Multivariate approach to reveal relationships between sensory perception of cheeses and aroma profile obtained with different extraction methods

A new and original statistical approach was used to compare the effectiveness of 4 different methods to analyse aroma compounds of seven different commercial semi-hard cheeses with regard to their orthonasal sensory perception. Four extraction methods were evaluated: Purge and Trap, Artificial Mouth, Solid-Phase Microextraction (SPME) and Solvent-Assisted Flavour Evaporation (SAFE). Among the headspace methods, Artificial Mouth gave the closest representation of the studied product space to the sensory perception one. The SAFE method was complementary to the dynamic headspace methods, as it was very efficient in extracting the heavy molecules but less efficient for extracting the most volatile compounds. SPME and Purge and Trap gave intermediate representations. At the product level, results indicate that carboxylic acids are present in similar amounts in the different commercial marks of semi-hard cheeses, that esters and aldehydes vary in small proportions and that the proportions of sulphur compounds, alcohols and ketones mainly explain the sensory differences between the commercial marks. •Similarity between sensory and physico-chemical data was evaluated by RV coefficient.•Headspace extraction methods are more reconcilable with sensory perception than SAFE.•Artificial mouth allows the best sensory-like discrimination of semi-hard cheeses.•Sulphur compounds, alcohols and ketones explain sensory differences between cheeses.