Characterisation of tea leaves according to their total mineral content by means of probabilistic neural networks

The concentrations of aluminium, barium, calcium, copper, iron, magnesium, manganese, nickel, phosphorus, potassium, sodium, strontium, sulphur and zinc in white, green, black, Oolong and Pu-erh teas have been determined by inductively coupled plasma atomic emission spectrometry (ICP-AES). Samples were microwave-digested and the performance characteristics of the method were verified by analysing a certified reference material. The measured elemental concentrations in tea leaves were used to differentiate the five tea varieties. Non-parametric analysis was applied to highlight significant differences between types, and pattern recognition methods were used to characterise samples. For this aim, linear discriminant analysis (LDA) and probabilistic neural networks (PNN) were used to construct classification models with an overall classification performance of 81% and 97%, respectively. (C) 2010 Elsevier Ltd. All rights reserved.