Onion is the second most economically important commercial vegetable crop for fresh market in the United States. It is estimated that approximately 6.2 billion pounds (2.8 million tons) of onions are produced each year in the U.S. Across the world, average annual onion consumption per person is estimated to be over 13 lb (6 Kg). Onions like any other vegetables are threatened by various bacterial or fungal diseases.
US Scientists have investigated a method of detection of sour skin caused by the bacteria Burkholderia cepacia that is one of the most important post-harvest bacterial disease in onions.
The general objective of the study was to test the automated customized electronic nose system in detecting the presence of sour skin disease in onions; while the specific objectives were to:
1. compare three baseline correction methods and three features for data pre-processing;
2. conduct principal component analysis (PCA) and develop classification models to distinguish healthy and sour skin infected onions;
3. select the best combination of metal oxide semiconductor (MOS) gas sensors from the seven available sensors (TGS 813, TGS 822, TGS 825, TGS 826, TGS 2620, SB 11A, SB AQ8).
The sensor array consists of seven metal oxide semiconductor gas sensors and a microcontroller-based automatic data logging system. Three features (relative response, area, and slope) were extracted from the sensor signal and three baseline correction methods were employed to correct the sensors' responses. The gas sensor array was tested in two separate experiments with two treatments (control and sour skin). The multivariate data analysis revealed that the ''relative responsè' feature combined with relative baseline correction method provided the best discrimination of infected onions among healthy ones.
Scientists conclude that this study proved the efficacy of using a customized gas sensor array to detect sour skin infected onions among healthy onions. The relative response feature combined with relative baseline correction method performed the best among the nine feature-baseline correction combinations. The sensor responses showed significant difference between the volatiles released by control onions and sour skin diseased onions starting from 4 to 7 days after inoculation. TGS 826 and SB-AQ8 contributed the most in detecting the diseased onions whereas TGS 813 and TGS 2620 contributed the least. When all the seven MOS sensors were used, a classification accuracy of 85% (in validation) was achieved by using support vector machine. It was possible to achieve comparable results by removing the least one or two contributing sensors.
The tested customized gas sensor array shows great potential to be used as an automated detection tool for onion postharvest diseases in storage.
