Estimation of the fat content of minced meat using a portable minced meat microwave fat meter
M. Kent, A. Lees and A. Roger
A portable device for the estimation of the fat content of fish has been previously described and its use for the measurement of meat products is now reported. Unlike fish flesh, where the fat accumulates at the expense of water and protein, in minced meat products it can be shown that fat accumulation is independent, i.e. additive. When other components are added the relationship between the meter response and the fat content does not hold. The meter may then be used only for estimation of water content.
Keywords: minced meat; fat content; water content; hand-held instrument; microwaves
An instrument has been described which, by measuring the water content of fish flesh, can provide an estimate of the fat content (Kent 1990).
It has been shown that , for all the fish species studied, the fat accumulates at the expense of water and protein. Some of the water loss occurs because of the loss of protein and these changes are a consequence of the breeding and feeding cycles of the fish (Iles and Wood, 1965). It is possible that this instrument could be used similarly to estimate the fat content of other foodstuffs and of particular interest, minced meat products.
It is important at present to consider only food materials of reasonable homogeneity or with small spatial variations in comparison with the sensor and the wavelength of measurement. The present instrument operates at a frequency of 2GHz for which the wavelength in air is 0.15m but of the order of 0.02m in moist materials. Ohlsson et al. (1974) showed that in meat homogenates the real component of the complex dielectric permittivity has a fairly linear dependence on water content. It is to this property that the sensor responds (Kent, 1973).
Figure 1. Triangular diagram for ternary mixture of fat, water and solids. The point P represents a starting composition of 15% solid, 10% fat and 75% water Locus DPB represents addition or subtraction of fat FPA similar variation In solid content and EPC similar variation in water content. For any of these conditions the ratio of the other two components remains the same.
For ternary mixtures, such as fat, water and non-fat solids (including protei, ash, etc.) the composition can be plotted on a triangular diagram (Figure 1). With the independent removal or accumulation of fat, starting from an initial composition at P (75% water, 10% fat, 15% non-fat solid), the composition follows a line such as DPB, passing through a corner of the triangle and preserving a constant water to solids ratio. Substraction or addition water would, however, follow a locus such as EPC and substraction or addition of solid, FPA. Natural accumulation such as occurs in fish, follows a quite different locus. Although the results published earlier (Kent, 1990) appeared to show no significant difference from the constant water to solids situation, this was not in agreement with Iles and Wood (1965) who had demonstrated that fat was accumulated at the expense of water and protein and not simply by addition. This discrepancy can be attributed to the weighting of the results by a number of scattered data obtained early in the experiment while the analytical procedures were being developed. If these data are excluded then agreement with Iles and Wood is achieved. This locus is shown in Figure 3. The nature of the product and its effect on the instrument is now described.
Materials and Methods
Beef mince is sold under at least three descriptions, (a) beef mince (high), (b) steak mince (medium to low fat) and (c) lean mince (low fat content). A further grade of ‘extra lean’ is sometimes available but will be considered only as part of category (c). These various grades of mince (here referred to simply as ‘beef mince’), were bought as 53 samples of 250g each from a variety of butchers’ shops and supermarkets. In addition, 26 types of sausages made from a variety of meats were bought, as were some samples of lamb, pork and turkey mince. Skinless sausages were not examined.
Water contents were determined by oven drying at a temperature of about 105(C for 24 hours and measuring the weight loss. Measurements were made in triplicate on 20g aliquots. Fat contents were determined, also in triplicate, using a Foss-Let apparatus. Nitrogen was determined according to the method of Kjeldahl. Ash was determined by heating dried samples at 550(C in a furnace and weighing the residue.
Prior to use, the samples were kept in a laboratory refrigerator at 5(2(C. The temperature of the minces was measured at the time of measurement and was found to be 11.5(2.2(C. This range of temperature does not significantly affect the meter reading. Over the range 0-20(C, the reading for mackerel, for example, changes by 2% fat content.
The meter readings were obtained by hand-compressing the minces and pressing the sensing head firmly into the material. Further maceration of the minces then followed using a Magimix blender, type 2000. The meter measurements were then repeated. All meter measurements were performed in quadruplicate and the means calculated. r the sausages, the sensor was pressed firmly onto the surface and the measurements taken with the skin on. The skin was then removed and the measurements repeated on the sausage meat itself. The average reading for 8-10 sausages in each batch was used in the results.
Figure 2. Water content versus fat content for beef mince.
Results and Discussions
The samples obtained provided a wide range contents. The range of fat and water contents for the beef mince and the close relationship between them can be seen in Figure 2. The correlation coefficient between these components was 0.986 and the slope of the regression line was as expected for the situation involving addition of fat, i.e. independent accumulation.
Table 1 summarizes the parameters involved, where the gradient for a theoretical line starting with the same water content at the condition of zero fat is also shown for comparison. There was no statistically significant difference. The experimental regression line is plotted in Figure 3 on a triangular diagram where this behaviour was clearly seen as independent accumulation. The dashed line (A) represents the experimental results where the solid line (B) is the theoretical line as in Figure 1 for independent fat accumulation. The broken line (C) is that found for herring flesh after the data of Iles and Wood (1965).
The ratio of water to nitrogen content was independent of variation in water content, having a value of 21.08±0.88. This invariance also indicates the addition only of fat these minces. No significant correlations were found between meter reading and ash content or nitrogen content, so only the fat and water contents need to be considered.
Figure 3. Triangular diagram for the experimental data Line A and the solid points represent the variation observed in beef mince composition over the range of fat contents measured. Line D and the crosses are the results for sausages. There is a much greater scatter of points in this case due to addition of other solids and possibly water. The deviation of the regression line from that for purely additive fat (line B) is significant There is no significant deviation in the case of the mince. Line C represents the data for herring flesh showing its significant deviation from the situation represented by line B.
Figure 4. Fat content as a percentage of total weight of all the minces studied versus meter readings in decibels. Linear regression shown is for beef mince only, circle, Beef; triangle, lamb; square, pork; filled circle, turkey.
Meat Fat Meter
Figure 5. Water content as a percentage of total weight of all the minces studied versus meter readings in decibels Linear regression shown is for beef mince only circle, Beef; triangle, lamb; square, pork; filled circle, turkey .
When either the fat or water content is plotted against the meter reading in dB (Figures 4 and 5) apparently linear dependencies were seen.
Table 2 summarizes the regression parameters for these calibration lines. Note that the minces from other meats lie on the same line but these were not included in the regression analysis.
With fish, the partial replacement of protein and water by fat results in a non-linear response of the instrument over this range of fat content and it was found necessary to use a logarithmic calibration for the best possible fit of the data,
i.e. Log 10 (fat) = 1 +bx (dB)
The results for mince can be treated in the same way but the fit between the log of the fat content and the meter response in dB is not significantly different from the linear fit (Equation 2 in Table 2, Figure 6). Nor is there improvement when the log of the water content is considered (Equation 4 in Table 2). The only advantage is that this the existing form of calibration used in the instrument. As a further test of the calibrations, the fat and water contents predicted by Equations 1 and 2 in Table 2 for an additional set of 14 samples were compared with conventional analyses on the same samples.
Meat Fat Meter
Figure 6. The data as in Figure 4 but with a logarithmic fit to beef mince results only- circle, Beef; triangle, lamb; square, pork; filled circle, turkey.
Regression equations for these data are shown in Table 3. For exact equivalence, the intercept should be zero and the gradient 1.0. From these data sets it is clear that the logarithmic calibration (Equation 2) performs better with no significant differences from the ideal relationship. The linear calibration (Equation 1) on the other hand gives very significant deviations from ideality.
In general; the fat contents of the sausages were higher than those of minces. Despite the addition of binding agents and flavouring materials, there was still some correlation between fat and water contents, although the correlation coefficient was reduced (Table 1). There was also little correlation between solids content and nitrogen as a result of the addition of non-nitrogenous materials to the mixture. The water to nitrogen ratio was also extremely variable for the same reason. The regression line for fat versus water is plotted in Figure 3. It should be noted that it departs considerably from the simple fat accumulation line due to the presence of these other added components.
The statistical data are shown in Table 2. There was no significant difference between pork or beef sausages and so the results have combined. The degree of correlation of the meter reading between either fat or water content is not significantly different (Equations 5 and 7 in Table 2). Although the removal of the sausage skins and direct measurement of the contents increased the correlation coefficients, these increases also were not statistically significant (Equations 8 and 10 in Table 2). However, the improvement in correlation between fat with skin-on (equation 5 in Table 2) and water for meat only (Equation in Table 2) is significant at the 10% level.
As with the minces, logarithmic calibrations were attempted for fat and water contents (Equations 9 and 11 in Table 2). Whilst not having the same degree of correlation as for the linear case, nevertheless, as stated, it is a useful for of calibration as the instrument is commercially manufactured with such logarithmic calibration incorporated. Correlations for the log of water content were not better than for the log of fat content against meter reading.
An independent set of sausage samples were obtained and, as with the minces, the performance of the meter using Equations 8-10 (Table 2) was examined. The regression equations for fat and water predicted from the meter readings versus those found by the conventional methods used for the calibration of the meter are also shown in Table 3. It can be seen that the meter performed far better in measuring the water content because of the lack of correlation between fat and water.
It would appear from these results that the instrument designed and constructed for the measurement of fish fat content can be readily adapted for use with some meat products such as minces. It may also be used with sausage meat, although not with the same accuracy because of additional components in the product.
Its use in this case would be better for estimation of water content. Although not offering the same accuracy as the more usual laboratory-based methods, it has the advantage of being portable instantaneous and non-destructive. It would also be suitable for on-line measurement and control as originally envisaged by Steele and Kent (1978).
Iles, T.D. and Wood, R.J. (1965). The fat/water relationship in North Sea Herring (Clupea Harengus) and its possible significance. J. Mar. Biol. Assoc. UK 45, 353-366
Kent, M. (1973) The use of strip-line configuration in microwave moisture measurements, II. J. Microwave Power 8, 189-194
Kent, M. (1990) Hand-held instrument for fat/water determination in whole fish. Food control 1, 47-53
Ohlsson, T., Enriques, M. and Bengtsson, N. (1974) Dielectric properties of model meat emulsions at 900 and 2800 MHz in relation to their composition. J. Food Sci. 39, 1153-1156
Steele, D.J. and Kent, M. (1978) Microwave stripline techniques applied to moisture in food material. 13th Microwave Power Symposium, Ottowa, International Microwave Power Institute, Clifton, Virginia, USA, pp 31-36
Received 27 November 1992
Revised 23 February 1993
Accepted 23 February 1993
Table 1: Regression Parameters for fat v. water
|Sample||Fat (%)= a + b x water(%)||No of samples||Correlation coefficient|
|Beef mince||F=96.72 ± 2.23 – (1.25 ± 0.03)xW||42||0.986|
|Independent fat accumulation||F= 100-1.29xW|
|Sausages (all meats)||F=72.51 ± 6.23 – (0.930 ± 0.111)xW||26||0.862|
|Independant fat accumulation||F= 100-1.28xW|
Table 2: Regression parameters for fat and water v meter readings
|Sample||Equation no.||Parameters||No of
|Beef mince||1||Fat (%) = 50.62 ± 2.16 – (5.10 ± 0.29) x dB||42||0.942|
|2||log 10 (Fat %) = 2.47 ± 0.09 – (0.188 ± 0.012) x dB||42||0.929|
|3||Water (%) = 36.90 ± 1.51 + (4.08 ± 0.20) x dB||42||0.955|
|4||log 10 (Water %) = 1.63 ± 0.01 – (0.0266 ± 0.0014) x dB||42||0.952|
|Sausages (all meats)||5||Fat (%) = 66.70 ± 8.07 – (6.29 ± 1.10) x dB||26||0.759|
|6||log 10 (Fat %) = 2.30 ± 0.18 – (0.136 ± 0.025) x dB||26||0.747|
|7||Water (%) = 8.08 ± 6.09 + (6.52 ± 0.83) x dB||26||0.848|
|Sausage meat only||8||Fat (%) = 55.38 ± 4.55 – (4.52 ± 0.59) x dB||26||0.842|
|9||log 10 (Fat %) = 2.08 ± 0.096 – (0.101 ± 0.0125) x dB||26||0.856|
|10||Water (%) = 21.87 ± 3.62 + (4.41 ± 0.47) x dB||26||0.887|
|11||log 10 (Water %) = 1.48 ± 0.03 – (0.349 ± 0.0038) x dB||26||0.880|
Table 3: Regression of component concentration predicted by calibration equation v. concentration measeured by conventional methods
|Sample||Equation No ( from table2)||Parameters|
|Beef mince (n = 14)||1||Fat (pred)=4.17 ± 1.26 + (0.711 ± 0.076) x fat (meas)||0.937|
|2||Fat (pred)=1.14 ± 0.92 + (0.977 ± 0.055) x fat (meas)||0.981|
|3||Water (pred)=17.48 ± 5.67 + (0.726 ± 0.085) x water (meas)||0.926|
|8||Fat (pred)=7.80 ± 3.19 + (0.558 ± 0.136) x fat (meas)||0.808|
|Sausages (n=11)||9||Fat (pred)=8.55 ± 3.31 + (0.522 ± 0.141) x fat (meas)||0.777|
|10||Water (pred)=1.37 ± 4.88 + (1.00 ± 0.089) x water (meas)|