Published 2009-12-01
Abstract
Mature-green ‘Booth 7’ avocados were stored at 20 °C and assessed for electronic nose (Cyranose® 320) analysis at three maturity stages: mature-green, mid-ripe, and ripe. The training session consisted of six replicates of one fruit for each maturity stage. The pulp (100 g of cubes; 1 cm3) was immediately placed in a 1.7-L glass jar, and sealed for 5 min prior to headspace sampling. The sequence of conditions for each sample reading was: 60 s baseline purge, 60 s sample draw, 30 s sample gas purge, 150 s air intake purge. The identification session was performed using seven ripe fruit under the same set of conditions. The results showed that the electronic nose was successfully trained. Cross-validation of the model was 100%, suitable for class discrimination. Canonical discriminant analysis separated the pulp into three clusters according to the maturity stage. Interclass M-distance was lower between mid-ripe and ripe fruit (15.32), and much higher between mature-green and mid-ripe (55.74) or mature-green and ripe (68.84). However, the electronic nose had poor performance on the identification of test samples (43%). Changes in methodology to improve sample identification are discussed.