A method of determining the content of a component in a fluid sample and an apparatus therefore
TECHNICAL FIELD
The present invention relates to a method an apparatus of determining the content of a component in a fluid sample, e g determination of fat in a milk sample, by IR analysis, as stated in the introductory part of claim 1
BACKGROUND ART
The presently preferred apparatus measures either a full IR spectrum or applies several filters providing information about the absorption at a plurality of wavenumbers.
The primary purpose of the present invention is to provide a simple IR apparatus that is able to determine the fat content in a milk sample A more general purpose is to determine any specific component in a fluid sample.
The problem Determination of the content of a specific component in a fluid, such as fat in a milk sample, through IR analysis is impeded by the fact that the attenuation through the material (usually a fluid material or liquid ) in many cases is very high. Accordingly any analysis must take place by use of a very thin cuvette. The pathlength of the light passing through a typical IR cuvette for analysing a fluid, e.g. milk, is about 10 to 200 μm, and the IR light beam is e.g. about 1 to 5 mm in diameter, i.e. from about 0.001 mm3 to about 5 mm3 of the fluid is analysed. Consequently the actually analysed sample is very little This is no problem when analysing homogeneous or properly homogenised fluids. However, if the liquid is an mhomogeneous mixture of a plurality of substances, which is the case when measuring raw milk, the actually measured sample may not be truly representative for the whole sample
SUMMARY OF THE INVENTION
In the broadest aspect it is suggested, according to the invention, to apply a method of quantitative determining the content or concentration of a component in a fluid sample by IR analysis, the method comprising the following steps a) directing one or more IR light beams through each of a number, n, of different parts of a fluid sample, the IR light including at least one wavelength within a waveband which the component absorbs,
b) for each of the n parts of the sample, detecting the IR light having passed through the respective part of the sample, c) obtaining, for each of the parts, at least one value on the basis of the detected IR light, the value representing information from which the content or concentration may be estimated, d) calculating at least two statistical parameters, such as statistical moments, eg. a mean value and a standard deviation, characteristic of a statistical distribution of the obtained values, and e) estimating or determining the content or concentration of the component on the basis of the at least two statistical parameters.
Also according to the invention it is suggested to apply a method of quantitative determination of a component in a fluid sample by IR analysis, comprising the following steps, introducing a fraction of the sample into an IR cuvette, directing an IR light beam through the cuvette,said IR light including at least one waveband in which the specified component in the fluid sample absorbs light, measuring the intensity of the IR light having passed through the fraction of the sample located in the IR cuvette, characterised by the further steps, storing at least one value, such as a measurement value, an intermediate calculation result or final result, derived from the measured intensity, and representing the content of the component, introducing a new fraction of the sample into an IR cuvette, repeating step b,c,d), repeating step e, f) a plurality of times, calculating an average value of the stored values, representing the content of the component, calculating the standard deviation, calculating the estimated content of the component as a function of the average value and the standard deviation.
Thereby the fat content may be determined fairly accurate even though the instrument is simple and the measured sample is not homogenised. The obtained accuracy depends on the number of repetitions/repetitive measurements on each sample. A better accuracy may be obtained by increasing the number of measurements.
A preferred embodiment of the new apparatus for a method according to the invention applies only a single filter. In an apparatus for determining the content of fat in milk such filter is preferably a 1.73 cm"1 filter. A further embodiment includes a small number of filters, such as four filters. Preferably the apparatus has no homogeniser. The apparatus is arranged to analyse raw milk without any homogenisation.
In one embodiment the fat content is determined for each measurement. An average fat content as well as the standard deviation is calculated from the plurality of measurements. In an alternative embodiment each measurement represents a value such as the measured intensity or an intermediate calculation result. A plurality of measurements represents a plurality of values from which an average value is calculated as well as a standard deviation.
Such average of an intermediate calculation may be applied for the further calculation of the specific content.
BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 shows as an example a schematic diagram of a first embodiment of an apparatus according to the invention.
Figure 2 shows as an example a schematic diagram of a second embodiment of an apparatus according to the invention Figure 3 shows measurements of a test set comprising natural samples in order to illustrate the principles of the present invention
Figure 4 shows the actual fat content of 37 samples versus the predicted content using the traditional method for homogeneous samples
Figure 5 shows - the actual fat content of 37 samples versus the predicted content using the new method according to the invention.
DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS
Figure 1 shows an exemplary set up/arrangement for carrying out a method according to the invention. A light source 12 emits a light beam [(such as MID-IR light or NIR light, and in the case of milk preferably comprising a wavelength corresponding to 1 73 cm 1, (also called "fat A"). The light beam passes through a sample cuvette 14 and reach a number of filters and detectors 16, preferably including means for recording, displaying and processing data representing the detected light signals. The filters may be arranged within holding means for the detectors 16 Such means may be a housing of conventional type (not shown) enclosing the light source 12, cuvette 14 and detectors as well as mounting or holding means therefore.
Figure 2 shows a similar arrangement. Further figure 2 indicates that the sample cuvette 14 preferably receives the samples through a flow system 22 for extraction or aspiration of samples Such flow system may comprise thermostatical control means. The sample is preferably ejected through a sample ejection system 24, such as a pump removing the sample to an outlet/drain
According to the invention the content of a fluid composition such as raw milk comprising several components such as fat globules, proteins, lactose and water can be determined by IR analysis without carrying out a homogenisation of the fluid.
A small fraction of a milk sample is aspirated into a sample cuvette through a flow system 22. A light source 12 directs an IR light beam towards a cuvette 14. The light beam is attenuated by the milk sample fraction. The attenuated and filtered light is detected by a number of detectors 16.
According to the invention a second small fraction of the same milk sample is aspirated into the same sample cuvette through the same flow system 22. The same light source 10 directs an IR light beam through the same filters 12 towards the same cuvette 14. The light beam is attenuated by the second milk sample fraction. The attenuated signal is detected by the same detector 16.
In the same manner as mentioned above a plurality of fractions of the same milk sample is measured in the IR-analysis apparatus.
The detected signals are converted into digital signals, stored and processed in means 30 arranged therefore, e.g. a computer. According to the invention a good estimate of the content of e.g. fat in raw milk can be calculated as follows when using a single filter and a single detector.
The fat content: Fat = b0 + b, * Fa + b2 * std(Fa) ,
wherein b0 = a first constant b, = a second constant
Fa = Intensities measured with a filter, e.g. 1.73 cm 1 F. is the mean value b2 = a third constant Std(Fa) = standard deviation of Fa and wherein each constant is a predetermined number, being determined from a number of calibration measurements on known samples.
in other words according to the invention a surprisingly good estimate of the true fat content of a raw milk sample may be determined by using an average value of a plurality values of first estimated fat contents based on a plurality of measurements, the plurality of values being calculated as if the milk sample were a homogenised sample, - and by applying a correction which is proportional to the standard deviation of the plurality of values.
The plurality, necessary to obtain a specified accuracy may be estimated from the standard deviation. If the standard deviation is high, then the content of fat is high, and correction needed to estimate the true content is high as well.
The above example relates to a milk sample measured by use of a single filter.
However, the same method may be applied to other mhomogeneous fluids using one or more filters. In general the unknown content of a specific component may be calculated as follows:
Content of component: n
Cc = b0 + ∑ b 1 ( * F , + ∑ b2 ] * std ( F,) , i=1 j=1 wherein n is the number of filters applied; and i and j relate to specific filters, and b0, b„-b1n, b2,-b2l are predetermined constants.
In an even more generalized version this may be expressed: n n
Cc= b0 + ∑ b , , * F, + ∑ b2| * F ( m 1 , m2 , m k) i=1 j=1
where mk is the k,h central moment, such as variance, kurtosis or skewness, of the obtained values derived from measurements on raw samples not being subjected to homogenisation.
When measuring raw milk - and when measuring the fat content in raw milk - the prior art recommends to homogenise the milk before introducing it into the IR cuvette. By homogenisation the fat globules are spread into much smaller items or particles. Further it is normal practice to heat up the samples to about 40 C° in order to melt the fat. The IR absorption is highly temperature-dependent, and the IR cuvette must be thermostabilised and kept at a constant and known temperature.
According to the present invention the measurement can be carried out without homogenisation of the samples.
When analysing the inhomogeneous raw milk, the fat globules will spread some of the IR light. Consequently an analysis result tend to indicate higher absorption and accordingly more fat than is actually the case.
The advantage of this method is that the new IR analysis apparatus does not need a homogeniser. A good accuracy can be obtained by repeating the measurements, in the way described in claim 1 , on a (large) plurality of fractions of the sample, i.e the fluid in a sample cup provided from the object to be tested, e.g. raw milk from a cow.
EXAMPLE 1
The inventive method described above is illustrated by Figure 3 showing measurements on a test set comprising natural milk samples. The measurements were carried out twice (first on the
natural raw milk samples and later on the same samples after they had passed through a homogenisor) on an apparatus according to the invention, calibrated to measure the fat content in homogenised milk correctly.
The fat content in milk can be determined in a number of ways. A reliable way is to determine the IR absorption at the wavelength 5,7 μm corresponding to a 1.73 cm 1 (a so-called fat A filter). Fat absorbs clearly at this wavelength. Hardly any other components in the milk contributes to the attenuation. Accordingly the fat content may be calculated from the measured attenuation of the IR beam passing through the cuvette.
At first the samples were measured as raw milk without homogenisation. Then they were homogenised and measured once more, and finally the fat content was determined by a reference method, such as by use of a conventional instrument for determination of the fat content in milk, e.g. a MilkoScan 120 FT.
The measurements on the right line marked by reference number 101 refer to the raw milk samples measured before homogenisation. The measurements on the left line marked by reference number 102 refer to the homogenised milk samples.
It appears that the repeatability of the measurements 102 on the homogenised milk samples are very good. All measurements are located approximately in the same small area, i.e. the standard deviation is small.
On the contrary the measurements 101 on the raw milk samples, having high fat contents, are spread on a length of line. 103. They indicate a considerable standard deviation. Also it is clear that the standard deviation increases with the fat content.
It appears from Figure 3 that there is a linear relationship between the content of fat and the attenuation measured on homogenised raw milk using a 1.73 cm filter. The method applied for the determination of line 102 is well known. The results can be considered as fairly true values
It also appears from Figure 3 that there is still a linear relationship when measuring raw milk not being homogenised in respect to an average value.
The slope of line 101 differs from the slope of line 102. This can be explained as follows: the measured attenuation is a result of a first contribution due to the pure IR absorption in the fat molecules and a second contribution due to the scattering of the fat globules.
A first important observation is that both contributions tend to increase with the fat content A second important observation is that by repeating the measurements a plurality of times, an average value can be determined. All average values seem to be located on a straight line The plurality, i.e. the number of repetitions must be selected to ensure that the repeatability of said average value is satisfactory, corresponding to a desired accuracy of the measurement. An acceptable repeatability is here to be understood as a repeatability which is in accordance with the desired accuracy of the apparatus.
A third observation appearing from Figure 3 is that the standard deviation tends to increase with the content of fat.
From the above-mentioned observations the inventor has concluded that the fat content in the milk sample may be calculated within appropriate accuracy from the below formula
The fat content: Fat = b0 + b, * Fa + b2 * std(Fa) , wherein b0 = a first predetermined constant b, = a second predetermined constant
F = Intensities measured with a filter, e.g. 1.73 cm 1 F is the mean value b2 = a third predetermined constant Std(Fa) = standard deviation of Fa
It appears from the above that an adjustment of the slope of the line 102 as well as the calculated standard deviation may be used to obtain a fairly true value representing the fat content
Example 2
The below examples shall illustrate the advantage by using the inventive method and apparatus Figures 4 and 5 show two series of measurements of fat in 37 samples using an instrument having a single 1.73 cm 1 filter. In Figure 4 the measurements are carried out according to the traditional method without use of the standard deviation In Figure 5 the measurements are carried out according to the inventive method with use of the standard deviation The accuracy was determined and is shown in the below tabel:
Description Accuracy, RMSEP
Without use of standard deviation 0 140 with use of standard deviation 0073
It clearly appears that the use of the standard deviation for correction of the measured values improves the accuracy substantially.
While a few particular embodiments of the invention have been shown, it will be understood, of course, that the invention is not limited thereto since many modifications may be made, such as determining other components than fat, and using other statistical parameters or moments of higher order, and it is, therefore, contemplated that the appended claims shall cover any such modification as fall within the true spirit and scope of the invention.