A new pollen detection method based on an electronic nose

It is shown how an electronic nose-a gas sensor array combined with a routine for pattern recognition—together with a pyrolysing unit can be used to distinguish pollen from other airborne particles. The samples investigated were three types of pollen, diesel and petrol soot, dust from windows adjacent to traffic and ordinary soil. Samples were heated in a specially designed furnace to 250°C and the gases evolved were analysed with an electronic nose. The data obtained were treated with principal component analysis (PCA) and with an artificial neural net (ANN). The PCA showed clear groupings between the various classes, but not so clear within the pollen group. The use of an ANN improved the results, leading to clear separations between the classes and partly also within the pollen class.