A statistical approach for the analysis of Proton Transfer Reaction Mass Spectrometry (PTR-MS) data aimed at a qualification of fruits based on VOC emission

Due to the high degree of sensitivity and the large amounts of data obtained by Proton Transfer Reaction Mass Spectrometry (PTR-MS), an appropriate method of data handling, analysis and interpretation is required. The experiments that we have carried out show a strong dependence of the measured Volatile Organic Compound (VOC) emissions from a large number of parameters that characterize a fruit. We have studied the preservation of Malus domestica ‘Red Delicious’ under Initial Low Oxygen Stress (ILOS) treatment and we have considered as co-variant parameters the weight and skin attributes such as hue, saturation and lightness. We have developed an approach based on a causal network model with respect to sets of parameters prior, intervening and consequent respectively. Through the statistical analysis of VOC emissions measured by PTR-MS, we have identified the masses of the VOC that best correlate to the different properties of the apples over the whole period of preservation and for shelf-life of about two weeks. It is very interesting to notice that such an analysis followed by a Relative Principal Component Analysis (R-PCA) allows a very nice discrimination of the apples and of the effects of the treatments on preservation and shelf life. We discuss the great potential of this method to identify specific VOC sets that could be used as true markers for different apples attributes and treatments.