15 Applications: Food Science
349
Distinguishing between wheat varieties based on their breadmaking quality was
the first qualitative analysis performed in the 1980s [18]. Downey et al. [19] correlated
wheat hardness with breadmaking quality, using a wheat hardness index, to differentiate between the wheat samples. Differences in particle sizes and the presence
of inorganic additives enabled 97% correct classification of a range of commercial
white flours, i.e. biscuit, self-raising, household, cake, bakers’ and soda bread mix
[20].
Wheat used for food applications comprised bread wheat (Triticum aestivum)
and durum wheat (Triticum durum). The latter is used for pasta production and
has different chemical and physical properties compared to bread wheat. In some
European countries such as Italy, pasta is required to be produced using only durum
wheat semolina and water. The addition of bread wheat results in a lower-quality
product which would have inadequate resistance to cooking. Potential adulteration
of durum wheat with bread wheat is thus of great concern. The potential to detect the
addition of bread wheat flour to durum wheat flour was illustrated with uncertainties
associated with the models to be about half of that of the official Italian method [4].
Although NIR spectroscopy is extensively used to quantify chemical composition in cereals (e.g. protein, moisture, oil), limited studies are available on cultivar
discrimination and traceability of cereals [21].
15.3 Meat and Meat Products
NIR spectroscopy is extensively used to determine the content of meat components.
The first NIR spectroscopy models developed included those which could predict
intramuscular fat and moisture content. These could be predicted at excellent accuracies with low SEP results (0.18% for intramuscular fat; 0.37% for moisture) and
high RPD values (9.17 for intramuscular fat; 7.21 for moisture) demonstrated [22].
Quantification of protein also with excellent prediction accuracies (SEP = 0.35%;
RPD = 5.13) followed soon. NIR prediction of technological properties is more challenging as can be seen in SEP results and RPD values obtained for pH (0.05; 1.28),
colour (0.42; 2.16) and water-holding capacity (WHC; 2.355; 1.27). Better results
were obtained for pH when spectra were collected from intact meat samples. When
minced meat was used, chemical composition predictions were more successful
than for intact meat. Some success was achieved with more complex predictions
such as ash content (SEP = 0.15%; RPD = 4.53). Adding the visible range enabled
improved NIR spectroscopy predictions for colour. Water-holding capacity and drip
loss measurements could, however, only be done with limited success thus far.
NIR spectroscopy measurement of sensory properties has not been successful
due to intact meat samples not being homogenous, which is the main reason for
poor predictions to date. The subjectivity of taste panels also contributes, in addition
to inconsistent sample preparation. Consistent presentation of the sample to the
instrument is also important and should receive the required attention when acquiring
NIR spectra. A reasonable accuracy was obtained when beef tenderness was predicted
349
Distinguishing between wheat varieties based on their breadmaking quality was
the first qualitative analysis performed in the 1980s [18]. Downey et al. [19] correlated
wheat hardness with breadmaking quality, using a wheat hardness index, to differentiate between the wheat samples. Differences in particle sizes and the presence
of inorganic additives enabled 97% correct classification of a range of commercial
white flours, i.e. biscuit, self-raising, household, cake, bakers’ and soda bread mix
[20].
Wheat used for food applications comprised bread wheat (Triticum aestivum)
and durum wheat (Triticum durum). The latter is used for pasta production and
has different chemical and physical properties compared to bread wheat. In some
European countries such as Italy, pasta is required to be produced using only durum
wheat semolina and water. The addition of bread wheat results in a lower-quality
product which would have inadequate resistance to cooking. Potential adulteration
of durum wheat with bread wheat is thus of great concern. The potential to detect the
addition of bread wheat flour to durum wheat flour was illustrated with uncertainties
associated with the models to be about half of that of the official Italian method [4].
Although NIR spectroscopy is extensively used to quantify chemical composition in cereals (e.g. protein, moisture, oil), limited studies are available on cultivar
discrimination and traceability of cereals [21].
15.3 Meat and Meat Products
NIR spectroscopy is extensively used to determine the content of meat components.
The first NIR spectroscopy models developed included those which could predict
intramuscular fat and moisture content. These could be predicted at excellent accuracies with low SEP results (0.18% for intramuscular fat; 0.37% for moisture) and
high RPD values (9.17 for intramuscular fat; 7.21 for moisture) demonstrated [22].
Quantification of protein also with excellent prediction accuracies (SEP = 0.35%;
RPD = 5.13) followed soon. NIR prediction of technological properties is more challenging as can be seen in SEP results and RPD values obtained for pH (0.05; 1.28),
colour (0.42; 2.16) and water-holding capacity (WHC; 2.355; 1.27). Better results
were obtained for pH when spectra were collected from intact meat samples. When
minced meat was used, chemical composition predictions were more successful
than for intact meat. Some success was achieved with more complex predictions
such as ash content (SEP = 0.15%; RPD = 4.53). Adding the visible range enabled
improved NIR spectroscopy predictions for colour. Water-holding capacity and drip
loss measurements could, however, only be done with limited success thus far.
NIR spectroscopy measurement of sensory properties has not been successful
due to intact meat samples not being homogenous, which is the main reason for
poor predictions to date. The subjectivity of taste panels also contributes, in addition
to inconsistent sample preparation. Consistent presentation of the sample to the
instrument is also important and should receive the required attention when acquiring
NIR spectra. A reasonable accuracy was obtained when beef tenderness was predicted