For the meat industry, reliable meat quality monitoring is of fundamental importance for profitability and consumer satisfaction [1]. Severe quality defects in meat are typically associated with unwanted attributes such as unacceptable appearance (pale color), reduced tenderness and juiciness or structural disintegration [2, 3]. The latter is especially relevant for pork, where “destructured” meat poses a challenge to high-quality ham production [4]. Destructured meat resembles typical features of PSE (Pale, Soft, Exudative) meat, which has been a global challenge for the pork sector in the last few decades [5]. The defect has been linked to various factors, such as rapid post-mortem pH decline, specific mutations, oxidative stress, and enhanced apoptotic processes [6–9]. While efforts have been made to eradicate mutations that can cause the defect [5], the prevalence of severe pork quality defects can be as high as 20 and even 50% in several European countries [3, 10, 11]. The terms PSE-like and destructured meat are often used interchangeably and describe a syndrome where several quality loss features coincide, i.e., color-, texture-, and pH anomalies with tissue disintegration and very low water holding capacity [12, 13]. Especially, the latter two features render PSE-like loin or ham unsuitable for costly curing processes. This is because the loss of these technological qualities often results in final products that fall apart when sliced and, hence, cannot be sold [3]. To at least reduce the financial burden of PSE-like pork, there is a need for objective test methods that can be used to detect and sort raw ham or loin with defects, before entering the costly processing streams for dry and wet curing.
Traditional methods for PSE-classification have been based on pH, color, and drip loss testing or on visual evaluation [14–18]. Yet, such PSE meat defect classification systems are not consistent among authors and are typically made for specific muscles, most often loin and less often ham muscles. Additionally, these methods have limitations in detecting specific quality features and may be less suited to predicting the prevalent structural defects that are typical for PSE-like ham [19]. Apart from PSE detection, other technologies like ultrasonic spectral analyses, X-ray measurements, and optical spectroscopy have been proposed as alternative quality assessment tools for pork meat quality [20–23]. In the absence of accessible instruments for objective measurements in meat-cutting plants, subjective, visual evaluation can offer a low-cost approach. Several subjective visual guidelines have been developed for pork defect evaluation [24–26]. As with other subjective tests, however, these methods have short-comings, e.g., a need for experienced users and poorer reproducibility compared to instrument-based testing. In addition, subjective color evaluation can also be biased by context, i.e., by different lighting, and also can be prone to day-to-day variations because of imperfect color memory [27].
Bioimpedance spectroscopy can offer a more reproducible, inexpensive, and fast method for testing diverse technological qualities of meat. Studies suggest the use of bioimpedance-based testing, particularly by using the
To this end, we studied the relationship between objective, instrument-based tests, including bioimpedance spectroscopy, and a subjective, visual quality test method. The visual test was based on an established reference for PSE-like ham defect evaluation. This allowed for scoring the severity of tissue and color defects [24], and in the following will be referred to as visual DES scoring. We asked how the different instrumental test methods predicted the visual DES score and how such prediction can be improved by establishing multivariate prediction models.
Pork (
The ham cuts were tested using subjective visual scoring and objective, instrumental methods. The latter included bioimpedance spectroscopy and the two most widely used meat quality tests, i.e., pH- and color-testing. Performing also drip-loss testing was not feasible, as sampling for drip loss would have largely destructed the valuable ham cuts, and transporting the cut up, smaller pieces to a laboratory would have caused drip to set in before controlled testing can commence. Measurements were made in the chilling room of the slaughterhouse. Common meat color coordinates (CIELAB/
To allow for blooming, meat color was determined after the surface has been exposed to air for at least one hour. For calibration, a white ceramic calibration cap CR – A43 was used. The light source was a pulsed xenon lamp (Konica Minolta Sensing INC, Tokyo, Japan). The ultimate pH (pHu) was measured using a WTW pH 3110 (WTW, Germany), equipped with a BlueLine 21 pHT electrode. The pH-meter was calibrated before measurements using fresh buffer solutions with pH of 4.0 and 7.0. For each ham cut, pH and color values were recorded at two different anatomical locations, i.e., in the central region of each muscle (SM, AD; Fig. 1). The two muscles were included as there are known muscle-specific differences, e.g., in pH and color [3].
Next, bioimpedance was measured using a Zurich Instruments MFIA impedance analyzer (Zurich Instruments AG, Switzerland). Spectra were recorded for a frequency range from 10 Hz to 1 MHz, with 40 distinct frequency points and an applied voltage of 300 mV rms. A tetrapolar electrode setup with two signal-generating and two receiving electrodes was used. Electrode pins were made of stainless-steel with a diameter of 2 mm and a length of 12 mm. Pins were aligned in one row with 18 mm spacing between the middle, voltage pick-up electrodes, and 14 mm between the middle, voltage and the outer, signal-generating electrodes.
Shielded cables of 1 m length were used to connect the impedance analyzer and the electrode socket. Similar to pH and color testing, bioimpedance was measured at the two anatomical locations indicated in Figure 1. Two readings were recorded for each location of the electrode. The electrode was cleaned after every measurement.
Bioimpedance is a passive electrical property which is defined as the ability of the biological tissue to impede (oppose) the flow of electrical current. Bioimpedance can be measured by applying an electric excitation signal (either current or potential) and picking up the response of the tissue through electrodes, which convert the electronic charge to ionic charge and vice versa [33]. The complex electrical impedance produced by biological tissues is also called bioimpedance and can be expressed by the ratio of the voltage (V) and current (I).
The complex electrical impedance of various organic tissues, including meat, results from contributions of both, tissue capacitance and conductance. This relation is expressed as,
Here R∞ and R0 are the resistance or the electrical impedance modules at high and low frequency, respectively. τ is the characteristic time constant of the system corresponding to a specific angular frequency ω = 1/τ = 2πfc, where fc is the characteristic frequency, which corresponds to the frequency at which the absolute value of the imaginary part of impedance is largest, and α is the distribution or shape adjustment and interaction parameter.
The specific relations of these parameters to structural changes in biological tissues are not completely clear. However, the normalized difference of
The
For visual evaluation, we adapted a ranking system for features related to tissue destruction (DES) and color established by IFIP [19, 24]. Briefly, the first step was to expose the inner part of the SM and AD muscles that were in close contact with the femur bone (
For testing relations between subjective DES scoring and instrument-based quality monitoring, average (N = 111 samples) and joint values (“training”, N = 25) of the two judges were used as a final DES score for each individual sample (N = 136). The observers gave DES scores according to the feature sets detailed in Table 1.
The four-rank visual DES scoring scheme for destructured ham cuts, including the semimembranosus and adductor muscle. Scoring was based on evaluating structural disintegration and visual color (adapted from [19, 24]).
Score | ||||
Visual colour | Reddish (> 3) | From pale to reddish (1 - 3) | Very pale (1 - 2) | |
Muscle structure defect | Compact fibre structure | Absence of fibrillar structure in the affected area Destructured meat |
Absence of fibrillar structure in the affected area Soft and doughy, destructured meat Fluid exudate |
|
Area affected | None | Small areas on the surface with single patches of destructured zones | More than 50% of both muscle areas Lesions beneath the surface | |
Observations | No visible defects | Small, pale areas on the surface | Lesion less than approx. 2cm in depth | Lesion more than approx. 2cm in depth |
Average scores were grouped into different meat defect ranks as follows:
0 to 0.5 = DES0: with no defects detected, 1 to 1.5 = DES1: mild, 2 to 2.5 = DES2: moderate, 3 = DES3: severe.
The diverse sets of data were obtained from different analytical tools (electrical impedance analyzer, pH meter, colorimeter) and from visual scoring. Analysis of potential links among the DES score and quality parameters was conducted using the Pearson correlation coefficient (r). Calculations, correlation plots, and charts were generated with Minitab version 19 (LLC, Pennsylvania, USA). Significance levels were
The research relates to the use of animal products and complies with all the relevant national regulations and institutional policies for the care and use of animals.
Subjective visual evaluation for PSE-like defects was performed by two independent evaluators following the visual scoring scheme (Table 1). Figure 2A shows the distribution of averaged visual scores for 111 ham samples. We found that all visual DES scores from 0 = no defect to 3 = severe defects were present in the sample population. This suggests that the sample population showed sufficient heterogeneity with samples ranking from normal to very poor quality, which is a prerequisite for exploring correlations with other test parameters.
There were detectable differences in subjective defect scoring among the two observers (compare Figs. 2B and C). Yet, r = 0.62 supported a strong correlation between the two observers’ DES scores. When calculating how good observer 1 DES scores can predict observer 2 data, we found a prediction error of 0.85 for a second-degree polynomial regression.
We tested the bioimpedance response, based on the
We then tested how visual DES scores compare with
This prompted us to ask, if the prediction of visual DES scoring can be improved by stepwise regression modelling and by including more parameters than just
We show that a linear multiple regression model (N = 136, see below) can provide a markedly higher correlation (r = 0.71) than we found for correlations of visual DES scores with only one instrument-based parameter only. The prediction error of the model was 0.76, i.e., the average deviation of the model prediction from the actual scoring data is less than 1 DES score point. Supplement 1 shows the model statistics, including the P values for each estimated coefficient.
Lastly, we performed a stepwise multiple regression analysis of DES scores with
Using an established reference for subjective evaluation of pork quality defects (DES), we found the highest correlation for
Correlations of subjective quality scoring with the different physicochemical quality variables (
There are several likely explanations for different correlation strengths among different quality parameters and among different studies. For example, the well-known (“classic”) PSS or PSE defect is caused by a mutation that can generate a very distinct and easily detectable pH drop early post-mortem. In contrast, the “PSE-like defect” is a rather loosely defined term or syndrome, which is solely based on a set of detectable features, and is collectively applied to distinct, well-known defects (PSE/PSS, acid meat) as well as to defects with similar features (pale, soft, destructured), yet unknown causation. Depending on specific genetic or physiological predispositions, and also on causations linked to suboptimal pre- and post-slaughter practices, correlations between quality parameters may therefore vary within a test population or between studies. It is therefore conceivable that the relative expression of defect features and, hence, specific quality parameters, can be either dominated by pale discoloration (“color”) or destruction (“
Based on the above results and as before [13], we conclude that “
Our data corroborates a previous bioelectric study, which proposed the detection of PSE defects through combining the
Follow-up studies are needed to test and improve the multivariate model by including larger data sets, also from other pig populations. In addition, applied instrument-based PSE-like detection classification strategies must be developed, in particular for segmenting continuous output data into binary (defect/no-defect) or ordinal ranking data (no/moderate/severe defect).
Subjective defect scoring can be a relatively cheap, low-barrier tool to classify raw pork meat and prevent cuts with severe quality defects from entering pricey and long-lasting processing streams, such as curing. However, subjective scoring schemes, including the one we used here, can suffer from insufficient normalization or reproducibility. Specifically, color perception can change in cutting plants with windows such as the one we performed our tests at in autumn, a period with rapidly changing daylight conditions. In addition, subjective bias inherently affects reproducibility, as we demonstrate with Figure 2. We cannot exclude that performing more tests over a longer period of time may have gradually reduced inter-individual observer differences. However, assessing pork defects involves haptic evaluation, visual color, and pattern of a cut with a relatively complex muscle anatomy [41]. Also, unlearning from visit to visit, and changed perception with context changes, i.e., with meat from different cutters, cannot be completely overcome in our or similar studies. However, for real-life applications, different evaluation (“perception”) by different cutting plants, cutters or sensors might contribute most to reduced reproducibility in industrial settings. Together, all this makes subjective scoring a typically less reliable reference standard for developing instrument-based detection tools. However, while it would be preferable to improve the detection capacity of novel tools by using instrument-based data as a standard, there is no unified physicochemical standard for PSE-like defects established yet [see [13] and references therein].
In addition to physiological and likely genetic variation among samples with PSE-like defects (see above), also instrument-specific challenges can reduce reproducibility and correlation-strength between different quality parameters. Specifically, meat pH is very location dependent and can vary markedly even in the same muscle and with measurements taken close to one another. This can add to the technical variation, that is known for measuring different muscles within the same individual (compare Fig. 3 for
Our study shows how subjective pork defect scores can be predicted by bioimpedance response (