A. Amari

Application of a portable electronic nose device to discriminate and identify cheeses with known percentages of cow's and goat's milk

A portable electronic nose comprising an array of 6 metal oxide semiconductor sensors was developed and used, jointly with pattern recognition methods, to discriminate and identify several cheeses made from goats, cows milk and their mixtures. Principal Component Analysis (PCA) was used to visualize the different categories of aroma profiles and Multivariate Analysis of Variance (MANOVA) was performed to test the significance of the differences between cheeses groups. Database was then elaborated using supervised classifiers such as Discriminant Factor Analysis (DFA) with leave one out approach. The results indicate that the portable electronic nose device can clearly and rapidly distinguish between cows milk cheese, goats milk cheese and cheeses containing variable amounts of cows and goats milk. So, this system can be used in order to avoid fraud and to fulfill customer expectations.

Data Fusion from Voltammetric and Potentiometric Sensors to Build a Hybrid Electronic Tongue Applied in Classification of Beers

A hybrid electronic tongue based on data fusion of two different sensor families was built and used to recognize three types of beer. The employed sensor array was formed by three modified graphite‐epoxy voltammetric sensors plus six potentiometric sensors with cross‐sensitivity. The sensors array coupled with feature extraction and pattern recognition methods, namely Principal Component Analysis (PCA) and Discriminant Factor Analysis (DFA), were trained to classify the data clusters related to different beer types. PCA was used to visualize the different categories of taste profiles and DFA with leave‐one‐out cross validation approach permitted the qualitative classification. According to the DFA model, 96% of beer samples were correctly classified. The aim of this work is to prove performance of hybrid electronic tongue systems by exploiting the new approach of data fusion of different sensor families, in comparison of electronic tongue with only one sensor type.

Geographical classification of Virgin Olive Oils by combining the electronic nose and tongue

Although the great interest of development of performed gas and liquid sensors, lack of cross-sensitivity still remains the major drawback of electronic sensing systems such as electronic nose and tongue. We propose here an approach aimed at overcoming this shortcoming. So a performed data fusion method of electronic nose and tongue was used in order to classify five Virgin Olive Oils (VOOs) picked up from five Moroccan geographical areas. The electronic nose instrument consists of five commercial available MOS TGS gas sensors and the electronic tongue was designed using four voltammetric electrodes. Two techniques, i.e., Principal Component Analysis (PCA) and Support Vector Machines (SVMs) were used to develop classification models using as inputs specific features extracted from the collected sensor signals. Great enhancement in successful discrimination between all VOOs was achieved when compared to the individual systems due to a performed low-level of abstraction data fusion.

Potential application of the electronic nose for shelf‐life determination of raw milk and red meat

The present study describes the performance of an electronic nose in food odor analysis. This methodology was successfully applied to odor characterization of milk stored at 4° C during 4 days and of beef and sheep meat stored at 4° C for up to 15 days. The electronic nose sensor system coupled to PCA as a pattern recognition technique, is able to reveal characteristic changes in raw milk and red meat quality related to storage time. Additionally, a bacteriological method was selected as the reference method to consistently train the electronic nose system for both beef and sheep meat analysis.