A NON-INVASIVE TECHNIQUE FOR DETECTING SEED VIABILITY BY USING ELECTRONIC NOSE
Xin, X., Yin, G.K., Zhu, L.Y., He, J.J. and Lu, X.X.
National Crop Genebank, Institute of Crop Sciences, Chinese Academy of Agricultural Sciences, Beijing, China;
Contact: Xin Xia, email@example.com
Seed viability test is one of the routine work at genebank. Germination and TTC tests were classical methods to detect seed viability, however, both methods consume many time and seeds. Our study aimed to develop a new technique that can detect seed viability in a non-invasive and quick manner. Many reports have demonstrated that seeds volatile components changes with reduction in viability. In this study, we tried to draw the smell fingerprint of seeds volatiles by using electronic nose technique. Wheat (Triticum aestivum L.), soybean (Glycine max L.), and oilseed rape (Brassica napus L.) seeds were used as models to represent starch, protein, and oil seeds, respectively. Each crop had three or six varieties, and each variety had four or five different germination rates. Seeds were sealed in vials, and the volatiles signals were collected using a PEN3 electronic nose with oxide sensors array. The results showed that PEN3 electronic nose technique can separate seeds of different viabilities using PCA or LDA. Using BP neural network analysis, PEN3 electronic nose technique could predict viability of unknown seed samples. The BP neural network training accuracy were higher than 96%，meanwhile the predicted accuracy was over 98%，indicating that this method is reliable. The process of this technique was without any treatment, indicating this technique is totally non-invasive. In addition, this technique could save a lot of time and labor, as it only needs about 1 minute to collect the volatile signals. Therefore, PEN3 electronic nose technique could be a non-invasive, reliable, quick and low-cost technique for seed viability detection.