News and analysis on the global poultry
and animal feed industries.
on June 25, 2009

Testing feed quality: the 'artificial senses'

Electronic senses may mean new tools for accurate monitoring of feed ingredients, opening new possibilities for precision of quality management and safe feed programmes.

Increasing interest in available feed and food supply worldwide has stimulated concerns over safety and quality issues, and given rise to legislation on traceability, control and labelling in order to prevent food crises in the future. This pressure for safe and high quality food, as well as company policies for enhancing internal quality assurance (QA), has also accelerated the need for better analytical control of feed and food. Moreover, today's increasing volume, rate of production and competitive marketplace needs fast, reliable information to allow manufacturers to make sound business decisions.

The quality of a product is closely related to its physical and organoleptic properties. The use of the senses, such as vision and smell, are rapid field methods for feed quality control. Techniques based on the use of electronic nose, electronic tongue and image analysis may represent an evolution of sensorial analyses traditionally entrusted to human senses. They can provide quantisation of quality and safety in real time with the objectivity of an instrumental response and are likely to have wide applications in feed research and industry both as tools for rapid screening and quality control and as a support for decision-making in the area of product quality.

Electronic nose

Unlike the case with hearing or vision, some of the physiological principles of the sense of smell are still unclear. Odours are made up of up to several thousand chemical components that are generally light, small, polar and often hydrophobic. Only sporadic work on the design of artificial olfactory systems was performed up to end of the 1980s, when researchers introduced the term "electronic nose" and the concept of "artificial olfaction." In the early 1990s, commercial instruments became available on the market and immediate use in the feed and food industry became apparent.

An electronic nose is composed of an array of sensors, a data pre-processor and a pattern recognition system. Each sensor in the array is sensitive to a different range of chemicals. The development of a sensor array is different from that of conventional chemical sensors which selectively detect specific chemical substances. The electronic nose does not distinguish each volatile substance, but instead expresses the global odour of a product. Sensor array formats interact with different volatile molecules and provide an electronic signal that can be utilised effectively as a fingerprint of the volatile molecules associated with the product.

Different sensor technologies for electronic nose are available. The main commercially available are MOS (Metal Oxide Semiconductor), MOSFET (Metal Oxide Semiconductor Field Effect Transistors), polymer and surface acoustic. The response rapidity and recovery time, the operative temperature, and the sensitivity to disturbance represent the main differences among sensors in response to organic vapour. Sensor arrays may be tailored for a specific product since it is almost impossible to have a universal sensor array capable of analysis of all odours associated with a product.

Interpreting data

Two main data analysis methods can be applied for electronic noses: exploratory and predictive methods. The first one allows the user to discriminate and classify the odour of different products. The second one is designed to find a prediction model to describe product characteristics when there is prior knowledge of their existence. Whatever is the goal of the data analysis, an electronic nose has to be trained to fulfill discrimination or prediction capability of the pattern recognition system. The methods more often used for electronic nose data analysis are chemometric approaches such as Principal Component Analysis (PCA), Cluster Analysis (CA), Discriminant Analysis, Multiple Regressions and machine learning method. Once the electronic nose has been trained and the appropriate data analysis method has been selected and applied, the response can be obtained in real time. Therefore compared with traditional odour analysis methods (sensory panel and gas chromatography), the advantages of the electronic nose are high sensitivity, easy sample preparation, non-destructive operation, fast detection, lower cost and objectivity.

Applications in feed

The application of the electronic nose technology in the feed industry is not as developed as its use in the food industry. However, a promising field of application is in feed safety control in order to evaluate mould spoilage. Recent studies have demonstrated the electronic nose capability in order to discriminate between non-contaminated and infected samples with different species or different strains of toxigenic fungi. Studies indicate that electronic nose technology can be used as a screening method for simple and rapid detection not only for fungal contamination but also for mycotoxin presence in feed and food. Knowledge and control of mycotoxin contamination and distribution in feed and food is a world-wide objective of producers, manufacturers, regulatory agency and researchers due to the high economic and sanitary impact on food safety and human and animal health.

A further application of electronic nose in the feed industry and an important issue in feed safety comes from the identification of products of animal origin, generally known as processed animal proteins or PAP, in feedstuffs. In addition, recent results suggest that the electronic nose may have a potential application in the evaluation of fat quality in pet food. Therefore, electronic nose represents a promising and powerful tool able to provide immediate and satisfactory answers in a wide range of feed industry applications.

Image analysis

Image analysis has a longer history than electronic nose. Today, image processing and image analysis are recognised as being at the core of computer vision. Image analysis requires the acquisition of an image of the object, which can be performed by different equipments such as cameras, scanners, microscopes, and laser system. The optical image is converted by digitisation into digital format (numerical form) and further processed in order to assesses morphological, geometric and chromatic variables and produce quantitative information, which are used in the subsequent control systems for decision making. Data analysis in computer vision uses a similar statistical approach as already described for electronic nose. This is an important stage of the process, and must be performed correctly in order to obtain reliable data that is useful for decision-making in the area of product quality. In general, 2-dimensional (2D) data are needed for grading, classification and analysis of most agricultural images. However, 3-dimensional (3D) image analysis may be needed in order to obtain more details on structure and characteristics of a product.

Analysis application

Computer vision systems have been investigated for many applications involving grains and oilseeds and may have a wide application in the field of animal nutrition, feeding and health. Various computer vision rapid testing technologies based on color, size, morphology and defect of grain are already commercially available. In the feed industry, determination of the optical shape, size and distribution of individual particles is widely used and represents one of several nutritional and quality control parameters. Instrument devices and market systems of measurement, via image analysis, of particle diameters and relative proportions of the various sizes are already available and used. This analytical approach, compared with traditional ones (e.g. analysis of particle distribution by mechanical sieving), enables the user to obtain the granulometric features of a population rapidly and can be easily and quickly used in laboratory and/or routine conditions.

Computer vision has been applied with success in order to understand the structure, ultra-structure and chemistry of grain components, which are the basis for developing industrial processes for cereals. A variety of microscopic 2D and 3D techniques are available. Moreover, there is the possibility to associate image analysis (visible image) and spectroscopy analysis (e.g. "chemical image" generated by FTIR microspectroscopy) in order to explore molecular and chemical features of the microstructure of a product. This approach has been applied to reveal microstructural-chemical features of corn, usually destroyed with the traditional "wet" chemical methods.

Researchers have concluded that this molecular chemical information can be linked to nutritional information and can be used to select varieties of corn for targeted feed purposes and for prediction of corn quality and nutritive value for animals. Although these last image analyses are very promising and well-developed at the research level, they are still far from being used as rapid testing technologies because a complex and time- consuming sample preparation is required. Another interesting example of image analysis application in the feed industry is represented by the control for the presence of animal by-products.

Electronic tongue

Besides olfaction, taste is the other "chemical" sense. Several chemical substances are responsible for the taste of a product. Humans do not distinguish each chemical substance, but express the taste itself that is comprised of five basic qualities: sourness, saltiness, sweetness, bitterness and deliciousness. Like the electronic nose, the electronic tongue is an analytical instrument comprising an array of non-specific, low-selective chemical sensors with partial specificity (cross-sensitivity) to different components in solution, an appropriate method of pattern recognition and/or multivariate calibration for data processing. The methods for electronic tongue data analysis are the same already described for the electronic nose. If properly configured and trained, the electronic tongue, like the electronic nose, may be capable of recognising complex liquids of different natures and eventually be used to determine quantitative composition. Like the electronic nose, a unique feature is the possibility to maintain a correlation between the output of the electronic tongue and human perception. Sensor array of the electronic tongue mainly consists of arrays for liquid sensing based on potentiometry and voltammetry electrochemical methods. The main differences are in regard to the sensor array sensitivity, selectivity, versatility, and robustness.

Electronic tongue applications

The electronic tongue can recognize the taste of a liquid or liquefied product. Currently the main application of this technology is as a classification tool in food analysis with the aim of discriminating among different kinds of liquid foodstuffs according to different quality parameters. Applications can be found in beer, tea, coffee, milk, mineral water, and wine. The applicability of the electronic tongue to solid food has been demonstrated in tomatoes and in fish. No data are yet available regarding specific applications of the electronic tongue in the feed industry, but applications for the analysis of feedstuffs are possible and foreseen. An interesting application area of the electronic tongue alone or associated with the electronic nose could be in the evaluation and standardisation of palatability of raw material and complete feeds when ingredients or additives with low acceptability, according to peculiarity and preferences of different animal species, are incorporated.

On-site technologies

The "artificial senses" can be considered a great example of how science and engineering work together to produce something of real utility. These "fit-to-purpose" analytical methods are rapid, user-friendly, adaptable, and coherent with the precision and accuracy level requested by the feed/food chain control and regulatory purposes, and useful for decision-making in the area of product quality. In the future, technological advances such as faster responding sensors, automated sampling systems and unsupervised data analysis will make it possible to use these technologies on-site for real time monitoring and control of feedstuffs and industrial processes.

Comments powered by Disqus