gas sensor array for blueberry fruit disease detection and classification
电子鼻技术用于蓝莓病害检测和分类changying li a,∗, gerard w. krewerb, pingsheng ji c, harald schermd, stanley j. kayse
a university of georgia, department of biological and agricultural engineering, 2329 rainwater road, tifton, ga 31794, usa
b university of georgia, department of horticulture, 4604 research way, tifton, ga 31794, usa
c university of georgia, department of plant pathology, 115 coastal way, tifton, ga 31794, usa
d university of georgia, department of plant pathology, 2311 miller plant sciences bldg., athens, ga 30602, usa
e university of georgia, department of horticulture, 1111 miller plant sciences bldg., athens, ga 30602, usa
a b s t r a c t
a conducting polymer gas sensor array (electronic nose cyranose 320) was evaluated for detecting and classifying three common postharvest diseases of blueberry fruit: gray mold caused by botrytis cinerea, anthracnose caused by colletotrichum gloeosporioides, and alternaria rot caused by alternaria sp. samples of ripe rabbiteye blueberries (vaccinium virgatum cv. brightwell) were inoculated individually with one of the three pathogens or left non-inoculated, and volatiles emanating from the fruit were assessed using the gas sensor array 6–10 d after inoculation in two separate experiments. principal component analysis of volatile profiles revealed four distinct groups corresponding to the four inoculation treatments. manova, conducted on profiles from individual assessment days or from combined data, confirmed that the four treatments were significantly different (p < 0.0001). a hierarchical cluster analysis indicated two super-clusters, i.e., control cluster (non-inoculated fruit) vs. pathogen cluster (inoculated fruit). within the pathogen cluster, fruit infected by b. cinerea and alternaria sp. were more similar to each other than to fruit infected by c. gloeosporioides. a linear bayesian classifier achieved 90% overall correct classification for data from experiment 1. tenaxtm trapping of volatiles with short-path thermal desorption and quantification by gas chromatography–mass spectrometry was used to characterize volatile compounds emanated from the four groups of berries. six compounds [styrene, 1-methyl-2-(1-methylethyl) benzene, eucalyptol, undecane, 5-methyl-2-(1-methylethyl)-2-cyclohexen-1-one, and thujopsene] were identified as contributing most in distinguishing differences in the volatiles emanating from the fruit due to infection. a canonical discriminant analysis model using the relative concentration of each of these compounds was developed and successfully classified the four categories of berries. this study underscores the potential feasibility of using a gas sensor array for blueberry postharvest quality assessment and fungal disease detection.
利用导电聚合物气体传感器阵列(电子鼻cyranose320)对蓝莓果实采后常见的三种病害进行了检测和分类：灰霉病、炭疽病、交链孢霉引起的炭疽病、成熟的兔眼蓝莓(牛痘)样品的交链孢腐烂病。um virgatum。brightwell)分别接种三种病原体中的一种或不接种，在两个单独的实验中，接种后6-10 d用气体传感器阵列评估从水果中释放的挥发物。挥发性成分分析显示四个不同的组分与四种接种处理相对应。manova对个体评估日或综合数据的资料进行分析，证实四种治疗方法有显著差异(p<0.0001)。层次聚类分析显示两个超级聚类，即控制聚类(未接种的果实)与病原聚类(接种的果实)。在病原菌群中，灰霉病和交链孢杆菌侵染的果实比灰霉病侵染的果实更为相似。线性贝叶斯分类器对实验1的数据实现了90%的整体正确分类。采用短程热解吸和气相色谱-质谱定量法对四组浆果挥发物的tenaxtm捕集进行了表征。6种化合物[苯乙烯、1-甲基-2-(1-甲基乙基)苯、桉树醇、十一烷、5-甲基-2-(1-甲基乙基)-和thujopsene]被鉴定为助于区分由感染引起的水果挥发物差异。建立了一个基于上述化合物相对浓度的典型判别分析模型，并成功地对四类浆果进行了分类。本研究强调了使用电子鼻进行蓝莓采后质量评估和真菌病检测的潜在可行性。