In the first part of the article, the authors explained the physical mechanisms of flavour and fragrance perception. They introduced the two basic flavour-profiling methods. The second part of the article provides an analysis of the applications of the dynamic flavour profiling method and the aplications of the electronic nose in the food processing industry.

Applications of Dynamic Flavour Profile (DFP)
The following examples illustrate how the DFP approach can be applied to address various problems:

1. The product profile dissipates at the end upon the addition of another ingredient. The ingredient imparts a desired characteristic elsewhere, but not in the background. Therefore, its elimination is not recommended.

The first choice is to modify the ingredient so that its effect on the product profile is minimised or eliminated. If the ingredient has a flavour itself, profiling it might help in determining what adjustments are feasible. Another approach might be to determine which of the 14 attributes have been affected by the addition of this ingredient. A before and after comparison of the DFP curves will illustrate this. By pinpointing the specific characteristics or nuances that are lacking, we can make adjustments by modifying a flavouring to augment those notes.

2. A product is modified to reduce the fat content. A change in profile is observed in the top note and a loss of overall creaminess is detected in the middle.

One approach might be to determine the effect of different types of vegetable oil or oil replacers on the product. The DFP curve for each variable can be analysed, and the oil with the closest overall profile to the original could be selected.

3. When the product is processed, an off-note develops. The DFP method can be used as stated
above, to select an additive that appears at the same time as the off-note and masks it.

4. The product flavour fades upon extended shelf life, but previous direction to suppliers has been fruitless.

Since the 10 odour types used in the DFP method are analogues to chemical structure and terminologies commonly used by flavourists, pinpointing specific time- and component-oriented descriptors should make the job as defined and targeted as possible.

5. The fruitiness of the product has been reduced. Fruitiness is usually associated with esters. Since esters are volatile, the heat added to the product has to be minimised. However, all attempts to address the problem in this manner have been of no avail. The DFP method determines that the ester profile is in fact high in the front but lower in the middle; this seems to be the area that needs support. Therefore, fruit flavours that contain middle volatile esters (amyl C-2, ethyl C-5 to C-7), such as apple, pineapple and banana flavours, would be an appropriate choice to minimise the problem.

Electronic nose
Since the graphical depiction of the DFP is somewhat complex, the later developments in computer technology come at the most opportune time. One such development is the electronic nose, which is merely a series of sensors that respond to volatile components of the headspace above a sample (Michael et al., 2000). The experimenter optimises operating conditions and teaches the electronic nose what to recognise through teaching sets and data libraries.

The software of the electronic nose uses a series of algorithms and chemometric methods to provide meaningful data from the sensor response. As this is a correlation technique, one must remember that the electronic nose will only be, at best, as accurate as the analytical or sensory data it is correlated against.

In recent years, significant interest in the use of sensor arrays to discriminate between odours has arisen. If an array of non-specific sensors could be compiled to rival the human olfactory system, then the samples need not be separated and could be monitored analytically as a whole (Gardner and Bartlett, 1992; Kress-Rogers, 1997).

The term ‘electronic nose’ is applied to such an array of chemical sensors, where each sensor has only partial specificity to a wide range of odourant molecules, coupled with a suitable pattern recognition system.

A sensor array used for discrimination between odours was first demonstrated by Persaud and Dodd in 1982. The subsequent improvements in both sensors and data analysis methods have led to the development and marketing of several electronic noses for commercial applications.

Principles of olfaction
To appreciate the operation of an electronic nose, it is necessary to have a basic understanding of the principles behind the human olfactory system. The sensation of smell is dependent upon the interaction of odourant molecules with a group of specialised nerve cells. The olfactory receptors are situated just above the bridge of the nose out of the main air stream. The hydrophobicity of a molecule is important, since the first step in the process of olfactory recognition is the dissolution of the molecule in an aqueous mucous layer covering the olfactory receptor cells. Each olfactory cell has a number of cilia, effectively increasing the surface area, which contains different guanine nucleotide-binding proteins (Breer, 1994, Clapham, 1996). The genes encoding the proposed olfactory receptor proteins have also been identified (Axel, 1995). The significance of the smell is illustrated by the sheer weight of genetic information used to control it. About 1,000 genes encode 1,000 different odour receptors. Thus, the odour receptor genes account for approximately one percent of all human genes. In the olfactory receptor structure, each protein is thought to transverse the cell membrane seven times, forming a pocket-like structure into which the odour may bind specifically (Brur, 1994).

Many theories, such as the stereochemical theory of Amoore (1970) and the vibration theory of Wrigh (1982), have been proposed to explain the process of olfactory recognition. The exact method of interaction between an odour molecule and a receptor site has not been proved conclusively, but it is known that interaction results in excitation of the receptor cell. This produces a cascade of reactions with activity control in channels within the cell membrane, resulting in an electrical signal that is passed along the axon to the olfactory bulb. Two transduction pathways have been identified using separate second messengers, i.e. cAMP and IP3, which are thought to act in opposition to each other (Breer, 1994).

The human olfactory system is quite remarkable in its ability. The number of distinct types of binding proteins is small (about 1,000), while the number of olfactory cells is large (about 100 million). The same proteins must be present in many different olfactory cells. Humans can detect at least 10,000 odours; therefore, each odour must bind to several receptors, producing receptors which are described as having partially overlapping sensitivities. However, there are only about 50 identified odours, far fewer than is possible.

The olfactory bulb is composed of three main layers. Thus, stimulation of a combination of olfactory receptors results in the formation of a two-dimensional topographical map representative of a particular odour. The electrical signals produced are further processed by the mitral cells and finally sent via the granular layer to the brain. The overall function of this stage is to reduce the noise associated with the signal and amplify it, effectively increasing both the sensitivity and the selectivity of the system. The electronic nose system parallels the human olfactory system in the following manner: Each chemical sensor represents a group of olfactory receptors and produces a time-dependent electrical signal in response to an odour.

Any noise and sensor drift may be reduced using signal pre-processing techniques. The final stage in the human olfactory process is the cerebral cortex of the brain, which classifies and memorises odours (Schild, 1990); the equivalent process in the artificial nose is the use of pattern recognition software. So the advent of artificial sensory systems able to minimise chemical senses, such as those of electronic noses, has opened up a variety of practical applications and new possibilities in many fields where the presence of odours is the phenomenon under control (Kress Rogers, 1996).

Application of the electronic nose

1. To monitor maturity and shelf-life of tomatoes
An electronic nose (E-nose) was used to evaluate the maturity and monitor the shelf life of tomatoes. Results showed that fruits in the pink stage could be distinguished from those in the light red stage and red stage by the E-nose using principal component analysis (PCA) and linear discriminant analysis (LDA).

The E-nose was able to classify samples having different firmness using PCA. By means of partial least square (PLS) based on E-nose sensor response signals, fruit firmness was predicted and the correlation between the predicted and measured values was 0.936. A clear distinction of initially light red fruit during storage from Day 1–6, 7–11 and 14–17 was obtainable by the electronic nose using PCA and LDA (Jun and Yibin, 2007). Research has shown that the E-nose has sufficient sensitivity and resolution to evaluate tomato fruit maturity and product firmness. The shelf life of light red stage and red stage fruits can be satisfactorily evaluated also.

2. Correlation between electronic nose response and head-space volatiles of Tongkat Ali (Eurycoma longifolia)
Most herbs have their own characteristic smell due to the presence of volatile  compounds. Traditionally, volatiles are analysed using sophisticated and expensive gas chromatography (GC) in tandem with selective mass spectrometric (MS) detector. An alternative approach on the use of electronic nose is described.

The approach is simpler to the traditional GC-MS counterparts, but provides key information of the samples analysed. Using lipids and gas chromatography stationary phase materials with different polarity as sensing membrane, a quartz crystal microbalance smell sensor array has been developed for the analysis of a traditional medicinal plant. The headspace vapours of different types of Tongkat Ali (Eurycoma longifolia) extracts were analysed by the smell sensor and GC-MS. The correlation between the sensor’s response and identified compounds were studied using PCA. (Shafiqul Islam et al., 2005).

3. Correlation of sensory analysis with electronic nose data
Gutierrez-Osuna et al, (2001) had evaluated the electronic nose as an alternative to sensory analysis for assessing the effectiveness of biofilters. An aroma San@ A32S electronic nose and a human panel at Duke University's Taste and Smell Research Lab were used to measure typical volatile compounds from swine confinement buildings. Chemo metrics techniques were employed to predict the olfactory scores of the human panel from the electronic nose data. The cross-sensitivity of the sensor array to the humidity of the samples is described. The results indicate that the Enose generates responses that are correlated with sensory analysis rating of swine malodours at different concentrations.

4. Forming odour categories using an electronic nose
The ability to use linguistic concepts to describe perceptions or measure data is an emerging feature for artificial sensing systems. Odour categories need to be symbiotically represented using the data from an electronic nose. One objective is to facilitate human and computer interaction and therefore, the name given to the odours are correlated with a human user. Perceptual differences that arise between the human perception of odours and the electronic one, represents a challenge to the system. Therefore, to cope with these differences, the system maintains the freedom to evaluate how appropriately the linguistic concepts represent the sensory perceptions.

Finally, some experimental results are shown where odour categories are formed and new odours are described using these categories.

5. Quartz resonator sensor arrays
Nakomota et al (1993) successfully identified perfumes and flavours using a quartz resonator with 8 different types of resonator membrane systems initially designed to inspect the aromas. Both neutral networks and principal components analysis were used to aid data separation.

Five different perfume types were clearly separated using PCA. Nato et al (1995) showed that an epoxy-coated quartz crystal resonator sensor, used with associative neutral network analysis and principle component analysis, could discriminate between red, white and rose wine.

Nine parameters representative of the transient sensory response curve were chosen and analysed. The neural network after initial training with 5,000 known samples was able to identify each kind of wine with a 100% success rate.

Conclusion
The quality of the food supply depends at least in part on the sensory effects that foods have on consumers. Sensory research and its application have played a key role in the development of food products that are well accepted by consumers. This area of research will continue to be a vital link in modern food supply.

References

• Acree T.E. 1993. Bioassays for flavour. In: Flavour Science - Sensible Principles and Techniques, ed. T.E. acre and R. Teranishi. pp. 1-20. ACS, Washington.

• Acree T.E., Barnard J. and Cunningham D.G. 1984. A procedure for the sensory analysis of gas
chromatographic effluents. Food Chem. 14:273-286.

• Aishima T. 1991c. Discrimination of liqueur aromas by pattern recognition analysis of responses
from a gas sensory array. Anal Chim Acta. 243: 293-300.

• Aishima T. and Nahal S. 1991. Chemo metric of flavour research. Food Rev. Intl. 7(1): 33-101.

• Amoore J.E. 1970. Molecular basis of odour classification. C.C. Thosam, Spingfield, III.

• Axel R. 1995. The molecular logic of smell. Sci.Am. Oct. pp. 130-137.

• Bartlett, P.N., Elliott, J.M. and Gardner J.W. 1997. Electronic noses and their application in the food industry. Food Technol. 51:44-48.

• Breer H. 1994. Signal recognition and chemoelectrical transduction in olfaction. Biosensors Bioelectronics. 9:625-632.

• Catellan G.W. 1971. Physical chemistry, 2nd ed., Addison-Wesley Publ., Reading, Mass.

• Clapham D.E. 1996. The G-protein nanomachine. Nature. 379: 297:299.

• De Roos K. 1996. Personal communications.

• Dolf D.1996. The dynamic flavour profile method. Food Technol. 2:55-60.

• Fiorella Sinesio, Corrado D.N., Giovanni B.Q., Francesco M.B., Elisabetta M., Antonella M., Reberto P. and Amaldo D.A. 2000. Use of electronic nose and trained sensory panel in the evaluation of tomato quality. J. Sci. Food Agric. 80:63-71.

• Fischer N. and Widder S. 1997. How proteins influence food flavour. Food Technol. 51(1): 68-70.

• Fisher R.G. 1994. Personal communication. Kraft Foods, Glenview, 12.

• Fisher R.G., Toui I.C., Bodett, S.P., Leland J.V., Croasmun W.R. and White J.D. 1994. Effects of solids on equilibrium flavour releases. Presented at Ann. Mtg., Am. Chem. Soc., Washington, Aug. 21-26.

• Gardner J.W. and Bartlett. P.N. 1992. Sensors and sensory systems for an electronic nose. NATO ASI series E: Applied Sciences, Vo. 212, 161-179. Kluwer, Dordrecht the Netherlands.

• Gardner J.W. and Hines E.L. 1997. Pattern analysis techniques in Kress-Rogers. pp. 633-652.

• Grosch W. 1993. Detection of patent odourants in food by aroma extract dilution analysis. Trends in Food Science and Technology. 4:68-73.

• Grosch W.1993. Detection of patent odourants in foods by aroma extracts dilution analysis. Trends in Food Science and Technology. 4:68-73.Z

• Gutierrez-Osuna R., Schiffman S.S. and Nagle H.T. 2001. Correlation of sensory analysis with electronic nose data for swine odour remediation assessment. Proceeding of the 3rd European Congress on Odours, Metrology and Electronic noses, Paris, France, June 19-21.

• Jane V. Leland. 1997. Flavour interaction: The greater whole. Food Technol. 51(1): 75-80.

• Jun W. and Yibin Z. 2007. Electronic nose technique: Potential for monitoring maturity and shelf life of tomatoes. New Zealand J. of Agricultural Research. 50: 1219-1228.

• Kosikowski F.V. and Mocquot G. 1958. FAO Agric. Stud. 38: 133-147.

• Kress-Rogers E. 1996. Handbook of Biosensors and electronic nose, CRC Press, Boca Raton, FL.

• Land D.G. 1996. Perspectives on the effects of interactions on flavour perception: An overview In: Flavour-Food Interactions ed. R.J. Mc Gorrin and J.V. Leland pp. 2-11. ACS Symposium Series Book 633.

• Lawless H.T. and Lee C.B. 1993. Common chemical senses in food flavour. In: Flavour Sciencesensible principles and techniques, ed. T.E. Acree and R.Teranishi pp. 23-66, ACS. Washington D.C.

• Maur H.G. 1975. Binding of volatile aroma substances to nutrients and foodstuffs. Proc. Itern.
Symp. of Aroma Res. Zeist. Pudoc, Wageningin. pp. 143-157.

• Mc. Nullty P.B. 1987. Flavour release-elusive and dynamic. In: Food structure and behaviour. ed.: Blaushard J.M.V. and Lellford O. pp. 245-258. Academic London.

• Madsen G.M. and Roman D. Grypa. 2000. Spices, flavour system and the electronic nose. Food Technol. 54(3): 44-46.

• Nakamoto T., Fukuda A. and Moriizumi T. 1993.Perfume and flavour identifications by odour-sensing system using quartz resonator array and neutral network pattern recognition. Sensors and Actuators. B, 10: 85-90.

• Nanto H., Tsubakino S., Ikeda M. and Fumitaka E. 1995. Identification of aromas from wine using quartz-resonator gas sensors in conjunction with neural-network analysis. Sensor Actuators. B, 24s-25: 794-796.

• Overbosch P., Afterof W.G.M. and Haring P.G.M. 1991. Flavour Release in the mouth. Food Revs. Int. 7(2):137-184.

• Pangborn R.M. 1964. Sensory evaluation of foods: A book backward and forward. Food Technol.
18:63-65.

• Persaud K. and Dodd G.H. 1982. Analysis of themammalian olfactory system using a model nose.
Nature. 299:352-355.

• Persaud K.C. and Pelosi P. 1995. An approach to an electronic nose. Artificial Internal Organs. 31:297-300.

• Persaud K.C. and Travers P.J. 1997. Arrays of broad specificity firms for sensing volatile chemicals in Kress-Rogers.

• Philip N., Bartlett P.M., Elliott and Julian W. 1997. Electronic noses and their application in food
industry. Food Technol. 51(12):44-48.

• Reineccius G.A. 1994. The flavorist. In: Source Book of Flavours, Second edition. G.A. Reineccius, pp. 691-712. Chapman & Hall. New York.

• Schild D. 1990. Chemosensory information processing NATO ASI Series H39: Cell Biology, 39.
Springer-Verlag Berlin.

• Shafiqul Islam A.K.M., Othman A.R., Ahmad M.N. and Md. Shakaff. A.Y. 2005. Correlation studies
between the electronic nose response and headspace volatiles of Eurycoma longifolia extracts. 1st International Conference on Sensing Technology. Nov. 21-23, Palmerston North, New Zealand.

• Stone H., Sidel J., Oliver S.M., Woolsey A. and Singleton R.C. 1974. Sensory evaluation by quantitative descriptive analysis. Food Technol. 28(1): 24-34.

• Wright R.H. 1982. The sense of smell. CRC Press. Boca Raton. FL.Dr VK Joshi is Prof and Head, Mutum Preema Devi is Research Scholar and Dr Dev Raj is Scientist (Food Technology), at the Department of Post Harvest Technology, Dr YS Parmar University of Horticulture and Forestry