MEAT
QUALITY SENSORS
Must be
-
very fast (maximum several seconds)
-
relatively non-destructive
-
suitable for hanging carcass
-
relatively inexpensive
-
strong, waterproof & idiot proof
-
absolutely no contamination
Most feasible
-
electrical impedance
-
electromechanical
-
fibre-optic
To assess
-
APPEARANCE (colour vs scattering)
-
FLUID EXUDATE
-
SOFTNESS
-
FAT DISTRIBUTION
To predict
-
TOUGHNESS
-
COOKING LOSSES
-
PROCESSING YIELDS
-
FINAL PRODUCT QUALITY
STRATEGY
[1] Innovative research on biophysical basis of meat quality
[2] Make apparatus portable or remote for abattoir use
[3] Correlate data to expert opinion or sensory panel responses
SUCCESS IS
-
advance the science of meat quality
-
make predictions with commercial value
-
help improve average meat quality
FAILURE IS
-
exploiting blind correlations
-
publishing boring, repetitive research
SLIDE 1. Fat is very important in the
meat industry because it determines yield. It is difficult to know
how much muscle there is in a carcass, so we assume bone is constant.
Then we estimate fat from subcutaneous fat depth. Then we predict
how much of the carcass weight is muscle. Hence, our technology for
fat depth detection is well developed. But, for quality, there
is a lot more to measure than just fat depth.
SLIDE 2. This shows a brief history of how technology for fat depth
measurement evolved.
-
The depth of fat used to be measured by cutting through the fat with a
scalpel.
-
Then a light pipe was used so that the operator could see the muscle:fat
boundary without slashing through the carcass.
-
The light pipe and human eye were replaced by a diode detector for the
muscle :fat boundary. An important feature was that the depth of
the boundary was found from a depth detector relative to a plate that remained
on the meat surface.
-
Optical fibers were then introduced which enabled us to see inside the
carcass.
-
Broad-band monochromatic measurements were replaced by a diode-array spectrograph
which allowed spectrophotometry and multiple regression.
-
Perhaps the future will allow us to measure both wavelength and angle of
measurement through the tissue on one array?
SLIDE 3. Remember the importance of measuring depth in the meat
in slide 2? Well, a classic fat-depth probe, such as those developed in
Denmark, provides a superior launch-platform for any type of measurement
to be made. On the x-axis is depth in the meat. On the y-axis is whatever
is being measured at the tip of the probe. THIS ADDS A STEREOLOGICAL
DIMENSION TO THE MEASUREMENT.
SLIDE 4. This is an example of what we might do with a
hand-held probe. Red light from a laser has been used to illuminate the
meat, then a small-window measurement of reflectance has been used to separate
fat from muscle. FAT DEPTH is the high plateau from 0 to 4 mm depth.
There are two measurements: WAY-IN and WAY-OUT. There
is an offset between way-in and way-out because of tissue deformation (which
can be used to gather interesting data on the softness of the tissue -
provided temperature is known, because tissues get harder with refrigeration
and surface drying). Software has to be quite clever to spot the muscle:fat
boundary because, as we will see later, sometimes the muscle can be very
pale. Within the muscle can be seen MARBLING (INTRAMUSCULAR) FAT.
SLIDE 5. A lot of things affect the quality of
meat. Some of them originate on the farm while others, especially
pH, are determined by transport of animals, method of slaughter, and rate
of refrigeration. As well as causing protein denaturation and major
differences in light scattering, pH affects the negative electrostatic
repulsion between myofilaments. If filaments
come close together, the native water between them moves out of the filament
lattice, and ultimately ends up as a massive fluid loss (drip and
evaporation) from the meat, as well as causing obvious problems in meat
packaging. In the laboratory, we follow the fluid movements by low-angle
x-ray diffraction, transmission electron microscopy and interference microscopy.
The challenge is to make comparable measurements with simple sensors!
SLIDE 6. Remembering or goal of understanding what we are measuring,
here is a way to investigate the effect of pH on the light scattering that
occurs within pork. Light scattering is a very strong effect.
High scattering causes a short light path with little selective absorbance
by chromophores so the meat appears pale. Low scattering allows a
long light path, strong absorbance by chromophores, so the meat appears
dark. The disk of pork (shown in red) is flushed with buffer (shown in
yellow) and measured using the optical fibre.
SLIDE 7. The pH of the 0.2 M phosphate buffer is controlled by
computer using acid, base and dump valves. Some flushing was required
to remove sarcoplasmic proteins. In retrospect, the ionic concentration
was a little too high, causing some solution of myofibrillar proteins.
Despite this, pH effects were reversible and predicable from what we know
of the effect of pH on the negative electrostatic repulsion of thick and
thin myofilaments in their lattice.
SLIDE 8. This shows the components connected to the fibre-optic light
guide in the experiment to study the effect of pH on the reflectance of
the disk of pork. A solenoid-shutter is required in the illumination pathway
to find the dark-field current of the photomultiplier. Programmable
stray-light filters are required after the grating monochromator to remove
high-order harmonics. Everything is driven from an IEEE488 bus.
SLIDE 9. This shows a dynamic test of reflectance (primarily
scattering) change with pH. The key point to note is that reflectance INCREASES
as pH DECREASES. A second point to note is that in this
sample of NORMAL pork, changing the pH (by an amount similar to that occurring
as a result of a normal amount of post-mortem glycolysis), caused a change
in reflectance of about 0.02 (relative to magnesium oxide = 1).
SLIDE 10. Now we repeat the experiment with PSE
(Pale, Soft, Exudative) pork. The reversible change in pH is about
the same order of magnitude as found with normal pork, but the relatively
small change is superimposed on a very high level of reflectance which
originates, now almost certainly as proposed by Bendall, from precipitated
or denatured sarcoplasmic proteins (not washed out because they are deposited
around the myofibrils).
SLIDE 11. Spectrophotometry through optical fibres is relatively
simple, but colorimetry is more complex. The main point is that colorimeters
have air spaces (from illuminator to sample, and from sample to spectrophotometer).
But when we insert optical fibres directly into soft tissue there is direct
contact between fibre cores and tissue fluids. Internal reflectance spectra
collected by fibre-optics tend to be related to reflectance spectra by
a third power of wavelength. After this transformation, the weighted ordinate
method may be used to calculate chromaticity coordinates for colorimetry
with reasonable success. Line 1 is the internal spectrum, line 2 is the
measured surface reflectance spectrum, and line 3 is the transformation
of the internal spectrum to predict the surface spectrum.
SLIDE 12. The shape of spectra collected by fibre-optics from
meat are very useful for steering a robotic probe. Although a robotic probe
can navigate relative to the skeleton (using an ultrasonic image of the
skeleton produced by transducers surfing on a water jet), it is difficult
to know when the optical window of a probe is within a muscle. Clearly,
if the measurement is made from fat between muscles, then the spectrum
will be misleading.
Each measurement in a spectrum is compared to every other measurement,
to produce a probability matrix (Pmat) filled with -1
or +1. Zero is rare if the full result of A:D conversion is used.
Several matrices for classic examples of fat or muscle are averaged to
make a cumulative probability matrix (Cpmat) with a range from -1
to +1. The Cpmat for fat is subtracted from the Cpmat for muscle to give
a difference in cumulative probability matrix for muscle (Dcpmat-m)
with a range from -2 to +2. Similarities are canceled. Differences
are enhanced. The mirror image Dcpmat for fat (Dcpmaf-f) is
created by subtracting the Cpmat for muscle from the Cpmat for fat. Thus,
the probability of an unknown Pmat originating from muscle is
SUM (Pmat x Dcpmat-m) / ( SUM (Pmat x Dcpmat-m) +
SUM (Mat x Dcpmat-f))
Unfortunately, the photocopier reduced 10 shades of grey-map dithering
to black and white in the matrices plotted for this slide! Here the example
was for the robot to separate connective tissue (CT, including fat) from
very pale muscle (PSE, pale, soft, exudative pork). The advantage
of this method is speed, since the matrix summation can be done rapidly
while the robot is moving, thus allowing the window to be halted
within a known type of tissue with a known probability of success.
SLIDE 13. Another important point to remember is that, because
meat has strong scattering, the penetration of light at different wavelengths
varies considerably.
SLIDE 14. This gets us to stage 6 in slide 2 - the future. Measuring
at different wavelengths and at different directions through the
sample makes it possible to look at a correlation matrix to search out
the most useful features for prediction. Although true goniospectrophotometry
is possible through meat with optical fibres (angle changes - but path
length is constant, as used for slide 13), it is much easier to move the
optical fibre under a sample and measure light transmitted through the
sample at different angles and different path lengths. This method was
first developed for laser scattering in meat by Birth and Davies
in the US 20 years ago in 1978 - now we can add wavelength to angle.
A convenient way to look at the matrix is to use the coefficient of correlation
as altitude and plot a surface contour map. A key point to note is that
the sign of the correlation may be reversed from low (400 nm) to high (600
nm) wavelengths. This is very important for anyone planning an inexpensive
broad-band measurement out in the field. If the band includes plus and
minus correlations, then the device will fail!
SLIDE 15. Let us move on to another important attribute of meat
quality - the toughness of cooked meat.
SLIDE 16. By a fortunate coincidence, COLLAGEN
and ELASTIN, the two dominant protein fibres that cause connective
tissue toughness in meat are both fluorescent, emitting blue-white light
when excited with UV. Not only this, but peak excitation is at 370 nm,
which can be approximated by the very strong 365 nm emission peak of a
mercury source. Furthermore, because fluorescence quenching (fading)
proceeds from the outside to the inside of a connective tissue fibre, the
RATE OF FADING GIVES FIBRE DIAMETER, and
fibre diameter is correlated with tensile strong. Finally, with luck holding
out to an amazing extent, PYRIDINOLINE, one
of the dominant cross-links at the molecular level (it can make even a
small-diameter fibre heat resistant and strong) IS STRONGLY FLUORESCENT!
SLIDE 17. A probe plus depth detector (as in slide 3) can be
fitted with a single optical fibre for fluorometry, by splitting excitation
and emission pathways using a dichroic mirror (enhanced by appropriate
low-pass and high-pass filters). Not only is the use of a single optical
fibre a simple method, it is the best method. Thus near-field effects
at the distal window of the fibre dominate the signal.
SLIDE 18. As the single-fibre optical window passes through the
tissue it generates a signal revealing the microstructure and strength
of the connective tissue within the meat.
SLIDE 19. To achieve my goal of understanding
what I am measuring in simple, scientific terms, a simple signal
processing algorithm is preferable, although doubtless a lot could be achieved
with neural networks and fractal dimensions, but then I
would not be able to understand exactly what I
am measuring. Fortunately, it may be shown that the peaks tend to be symmetrical,
and this then allows peak height and half peak-width to be found with a
few simple flags. Roughly speaking, width can give thickness, and height
can give fluorescence intensity, but the stereology is complex because
the probe does not pass through all tissue sheets perpendicularly. Thus,
some fluorescent structures are seen tangentially. And it is usually
found that direction of measurement through the meat (which is has a strongly
anisotropic structure at all levels) is critical for both repeatability
and success of prediction.
SLIDE 20. Electromechanical probes for detecting
meat toughness have a bad reputation. Many designs of penetrometer
and torque deformation needles have been investigated and have produced
poor predictions of meat toughness. However, this is possibly because
the depth of penetration has been too shallow, and the logic has been to
seek a peak measurement to characterize the system. Peak measurements
are risky because they may include a transient spike from an extraneous
source not related to the overall tensile strength of the meat. As a probe
is being pushed through meat, useful electromechanical data may be collected
simply by looking at the depth vector. Commercial probes may trigger the
A:D from the shaft-encoder reading depth, which is very economical and
ideal for a hard-wired circuit with a fixed register size for data.
But, if a more flexible system is used, with data collection at a fixed
rate and a buffer to catch the data, disorders can be detected in the depth
vector. Thus, when the probe slows, stops, or even recoils (when the tip
hits tough tissue), the event can be quantified and compared with the incoming
signal from the primary sensory. In the example shown, the fluorescence
signal peaks when the probe decelerates, thus indicating the cause of the
deceleration to be connective tissue. This provides useful ancillary
information, especially when we remember that connective tissue is only
one cause of meat toughness.
SLIDE 21. The stereology provides a fascinating view of how beef
animals grow their intramuscular connective tissues.
SLIDE 22. In a large experiment with good standardisation
of transport, slaughter, refrigeration and aging, the difference
between tough and tender meat is very obvious.
SLIDE 23. It is important to remember, however, that connective
tissue is ONLY ONE SOURCE of meat toughness. Thus, correlations of
connective tissue fluorometry with taste panel responses are moderate but
not very strong. Other factors than connective tissue are involved!
SLIDE 24. Strong connective tissue is not always a problem as it is
in beef toughness. In turkeys, the opposite condition may occur at
heavy body weights: the connective tissue may not be strong enough to hold
bundles of muscle fibres together when cooked turkey roles are sliced thinly
for the delicatessen counter. A miniature version of the CT-probe
for beef may be used on turkey breast muscles.
SLIDE 25. After fluorometry, turkey breast muscles were tested
rheologically to see how strong they were. A non-destructive test was used
(no point in crushing the samples!), which explains the notch at P1 to
V in the hysteresis area of the stress-strain relationship where the mechanical
force release mechanism was activated.
SLIDE 26. Spectrofluorometry also may be used to assess collagen
content. For example, chicken skin is nice to eat and highly nutritious,
but becomes a problem when the level is too high in a processed chicken
meat product. Here we see the emission spectra of chicken meat pastes or
slurry with various collagen levels.
SLIDE 27. And here is the t-statistic for fluorescence versus cooking
loss. Thus, high collagen content caused high cooking loss, and the
best wavelength for the sensor was around 480 nm. The match between slides
26 and 27 is fairly obvious in this case, but it is not always like this.
Sometimes the best predictions originate from the edge of the spectrum.
It all depends where the desired information content is strongest relative
to spurious information, error and noise.
SLIDE 28. Let's leave fluorometry, although we have hardly touched
some of its possibilities, and move on to birefringence. Meat is
made of muscle fibres (giant, multinucleated cells) which contain contractile
organelles - MYOFIBRILS. Myofibrils are important
because they confer on meat many of its pleasant textural properties (without
which, we might as well eat texturised vegetable protein instead). Myofibrils
are transversely striated by A (ANISOTROPIC) and I (ISOTROPIC) bands.
This optical anisotropy originates from the precise longitudinal arrangement
of protein filaments within the myofibrils. Thus, myofibrils are
birefringent, with two speeds of light, one along and one across the myofibrils.
When the myofibrils contract, A band length is constant, and I band length
decreases. Thus, overall birefringence tends to increase as sarcomeres
get shorter. To cut a long story short, the key point is that SHORT
SARCOMERES CAUSE MEAT TOUGHNESS (because there is more overlapping of thick
and thin filaments which, when locked in rigor mortis, produce a dense
mat of protein for us to chew through). In slide 28, we are scanning down
the length of a myofibril with a polarised light microscope, showing
how we hope to measure sarcomere length from birefringence.
SLIDE 29. In the real world, however, we cannot isolate individual
cells and organelles very easily, so the challenge is to get the method
working on bulk tissue. Step one was to work with a 1-mm slice of
tissue, as shown here.
SLIDE 30. Here we are testing the relationship
between the transmittance of polarised NIR light and sarcomere
length using samples of cold-shortened pork versus restrained pork
(unable to shorten as a consequence of too rapid refrigeration).
As the analyser is rotated (degrees on x-axis) the correlation changes.
If the effect had nothing to do with polarisation, it would not change.
But it only changes with angle for restrained samples, not cold-shortened
samples with very short sarcomeres. NIR is used to reduce scattering
(which is inversely proportional to a power of wavelength).
SLIDE 31. Why doesn't the method work with very short sarcomeres?
The answer is that birefringence is a measure of ultrastructural neatness.
When filaments are neatly lined up, there are two refractive indices as
the electric vector of light interacts with charged groups of amino acids
on the neatly aligned filaments. Unfortunately, thick filaments are
2.5 micrometres long, and thin filaments are 1 micrometer in length. Thus,
as the sarcomere contracts from stretched length (say 3.5) down to less
than 2.5 micrometres, then the filaments overlap and bend as the thin filaments
jam into the Z-line. Thus, birefringence is reduced below 2.5 micrometres,
as we see here.
SLIDE 32. Now the sensor is getting complex!
It is detecting sarcomere length, provided that we do not have any severe
cold shortening, but it is also detecting other things with NIR that
may or may not be independent of the plane of polarisation. First we will
see what NIR transmittance per se is doing, then we will come back to the
problem using polarised light for bulk tissues. Here we see that the transmittance
of polarised NIR can predict the water-holding capacity of the sample (in
this case, a paste of turkey meat being used in food processing).
SLIDE 33. NIR reflectance and now transmittance
are popular methods with much to be said about them. To cut
a long story short, it is the fat content of the sample that is also being
detected by NIR. Fat also has a major effect on meat quality and
processing properties. Let us return to the more innovative aspects of
how to develop a probe for bulk tissue. Cutting slices of tissue
is no good - we must be able to push the probe into an intact system, as
we see here. The polarisers are mounted on the end of a bifurcated
light guide, the illuminating polar being perpendicular to the receiving
polar. There is no way to rotate the receiving polar so it is not
an analyser. A depolariser is need for the mirror which is used to
set 100% reflectance.
SLIDE 34. The bulk-state sensor still
works (fortunately!). Here we see it separating between rest-length
and stretched pork. On the y-axis is the back-scatter of NIR:
in other words, NIR that was rotated or depolarised to get back through
the receiving polar. We are cheating a bit here, however, because
we are looking at pre-rigor pork. In other words, scattering is at
a minimum (because the pH has not yet dropped from anaerobic glycolysis).
Scattering is the big problem - when it is high, the back-scatter is telling
us about pH as well as sarcomere length. Unfortunately, we are not trying
to measure pH, and we must be careful to examine the causality of correlations
(the method might appear to work in predicting meat quality, but it may
be detecting the effect of pH on meat quality rather than the effect of
sarcomere length on meat quality). Remember our goal of understanding
what we are doing, rather than exploiting blind correlations?
SLIDE 35. This brings us right up to date
with a bulk probe which IS capable of rotating its analyser. Although
polarisation preserving fibres certainly are available, they are very difficult
to use for bulk tissues - the fibres are too small, and the throughput
of light is severely limited. Thus, the problem is how to locate the rotary
analyser near the bulk tissue, before the optical fibres connected to the
photometer. One answer, as shown here, is to use a graded index lens
between the tissue and the analyser, so that the analyser can be rotated
within the housing of a hand-held probe under computer control. It is capable
of quantifying the amount of Fresnel reflectance from mirror-like
structures within the tissue - such as membranes and refractive index boundaries.
Thus, it has enabled us to measure the contribution of these factors to
overall meat paleness.
SLIDE 36. Using an extinction coefficient
(k) to quantify the proportion of reflected light maintaining its original
plane of polarisation (that is, Fresnel reflectance from reflective structures),
the probe can predict the processing properties of a saline chicken meat
paste.
SLIDE 37. Despite this progress in technology
transfer from innovative research to practical application, many
powerful methods remain trapped in the laboratory, such as the de Senarmont
method for ellipsometry which is exquisitely sensitive to the state of
the muscle fibres in meat and could have tremendous application in industry.
The difficulty is light scattering!
SLIDE 38. Thus, path difference provides a
neat explanation for dark cutting in beef. At a high pH, birefringence
is low, thus suggesting refractive scattering is low. Thus, light
incident on the meat is transmitted deep into its interior and the product
looks very dark.
SLIDE 39. Coming full circle back to the importance
of innovative research, sensors can be particularly useful when incorporated
into classical experimental apparatus, such as the rigorometer developed
in 1939 for the study of post-mortem metabolism in muscle. Here the scientific
question is why are 45 minute post-mortem measurements an unreliable guide
to the ultimate quality of the meat? Apart from the obvious point
that post-mortem metabolism has only just started at this convenient time
for industrial measurements (when carcasses move from the kill-floor to
the meat cooler), the problem may originate from transient osmotic uptake
of intercellular fluid because of glycogenolysis (big molecules splitting
into lots of smaller ones). Impedance measurements are particularly useful
because capacitance is strongly correlated with ATP level. This is
exploited in many of the commercially available meters for testing meat
quality - but they do not work very well at 45 minutes!
SLIDE 40. Here is the result of an experiment. The strip
of pre-rigor pork was periodically loaded. At it developed rigor mortis
(caused by lack of ATP and measured by NMR), the stress-strain hysteresis
area and amount of elongation decreased. These classic changes were
highly predictable from the fibre-optic sensor.
SLIDE 41. Electrical capacitance of cell membranes is a
good measure of ATP availability (ion pumps in cell membrane are open without
energy from ATP, thus providing a shunt between electrolytes on each side).
In the laboratory, the method can validated by NMR . Impedance measurements
are relatively simple in theory, but the main problem we have in the meat
industry is the high static charge on many hanging carcasses. This
is surprising, since the abattoir air is very humid, the carcasses are
wet, and the overhead rail may be grounded. Battery-operated apparatus
is the simplest solution.
SLIDE 42. Hybrids between dynamic rheology and optoelectric sensors
can be made portable and moved out into the real world. Optical fibres
are mounted in hypodermic needles, thus allowing spectrophotometry of what
ever is between the tips of the touching needles. Similarly, the
metal of the needle provides two electrodes for impedance measurements.
Moving one needle away from the other allows us to test the rheology
of the sample, while the amount and rate of entry of tissue fluid into
the space can be followed optically - before a final scan for spectrophotometry
and impedance at different frequencies. There are no results yet, this
one is still a prototype being programmed!
SLIDE 43. Finally, it is important to remember that slaughtering
is one of the main operations in the meat industry. Many parameters
can be monitored indirectly using load cells and electrical transformers
- including the completeness of exsanguination (bleeding), the voltages
and amperages of stunning currents (monitored to ensure that stunning is
effective and humane), and reflex activity. Reflex
activity is particularly important because it accelerates glycolysis,
enhancing the pH-decline rates which earlier we saw had a profound effect
on meat quality.
SLIDE 44. The patterns of reflex activity detected by a load cell as
the carcass bounces may be used diagnostically, sometimes enabling the
source of activity to be tracked to a particular part of the carcass.
Different types of reflex activity are useful in checking whether stunning
has been done humanely.
SLIDE 45. EMG can take us down to the cellular level in understanding
which muscle fibres are contracting, where we can relate to action
potentials and motor units. Thus, from a relatively simple device
such as a load cell, it is possible to gain an amazing amount of information.
Perhaps, eventually, we may have sensors throughout the meat industry enabling
true quality management to improving average meat quality. In doing so,
we also have the opportunity to enhance our scientific understanding of
meat quality.
WANT MORE INFORMATION
?
[1] MEAT PRODUCTION
H.J. Swatland (1994) Structure and Development of Meat Animals. Technomic
Publishing, Basel ISBN 1-56676-120-4
[2] MEAT SENSORS
H.J. Swatland (1995) On-line Evaluation of Meat. Technomic Publishing,
Basel ISBN 1-56676-3339
[3] SOFTWARE
H.J. Swatland (1998) Computer Operation for Microscope Photometry.
CRC Press, Boca Raton, Florida ISBN 0-8493-1697-9