analysis - Google Patents

Abstract

A method of identifying animal tissue using a benchtop NMR spectrometer with a magnetic field strength between 1.0 and 2.5 Tesla is provided. Triglyceride molecules from an animal tissue sample are dissolved using a solvent to obtain a sample. The sample is then scanned using a benchtop NMR spectrometer and the resulting signal is converted to a frequency spectrum. The spectrum is then analyzed and characteristic data for the animal tissue is output, the characteristic data indicating the identity of the animal tissue.

Description

Field of the invention

The present invention relates to a method for animal tissue analysis using a "benchtop" NMR spectrometer with a permanent magnet.

Background of the invention

The issue of meat contamination has recently attracted public attention. In January 2013, it was found that a number of European supermarkets sold meat products advertised as containing beef, but which was actually identified as undeclared horsemeat, in some cases up to 100% of the meat content. Similar cases were discovered in which undeclared pork was found, which was advertised as beef balls. The presence of undeclared pork is especially annoying for those who deliberately do not eat pork. In the light of this scandal, there has been a growing need for testing the origin of meat products.

It should first be noted that the terms "meat" and "animal tissue" are used interchangeably throughout this document and both are intended to refer to the meat of a nonhuman animal (including mammals, fish, birds, reptiles, molluscs, crustaceans, etc.). ).

The standard method for testing meat products is a polymerase chain reaction (PCR) which amplifies the DNA in a sample, allowing the identification of a contaminating species.

Running the test requires skilled and expensive equipment, and the fastest time to get results is typically between 24 and 48 hours. This process also represents a significant cost factor since the cost is more than £ 100 (about € 127) per sample. Therefore, there is a need for a cheaper and faster solution for testing animal tissues.

Summary of the invention

• (i) Inkontaktbringen des Tiergewebes mit einem Lösungsmittel, um Triglyceridmoleküle aus dem Tiergewebe in dem Lösungsmittel aufzulösen, um eine Probe zu erhalten;
• (ii) Platzieren der Probe in den Probenhalter des NMR-Spektrometers;
• (iii) Anwenden einer Hochfrequenz(HF)-Pulssequenz auf die Probe und Detektieren eines freien Induktionszerfall(FID)-Signals unter Verwendung des NMR-Spektrometers;
• (iv) Umwandeln des detektierten FID-Signals in den Frequenzraum, um ein Frequenzspektrum mit einer Triglyceridsignatur zu erhalten, und
• (v) Analysieren der Triglyceridsignatur und Ausgeben von charakteristischen Daten für das Tiergewebe, wobei die charakteristischen Daten die Identität des Tiergewebes anzeigen.

According to the first aspect of the present invention, there is provided a method of identifying animal tissue using an NMR spectrometer with a permanent magnet having a magnetic field intensity at a sample holder of the spectrometer between 1.0 and 2.5 Tesla, the method comprising:

• (i) contacting the animal tissue with a solvent to dissolve triglyceride molecules from the animal tissue in the solvent to obtain a sample;
• (ii) placing the sample in the sample holder of the NMR spectrometer;
• (iii) applying a radio frequency (RF) pulse sequence to the sample and detecting a free induction decay (FID) signal using the NMR spectrometer;
• (iv) converting the detected FID signal into the frequency domain to obtain a frequency spectrum with a triglyceride signature, and
• (v) analyzing the triglyceride signature and outputting characteristic data for the animal tissue, the characteristic data indicating the identity of the animal tissue.

The present invention provides an inexpensive, fast and effective method of analyzing the nature or origin of non-human animal tissues. Fats are stored in animals predominantly as triglycerides. Triglyceride molecules consist of a glycerol molecule (called the "backbone"), which is bound to three fatty acid molecules via esterification. Many types of fatty acids occur in nature, and various fatty acid molecules can be attached to the three respective sites on the glycerol backbone. The presence of triglycerides with fatty acids having different saturations (i.e., monounsaturated, polyunsaturated, or saturated) varies between different species of animals and, to a certain extent, between different cuts of meat, that is, tissues of different parts of the same type of animal. A fatty acid is saturated if there are only single bonds between adjacent carbon atoms in the chain, monounsaturated if a double bond is present, and polyunsaturated if more than one double bond is present. The present Applicant has recognized that by analyzing the triglyceride components in a meat sample, a simpler solution than DNA sequencing after PCR for determining the origin of the meat can be found.

Nuclear magnetic resonance (NMR) spectroscopy provides a method to study the electronic environment of NMR-active nuclei and the consequent structure of the molecules to which these nuclei are bound. NMR spectroscopy involves bringing a sample of NMR-active nuclei (such as hydrogen-1) into a static external magnetic field B ₀ generated by an NMR spectrometer. An electromagnetic radio frequency (RF) pulse is then applied from the spectrometer to the sample at a frequency which is characteristic of the NMR active nuclei of interest. The energy delivered by this pulse is 'absorbed' by these nuclei and re-emitted as radio waves at the resonant frequency of the active nuclei. This signal is known as Free Induction Decay (FID) signal and is detected by the spectrometer and analyzed to infer the chemical structure of this sample.

In the context of meat sampling, we have found that the relative triglyceride compositions differ significantly between different species of animals and between different cuts of meat. We have further recognized that the large amount of hydrogen nuclei and their different chemical environments in triglycerides make them suitable candidates for analysis using NMR. We have further recognized that these triglycerides can be extracted from animal tissue samples by a simple solvent-based process with few impurities. Triglyceride samples can therefore be obtained from an unknown meat of unknown identity and analyzed using NMR to obtain characteristic data for the animal tissue indicating the identity of the animal tissue.

The term "identity" is intended to refer to the origin, type or species of meat as far as the source animal and, in some cases, the particular type or cut of meat of that animal is concerned. In the case that the sample is made using animal tissue, that is, actually a mixture of different animal tissues, multiple identities may be indicated by the characteristic data. Preferably, however, the relative weights of the identities may also be indicated.

The identity may be output as a tag, for example, the word "pork" is displayed on a monitor, or it may be indicated in a more abstract manner by e.g. As a sound, a light display or a mechanical output is triggered, or that raw data are output, as long as it was possible to easily identify the identity of the animal tissue with this issue. In case the device is configured to detect only an animal tissue identity, the output of the characteristic data may include a simple yes / no answer indicating whether the triglyceride signature shows a predetermined or preset response.

In the frequency domain analysis, the portion of the free induction decay signal originating only from the triglyceride content is referred to herein as the "triglyceride signature". The triglyceride signature is preferably the response of the sample analyzed in the frequency domain to at least two of the olefinic, bis-allylic and glyceride regions (defined below) of the spectrum, and most preferably each of these three regions. In some cases, the signature may additionally contain the reaction in the linoleic region.

The triglyceride signature varies between different triglycerides because of the different saturation levels of the fatty acids in the triglyceride. For example, a hydrogen nucleus in the vicinity of a single double bond resonates between adjacent carbon atoms at a different frequency than one surrounded by single bonds. This is because the resonant frequency of the nucleus is proportional to the magnetic field to which this nucleus is exposed, and this magnetic field itself is affected by the electron density in the surrounding region. Circulating electrons create a small magnetic field, which typically acts in the opposite direction to the external magnetic field; the higher the electron density, the weaker the magnetic field at the core. Double bonds have a higher electron density than single bonds and thus protect the external magnetic field to a higher degree. Hydrogen atoms in a fatty acid bonded to a carbon atom having a single bond and a double bond between the adjacent two carbon atoms in the chain (-CH = CH-) resonate in a particular region of the frequency spectrum known as the olefinic region. Similarly, hydrogen atoms located in the region of two double bonds (= CH-CH ₂ -CH =) resonate in another region of the spectrum known as the bis-allylic region. Likewise, the two hydrogen pairs in a glycerol molecule singly attached to an 'end' carbon resonate in another region, different from the olefinic and bis-allylic regions, and known as the glyceride region.

Low field strength "benchtop" NMR spectrometers have recently been developed as a cheaper, more compact and portable device for carrying out certain types of NMR experiments that have otherwise been performed in large NMR spectrometers with superconducting magnets. Instead of requiring a dedicated infrastructure, trained personnel, and extensive installations, benchtop NMR spectrometers use permanent magnets that generate a magnetic field between 1.0 and 2.5 Tesla on the sample holder of the spectrometer. They are generally standalone units that can be placed and moved directly on a bench or other surface as needed. These Spectrometers are also typically much easier to use by first-time users as they are used for some of the simpler NMR experiments. An example of such a benchtop NMR spectrometer is the Pulsar ^™ supplied by the Oxford Instruments Group. The present invention should provide an alternative, simpler and cheaper solution than current meat sampling. In the present invention, magnetic fields having a field strength of less than 1 Tesla are not desirable because it is difficult to distinguish between the peaks in the triglyceride signature in such weak fields. Also, a magnetic field strength above 2.5 Tesla is not desirable because the spectrometer is no longer Benchtop instrument at such high field strengths because superconducting magnets are needed.

The accuracy with which the triglyceride signature can be analyzed is an important consideration in a practical implementation of the method. An effective approach in step (v) involves integrating an olefinic region of the triglyceride signature. Typically, the chemical shift of the olefinic region is between 5.0 and 5.6 ppm. Preferably, step (v) also includes integrating a bis-allylic region of the triglyceride signature. Typically, the chemical shift of the bis-allylic region is between 2.5 and 3.0 ppm. Further preferably, if practicable, step (v) comprises integrating a linoleine region of the triglyceride signature. Typically, the chemical shift of the linolenic region is between 0.96 and 1.02 ppm. These chemical shift values can be treated as boundaries for the above integrals. Although the actual reaction of triglycerides varies for different animal tissues (i.e., different intensities are registered at different chemical shifts in the triglyceride signature), the chemical shift regions in which the triglycerides can be detected are common within the animal tissues. By comparing the integrals obtained for these regions, one can estimate the level of saturated, monounsaturated and polyunsaturated fat in the animal tissue. These data may indicate the type or origin of animal tissue. For example, characteristic peaks in these regions can also be identified by integrating portions of these regions to provide a more detailed analysis. The presence of a characteristic peak, usually associated with only one particular tissue type, may be particularly useful in identifying a trace contaminant of this tissue type in a meat composite product (or animal tissue), that is primarily of a different type.

Another useful method of analyzing the triglyceride signature is a method of principal component analysis. This method may be typically used in place of the peak integration and comparison method, but it may be carried out in combination with this method. Principal component analysis is a standard data processing technique, and we have found that it is particularly effective in distinguishing between different tissue types, especially between different species of animals. In addition, the major components may preferably be calibrated to determine levels of contamination or adulteration using a method known as major component regression. We have found that principal component analysis is a very effective technique in the present application because of the specific regions of interest in the chemical shift spectrum and the strong relationships that exist between the triglyceride content of the tissue and the animal species or tissue type from which they are derived , Such relationships are not always easy, and in such cases, principal component analysis is an effective technique for establishing more complex relationships. Other statistical analysis processes such. For example, neural network analysis may also be used in principle, although some such techniques may require a well-structured set of trial data determined by trained personnel.

The spectrometers used in the present invention have relatively small magnetic field strengths; typically, this means that there are higher noise levels in the signal compared to a higher field device. To handle this, the method of step (iii) preferably comprises comparing a set of peaks in the signal intensity at the beginning of the periodically acquired FID signal with a set of peaks at the end of the periodically acquired FID signal and computing a Signal-to-noise ratio from this comparison. The signal-to-noise ratio increases with the square root of the number of scans (i.e., the number of individual FID signals which are summed up). The signal-to-noise ratio can be measured and then an appropriate number of scans can be calculated and performed to achieve a desired overall signal-to-noise ratio / data quality in the final data. Typically, step (iii) is repeated according to the calculated signal-to-noise ratio to achieve a target signal-to-noise ratio for the data used in the following analysis.

Preferably, the output characteristic data shows an animal tissue candidate for the one used to obtain the sample Animal tissue on. In the case that the spectrometer has been set up or 'calibrated' to assign several different triglyceride signatures to different animal tissues, the characteristic data can be used to suggest an identity candidate of the animal tissue used to make the sample. This animal tissue candidate (or identity candidate) may be the particular species of animal from which the tissue was obtained, or it may indicate the particular cut or quality of the meat. This information may be output as a marker, for example the word "pork" is displayed on a monitor, or it may be indicated in a more abstract manner, e.g. As a sound, a light display or a mechanical output is triggered. In case the signature analysis indicates the presence of more than one animal tissue, multiple animal tissue identity candidates may be displayed, and preferably their relative amounts may be output (e.g., 80% beef, 20% horse).

Preferably, step (iv) further comprises performing a phase correction on the frequency spectrum. With ideal (theoretical) NMR detection, all the frequency components contained in the FID signal begin exactly in phase, and phase correction on the frequency spectrum is not necessary. However, in typical data acquisition, the frequency components are shifted from one another by unknown and different amounts, and phase correction is necessary.

The selection of the solvent to be used is important because it is effective in extracting the triglyceride molecules, is ineffective in extracting other molecules that can cause noise in the signal, and does not itself have an NMR spectrum that interferes with the analysis of the triglyceride signature. Preferably, the solvent is deuterated chloroform. Chloroform is an effective solvent for organic, carbon-based, hydrophobic molecules such. B. triglycerides. When mixed with meat, chloroform dissolves fat molecules known as triglycerides from the meat sample and other ingredients only slightly. Other types of fat (such as free fatty acids, phospholipids, monoglycerides or diglycerides) in the solvent are less than 1 percent. Typically, deuterated chloroform is used in NMR instead of chloroform to avoid 'falsifying' the spectrum by the appearance of a chloroform peak at 7.26 ppm. Deuterated chloroform has a similar composition to chloroform, but the hydrogen atom in chloroform is replaced by a deuterium. Deuterium resonates at a different frequency, that is, is not stimulated by the RF pulse applied by the spectrometer with typical hydrogen frequencies. In practice, however, a sample of deuterated chloroform is never deuterated to 100, and thus a reduced peak still occurs at 7.26 ppm, corresponding to the trace presence of chloroform. Fortunately, this is not a problem in this context because this signal does not overlap the triglyceride signature. Other solvents such. For example, toluene or benzene may also be used in principle, but have been found to be less effective at extracting triglycerides and may produce undesirable peaks in the frequency spectrum in the regions of interest.

Preferably, the carrier frequency of the RF pulse is between 40 and 110 MHz. This range is approximately the same as that needed to stimulate hydrogen nuclei in a magnetic field between 1.0 and 2.5 Tesla. This frequency range is sufficient to obtain a triglyceride signature which can be efficiently analyzed and yet supplied by a spectrometer which has small dimensions and relatively low cost.

Preferably, step (iv) further comprises integrating a glyceride region of the frequency spectrum. Preferably, the region indicating the glycerol content in the triglyceride molecules has a chemical shift between 4.0 to 4.5 ppm. The frequency spectra are also preferably adjusted based on a region of the spectra indicating the glycerol content in the triglyceride molecules. This glycerol signal is present in all samples containing triglyceride and has a common spectral response for each sample. It is therefore an appropriate choice with which spectra of different samples can be matched.

Preferably, step (iv) further comprises normalizing the spectrum based on the integral of the glyceride region. The intensity of the signal indicative of glycerol content is proportional to the total triglyceride content in the sample, and thus preferably the spectrum is always scaled such that the intensity of the peaks derived from the glycerol "triglyceride backbone" is the same in each of the spectra. This process explains the variation in the amount of triglyceride molecules between the samples.

Advantageously, step (iv) is performed using a fast Fourier transform technique. This is a mathematical technique that is performed and used using software (such as MNova ^™ ) to significantly reduce the processing time for performing a Fourier transform on the data obtained.

Preferably, the animal tissue is first cut small, finely chopped or otherwise homogenized prior to step (i). This ensures that the lean and fatty parts of the meat are well mixed with each other so as to reduce the variation in triglyceride content between different samples from the same source.

Advantageously, steps (i) to (iv) are initially performed using known animal tissues of known type so as to obtain target triglyceride signature data for use in the analysis of step (v). This approach effectively calibrates the spectrometer, and in particular the analysis method used, so that certain triglyceride signatures can be assigned to particular animal tissues. The characteristic data from a variety of different animal tissues can be entered into a database so that when the spectrometer is later used for unknown animal tissues, the database can be used to find a likely candidate for the animal tissue from which the sample is was obtained. The animal tissues for which the device has been calibrated to recognize the triglyceride signatures form a list of potential candidate animal tissue candidates. Alternatively, if the signature does not match the stored data of the database, the user may be informed so that the animal tissue may be subjected to further analysis, e.g. B. a DNA sequencing after PCR can be subjected. In this case, the identity of the animal tissue used to generate the sample is displayed as unknown but no animal tissue candidate is output. A database may be formed, which may include the spectral data itself, or it may contain information gathered therefrom after any analysis has been performed on it.

Advantageously, the results stored in the database can be used to define triglyceride signature information that is characteristic of a known meat type. This may be the case if a number of meat samples have been obtained from the same carcass of known origin. Variations in fat content, and hence triglyceride signature, in this carcass can be explained by constructing a database of information representative of the triglyceride signatures obtained from the various meat samples for that carcass. An average expected triglyceride signature contains triglyceride signature information that is characteristic of a known type of meat, such as e.g. Where the resonant frequency peaks can be expected, as well as the intensity, shape and width of these peaks. This database can be used to perform a principal component analysis procedure. In some cases, the NMR spectrometer can not be used to perform the analysis step itself, for example, the spectrometer can be connected to a computer which can operate with suitable software to perform this analysis. Alternatively, the analysis and database of known characteristic triglyceride signatures with relevant animal tissue candidate markers may be present in a remote control device accessible via an Internet connection.

Brief description of the drawings

Some examples of a method of analyzing animal tissue will now be described with reference to the accompanying drawings. Show it:

1 a perspective of a benchtop NMR spectrometer associated with a computer for use according to the discussed examples;

2 a flowchart illustrating the process steps that are performed for calibrating an NMR spectrometer according to the examples described;

3 an example of a frequency spectrum comprising a triglyceride signature obtained from beef according to the calibration example;

4 a flowchart illustrating the method steps for analyzing animal tissue according to a first example of the invention;

5 Fig. 12 is a graph obtained by a method of principal component analysis of triglyceride signatures according to a second example of the invention; and

6 an example of a frequency spectrum comprising a triglyceride signature obtained from beef and pork according to a third example of the invention.

Detailed description

To assist in understanding the invention, the following discussion includes a number of examples of the implementation of the present invention. An example is given first for calibrating a benchtop NMR spectrometer with a known animal tissue type. Examples of identifying unknown tissues are then discussed, along with typical experimental data that can be obtained.

contraption

1 is a representation of a system 1 that one with a computer 10 connected benchtop NMR spectrometer 2 for use according to the examples of the invention to be discussed. The spectrometer 2 includes a sample holder 5 , shown as a cylindrical cavity in the spectrometer 2 , and two NdFeB permanent magnets 7a and 7b , The magnets 7a and 7b have opposite north and south poles, so that a static, primary magnetic field B ₀ between 1.0 and 2.5 Tesla at the sample holder 5 acts. In use, a sample (not shown) is placed in the sample holder 5 in the primary magnetic field B ₀ . Two magnets are not necessarily needed, and other magnet geometries are also possible, such. B. a cylindrical magnet.

The NMR spectrometer also includes a set of radio frequency (RF) coils 4 operating between 40 and 110 MHz to generate RF magnetic fields perpendicular to the direction of the B ₀ field, and a set of B ₀ auxiliary coils (not shown) which are used to compensate for the inhomogeneities of the B ₀ field compensate. The RF coils 4 are capable of detecting RF-FID signals from a sample in the sample holder 5 as well as applying RF fields to the sample. Other features such. As heaters, a vacuum chamber and vibration dampers, etc., which are typically found in benchtop NMR spectrometers may also be provided, but these are not shown here for clarity.

The benchtop spectrometer 2 is a stand-alone unit that can be placed and moved directly on a bench or other surface (unlike superconducting NMR spectrometers). The spectrometer 2 is with a computer 10 connected by the spectrometer 2 analyzed results. Alternatively, this analysis may be performed remotely by a server (not shown), for example via Internet access. The computer 10 includes an indication to the user to view the results of the analysis.

Calibrating the spectrometer

2 is a flowchart illustrating the process steps for calibrating the spectrometer 2 with high quality triglyceride signature data according to a first example. In the present case, the method is to be used for the subsequent examination of ground beef for the presence of lower quality spare meat.

In step 101 A 50g serving of minced meat of a guaranteed origin is obtained and further finely chopped so that the lean and fat parts are mixed together. Optionally, the source or DNA of each portion is verified using PCR techniques to ensure that the tissue is identified. A 3 g aliquot is obtained and placed in a vial of 1 mL deuterated chloroform. The vial is shaken and allowed to stand for 5 minutes to dissolve the triglyceride molecules from the meat. The remaining solid animal tissue is then filtered out of the solvent and the liquid solvent, now containing triglyceride molecules from the meat, is transferred to a 5 mm NMR sample tube using a pipette to form a calibration sample.

In step 102 the calibration sample is placed in a sample holder 5 Benchtop NMR spectrometer 2 brought. In the present case, the NMR spectrometer has 2 a permanent magnet system ( 7a and 7b ), which generates a magnetic field of 1 Tesla at a target location at which the sample holder is in use 5 located. The magnet system generates a magnetic field B ₀ , which is very homogeneous, so that each part of a sample in the sample holder is exposed to a substantially identical magnetic field strength with a common direction. Once the sample is in the NMR spectrometer 2 For example, by stabilizing the temperature and, if necessary, applying "shimming" of the magnetic field, an NMR signal is obtained from the calibration sample.

In step 103 is through the RF coils 4 of the spectrometer 2 applied a "90 degree" RF pulse to the sample. This is accomplished using a 50 MHz radio frequency signal that is pre-calibrated with respect to the known magnetic field strength to which the sample is exposed so as to cause the hydrogen nuclei in the sample to be deflected out of alignment with the applied magnetic field and precess around the main field direction. This precession decreases and the resulting free induction decay (FID) back into alignment causes a release of energy through the RF coils 4 of the spectrometer is detected as FID signal. The FID signal is transmitted through the spectrometer 2 digitized and sent to the computer 10 communicates for analysis.

The signal-to-noise ratio in the detected FID signal is determined in step 104 calculated using a set of peaks at the beginning and at the end of the signal. In the present example, the NMR spectrometer detects 32,000 points during the acquisition time of the signal. In this example, an average of the first and last five percent of the points in the signal (ie, 1,600) is taken as the signal-to-noise ratio. It turned out that the signal-to-noise ratio increases with the square root of the number of times that the detection is repeated using the applied RF signal (known as "scans"), at least under 100 scans, and so the total number of scans necessary to obtain a target signal Noise Ratio are needed, calculated and then run through the spectrometer. The target signal-to-noise ratio varies between different spectrometers and the required quality of the spectrum, although for a particular spectrometer and a particular experiment an appropriate number of scans can be easily derived by the operator.

The resulting FID signals from the scans are detected by software on the computer 10 processed and combined together to form a composite FID signal.

In step 105 For example, the composite FID signal is converted into the frequency domain by suitable frequency analysis software so as to obtain a frequency spectrum. A suitable Fast Fourier Transform software may be used to achieve this to reduce the time required to perform this conversion into frequency space.

When the time domain FID signal is Fourier transformed, this produces "real" and "imaginary" components. Ideally, the real components contain all the information about the absorption spectrum. However, in the actual NMR detection (where the hydrogen atoms do not resonate in phase with each other), phase correction must be performed to obtain their absorption spectrum as a real result of the Fourier transform. This phase correction is performed by the software which adjusts the real and imaginary components of the Fourier transform by different amounts for each sample.

The frequency spectrum is then subjected to further processing so that the chemical shift is calculated for the spectrum based on tetramethylsilane as a reference (this is routine in NMR). This forms the abscissa of the spectrum, while the intensity forms the ordinate. The chemical shift data provides information about the particular chemical environments in which the hydrogen nuclei are located in the sample, such that there is a relationship between a particular chemical shift and a particular portion of a molecule having hydrogen atoms. One such part is the glyceride backbone region of the triglyceride molecules.

An example of a frequency spectrum that can be generated for a sample using ground beef is in 3 shown. In this case, the triglyceride signature is the intensity of the frequency spectrum in the olefinic region 150 (5.0 to 5.6 ppm), the glyceride region 151 (4.0 to 4.5 ppm), and the bis-allylic region 152 (2 , 5 to 3.0 ppm). In addition, a strong resonance is seen between 1.0 and 2.5 ppm, which is due to the methylene groups (-CH ₂ -) in the fatty acid chain (except the bis-allylene). The reaction between 0.5 to 1.0 ppm is due to the terminal methyl (-CH ₃ ) groups of the fatty acid chain. The range of the spectrum between 0.5 and 2.5 ppm is typically not analyzed because the peaks in this region overlap significantly and make it difficult to obtain any useful information about the sample. However, if linolenic acid is present in sufficient quantities in the sample such that a characteristic signal is visible in the spectrum, that signal region (0.96 to 1.02 ppm) can be included in the triglyceride signature for purposes of analysis.

The resulting frequency spectrum for this sample is in step 106 concatenated into a series of a data matrix so that it can be mathematically analyzed. The glyceride region 151 is integrated so as to provide a measure of the number of glycerol-containing molecules present. Glycerol is present in each triglyceride molecule, and so the integral of the glycerol associated signal provides a measure of the number of triglyceride molecules present in the sample (assuming that other glycerol-containing molecules are not present in significant quantities). The spectrum is normalized by this integral to accommodate varying amounts of triglycerides among the samples. The glyceride region 151 can also be used as an internal reference to which the spectra are to be adjusted, if necessary.

In addition to normalizing the spectra is in step 107 the peak integration method of analysis is also applied to other display parts of the obtained spectra. In particular, the bis-allylic region 152 and the olefinic region 150 of the spectrum are integrated and adjusted according to glycerol normalization. While in this example the olefinic, bis-allylic and glyceride regions are integrated, in other cases the triglyceride signature relies on only two regions of the olefinic, glyceride, bis-allylic and linoleic regions, and only these two regions are integrated.

The values for these integrals are then used in pre-installed or user-determined equations to obtain characteristic data for the triglyceride signature. These equations may be based on prior knowledge of, for example, the number of protons contributing to each of the peaks in the spectrum, and can be used to estimate the percentage of monounsaturated fatty acids, polyunsaturated fatty acids and saturated fatty acids in the animal tissue. An exemplary consideration that should be made pertains to the signal generated by the hydrogen atom, that is, attached to the middle carbon in glycerol since this occurs just below the olefinic band 150. This should be considered and negated when calculating peak areas for the estimation of component concentrations. These characteristic data are then in the step 108 and stored in a database and related to the relevant sample for which the spectrum was obtained.

The steps 101 to 108 are repeated for a number of subsamples (such as five) from the original 50 g sample of tissue. This process is then repeated for 6 different batches of ground beef, each from a known and approved source. Overall, therefore, data from 30 samples are obtained. Samples are obtained from different batches to ensure that all natural variations between batch contents are taken into account. The average characteristic data obtained from each sample is calculated and stored in the database with the associated animal tissue marker "Minced Meat".

In step 109 A statistical analysis is performed on the data to mathematically analyze the variance in the data and to allow confidence limits associated with the characteristic data. The results of the statistical analysis are also stored in the database, for example, providing expected "average" values for the integrals, their internal ratios and associated 'tolerance levels' of variance.

Analyzing an unknown animal tissue

First example

4 FIG. 11 is a flowchart illustrating the essential process steps according to the first example of the invention. FIG. The first example provides a process for checking ground beef for the presence of lower quality spare meat.

The benchtop NMR spectrometer 2 gets started in step 200 calibrated according to the above method using a source of verified minced meat. A serving of finely minced, unknown animal tissue, advertised as ground beef, is obtained and crunched 201 to a sample according to step 101 of the calibration example. The sample is placed in the sample holder 5 of the spectrometer 2 brought and in step 202 according to the steps 103 to 106 of the calibration procedure to obtain a frequency spectrum for the sample that has been normalized based on the integral of the glyceride region 151. In step 203 For example, characteristic data for the sample is obtained using a method of peak integration. The olefinic and bis-allylic regions (150 and 151) of the frequency spectrum (between 5.0 to 5.6 ppm and 2.5 to 3.0 ppm, respectively) are integrated and these values are input to a set of predefined equations to obtain characteristic data (as before) for the animal tissue used in the sample. In step 204 compares the computer 10 this characteristic data with the average characteristic data for minced meat stored in a database during the calibration process to see if they are in the accepted tolerance ranges. If yes, then step 205a running, with the mark "ground beef" being output, which the user displays on a display of the computer 10 so that the identity can be determined. However, if the characteristic data for the sample is outside the allowable ranges, the user is warned that the sample does not meet the predetermined reaction and thus may contain substitute meat. This is step 205b , The sample will then likely require further testing, such as B. by DNA analysis.

In some cases, the percentage of monounsaturated, polyunsaturated and saturated fat content may also be displayed to the user when calculated with an acceptable level of accuracy.

Second example

In a second example of the invention, triglyceride signatures formed by several different animal tissues (in this case, beef, lamb, pig and goose) can be identified and distinguished. In this example, principal component analysis is used in place of the peak integration method used in the first example.

First, it is necessary to generate a suitable database of characteristic data for each type of animal tissue that the user wishes to identify. The system 1 is in the step 200 However, in this example, characteristic data are obtained for beef, lamb, pork and goose meat. 30 Samples are made using guaranteed, finely minced beef. Various "cuts" will be used to give representative average data for beef. The steps 101 to 106 are then executed to obtain a scaled / normalized frequency spectrum for each sample. These spectra are adjusted based on the glyceride region and stored as a marker for the animal tissue in a database. This process is repeated for the remaining tissues (using verified sources of lamb, pork and goose meat) to obtain spectra which are also stored in the database.

The characteristic data is then generated for each type of tissue (lamb, pork and goose) using the Principal Components (PC) analysis. This is a standard data processing technique and has been found to be particularly effective in distinguishing between different tissue types, especially between different species of animals. The dataset is decomposed into the main components that describe systematic sources of variability in the dataset in decreasing order of significance. The spectra from the different tissue types in the database are effectively compared with each other to determine the components that vary between each spectrum. This is used to classify the tissue types based on an appropriate subset of the major components, by constructing tolerance intervals defining each group in the PC space by a statistical method, for example Hotelling's T-squared distribution or the Mahalanobis distance.

An animal meat sample of unknown meat (of unknown identity) to be tested is then prepared into a sample and according to the steps 101 to 106 scanned to obtain a frequency spectrum with a triglyceride signature. The frequency spectrum (or part of it) is then analyzed as a single data set. Suitable major components as defined by the calibration process are then extracted from the spectrum as characteristic data for the animal tissue. These main components are compared to the predefined values in the database to see if they fall within one of the predefined tolerances. If so, the appropriate marker indicating an identity candidate for that data is output so that the user can read it on a display (step 205a ). If not, the user is warned on a display that the animal tissue is unknown and requires further testing, such as: By DNA analysis to confirm its identity (step 205b ). Alternatively, the characteristic data can be easily entered in a graphical representation, such. B. 5 in which case the user can make the decision as to the origin of the animal tissue.

An example of data that can be generated according to the second example is in FIG 5 shown. Two major components (PCA1 and PCA2) are shown, and data obtained from several samples prepared using beef, lamb, pork and goose meat are shown by the various items on the graph. The variance or 'tolerance level' expected for each of these tissues is shown by the annulus areas of the graph and is calculated by the calibration process using verified sources of animal tissue. The samples that generate points that occur in a particular annulus area (ie, tolerance) are identified as containing the tissue type or animal tissue candidate associated with that area. It is apparent from the graph that, for example, results obtained from the goose appear in a range different from that obtained from the lamb. 5 also shows that in some cases, such as As in pork and beef, an overlap between the tolerance regions can occur. If it is found that the characteristic data output from a sample is overlapping in one of these areas, the sample is marked as requiring the user further testing to confirm its identity.

Further, major components may be calibrated to determine contamination or adulteration values using a method known as major component regression. This is a multivariate regression technique suitable for applying spectral data sets and discussed, for example, in Jolliffe, I.T. 2002, Principal Component Analysis, Springer.

Third example

In a third example of the invention, the presence of a trace contaminant of pork in a product containing primarily beef is identified by analyzing partial regions of the triglyceride signature. Certain types of tissue produce characteristic peaks in the olefinic or bis-allylic region which can be used to identify this tissue. If this tissue is present in only a trace amount, these peaks can not significantly change the total integral for that region, but they may nevertheless be detectable in the overall spectrum by the presence of a narrow peak. A quantitative analysis of this peak can also be done using a computer 10 according to the following example.

First, a calibration process is performed. A database of averaged, normalized spectra is obtained using several guaranteed tissue types (including beef and pork) according to the second example. The olefinic and bis-allylic regions (150 and 152) of each spectrum are then processed by a computer 10 displayed on a graph and analyzed by a user. The presence of a characteristic peak (or peaks) in a spectrum considered to be sufficiently different from the equivalent region for other tissue types is manually identified by a user, and the limits of chemical shift for that peak are entered into a database, to define a partial region 600.

6 Figure 4 shows an example of an average spectrum obtained using verified beef compared to an average spectrum obtained using verified pork. In this example, a characteristic peak that occurs for pork and not for beef (or for the other types of tissue analyzed) is identified by the user at between 2.6 and 2.7 ppm, and a partial region 600 is defined. This peak occurs because of the relatively high presence of polyunsaturated fatty acids in pork.

Subregion 600 is integrated for each tissue type and the value is stored in the database as the reference value for that tissue type. A "tolerance" or variance for the integral of this subregion is also calculated. If the user is only interested in the possible presence of pork, then only a partial region 600 needs to be entered into the database. Alternatively, the process may be repeated as desired for equivalent subregions associated with other animal tissues. A calibration method using the principal component analysis is also performed according to the second example to obtain a database of characteristic data for each tissue type analyzed.

An unknown animal tissue to be analyzed is then prepared into a sample, and a normalized frequency spectrum is obtained as before. The triglyceride signature for the sample is then analyzed using principal component analysis and the characteristic data is identified as falling within the beef tolerance level. In addition, the specific subregions of interest (as defined during the calibration process) are now also analyzed to identify the possible presence of trace impurities from other tissue types in the sample. The integral for the subregion 600 of the unknown meat is through the computer 10 compared with the values stored during the pork and beef calibration process. In this case, although the unknown tissue has previously been identified as beef, the integral of subregion 600 is above what is typically expected for beef and closer to what is expected for pork. Predefined algorithms then calculate the approximate relative weighting of beef and pork and through the computer 10 displayed. In addition, a graphical representation of subregion 600 is also provided by the computer 10 for verified pork and beef samples and for the unknown tissues visually displayed to the user. This allows the user to manually check to see if there is consistency with the analysis. Thus, an animal tissue candidate is displayed by showing the approximate weights of the animal tissues used in the sample, and it is shown that the raw data supports this analysis.

Claims (21)

A method of identifying animal tissue using an NMR spectrometer with a permanent magnet having a magnetic field strength on a sample holder of the spectrometer between 1.0 and 2.5 Tesla, the method comprising: (i) contacting the animal tissue with a solvent to dissolve triglyceride molecules from the animal tissue in the solvent to obtain a sample; (ii) placing the sample in the sample holder of the NMR spectrometer; (iii) applying a radio frequency (RF) pulse sequence to the sample and detecting a free induction decay (FID) signal using the NMR spectrometer; (iv) converting the detected FID signal into the frequency domain to obtain a frequency spectrum with a triglyceride signature, and (v) analyzing the triglyceride signature and outputting characteristic data for the animal tissue, the characteristic data indicating the identity of the animal tissue. The method of claim 1, wherein the analysis of step (v) comprises integrating an olefinic region of the triglyceride signature. The method of claim 2, wherein the chemical shift of the olefinic region is between 5.0 and 5.6 ppm. The method of any one of claims 1 to 3, wherein the analysis of step (v) comprises integrating a bis-allylic region of the triglyceride signature. The method of claim 4, wherein the chemical shift of the bis-allylic region is between 2.5 and 3.0 ppm. The method of claim 1, wherein the analysis of step (v) comprises performing a method of principal component analysis. Method according to one of the preceding claims, wherein after step (iii) the following steps are carried out: Comparing a set of peaks in signal intensity detected at the beginning of the FID signal over time with a set of peaks detected at the end of the FID signal in the time period, and Calculate a signal-to-noise ratio from this comparison. The method of claim 7, wherein step (iii) is repeated according to the calculated signal-to-noise ratio to achieve a target signal-to-noise ratio for the frequency spectrum. The method of any one of the preceding claims, wherein the output characteristic data indicates an animal tissue candidate for the animal tissue used to obtain the sample. The method of any one of the preceding claims, wherein step (iv) further comprises performing a phase correction on the frequency spectrum. A process according to any one of the preceding claims, wherein the solvent is deuterated chloroform. Method according to one of the preceding claims, wherein the frequency of the RF pulse is between 40 and 110 MHz. The method of any one of the preceding claims, wherein step (iv) further comprises integrating a glyceride region of the frequency spectrum. The method of claim 13, wherein the glyceride region has a chemical shift between 4.0 and 4.5 ppm. The method of claim 13 or 14, wherein step (iv) further comprises normalizing the spectrum based on the integral of the glyceride region. Method according to one of the preceding claims, wherein step (iv) is carried out using a fast Fourier transform technique. A method according to any one of the preceding claims, wherein the animal tissue is first chopped or finely chopped before step (i). The method of any one of the preceding claims, wherein steps (i) to (iv) are initially performed using known animal tissue so as to obtain target triglyceride signature data for use in the analysis of step (v). The method of any preceding claim, further comprising storing the results of the analysis of step (v) in a database. The method of claim 19, wherein the database is used to perform a method of principal component analysis. The method of any one of the preceding claims, wherein the triglyceride signature comprises at least two of an olefinic, bis-allylic and glyceride region of the frequency spectrum.

Images