Caio Bruno Wetterich, Ruan Felipe de Oliveira Neves, José Belasque, Reza Ehsani, and Luis Gustavo Marcassa, "Detection of Huanglongbing in Florida using fluorescence imaging spectroscopy and machine-learning methods," Appl. Opt. 56, 15-23 (2017)
In this study, we combine a fluorescence imaging technique and two machine-learning methods to discriminate Huanglongbing (HLB) disease from zinc-deficiency stress on samples from Florida, USA. Two classification methods, support vector machine (SVM) and artificial neural network (ANN), are used. Our classification results present high accuracy for both classification methods: 92.8% for SVM and 92.2% for ANN. The results from Florida are also compared to results from São Paulo State, Brazil. This comparison indicates that the present technique can be applied to discriminate HLB from zinc deficiency in both states.
You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Figure files are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Article tables are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Equations are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
Sensitivity, Specificity, and Accuracy Parameters of the SVM Classifier
Excitation at 405 nm
Excitation at 470 nm
F560
F580
F690
F530
F550
F690
Sensitivity
82.4%
98.0%
7.8%
81.4%
84.3%
32.4%
Specificity
37.2%
7.7%
94.9%
55.1%
50.0%
59.0%
Accuracy
62.8%
58.9%
45.6%
70.0%
69.4%
43.9%
Table 4.
Sensitivity, Specificity, and Accuracy Parameters Considering the Combination of the Fluorescence Bands to Discriminate HLB from Zinc Deficiency Using the SVM Classifier
Excitation at 405 nm
Excitation at 470 nm
Combination of the Two Excitations
All Fluorescence Bands
Sensitivity
92.2%
88.2%
88.2%
89.2%
90.2%
Specificity
38.5%
96.2%
59.0%
87.2%
98.7%
Accuracy
68.9%
91.7%
75.6%
88.3%
92.8%
Table 5.
Classification Results Comparing HLB Disease and Zinc-Deficiency Stress
Condition Determined by qPCR Test
Condition Positive
Condition Negative
Excitation at 405 nm
F560
Test outcome positive
Test outcome negative
F580
Test outcome positive
Test outcome negative
F690
Test outcome positive
Test outcome negative
Excitation at 470 nm
F530
Test outcome positive
Test outcome negative
F550
Test outcome positive
Test outcome negative
F690
Test outcome positive
Test outcome negative
Table 6.
ANN Classifier Sensitivity, Specificity, and Accuracy Parameters
Excitation at 405 nm
Excitation at 470 nm
F560
F580
F690
F530
F550
F690
Sensitivity
72.6%
96.1%
43.1%
67.7%
74.5%
73.5%
Specificity
56.4%
15.4%
69.2%
76.9%
71.8%
51.3%
Accuracy
65.6%
61.1%
54.4%
71.7%
73.3%
63.9%
Table 7.
Sensitivity, Specificity, and Accuracy Parameters Considering the Combination of Fluorescence Bands to Discriminate HLB from Zinc Deficiency Using the ANN Classifier
Excitation at 405 nm
Excitation at 470 nm
Combination of the Two Excitations
All Fluorescence Bands
Sensitivity
88.2%
87.3%
82.4%
89.2%
90.2%
Specificity
55.1%
89.7%
82.1%
89.7%
97.4%
Accuracy
73.9%
88.3%
82.2%
89.4%
92.2%
Table 8.
Performance of the Classifiers for Florida Samples according to the Combinations of the Spectral Bands from Brazil
SVM
ANN
Excitation at 405 nm
Excitation at 470 nm
Excitation at 405 nm
Excitation at 470 nm
Sensitivity
87.3%
86.3%
87.3%
88.2%
Specificity
94.9%
80.8%
73.1%
82.1%
Accuracy
90.6%
83.9%
81.1%
85.6%
Tables (8)
Table 1.
Fluorescence Bands Selected by the FIS-1
Diseases
Excitation 405 nm
Excitation 470 nm
HLB and zinc deficiency
Filters at 560, 580, and 690 nm
Filters at 530, 550, and 690 nm
Table 2.
Classification Results Comparing HLB Disease and Zinc-Deficiency Stressa
Sensitivity, Specificity, and Accuracy Parameters of the SVM Classifier
Excitation at 405 nm
Excitation at 470 nm
F560
F580
F690
F530
F550
F690
Sensitivity
82.4%
98.0%
7.8%
81.4%
84.3%
32.4%
Specificity
37.2%
7.7%
94.9%
55.1%
50.0%
59.0%
Accuracy
62.8%
58.9%
45.6%
70.0%
69.4%
43.9%
Table 4.
Sensitivity, Specificity, and Accuracy Parameters Considering the Combination of the Fluorescence Bands to Discriminate HLB from Zinc Deficiency Using the SVM Classifier
Excitation at 405 nm
Excitation at 470 nm
Combination of the Two Excitations
All Fluorescence Bands
Sensitivity
92.2%
88.2%
88.2%
89.2%
90.2%
Specificity
38.5%
96.2%
59.0%
87.2%
98.7%
Accuracy
68.9%
91.7%
75.6%
88.3%
92.8%
Table 5.
Classification Results Comparing HLB Disease and Zinc-Deficiency Stress
Condition Determined by qPCR Test
Condition Positive
Condition Negative
Excitation at 405 nm
F560
Test outcome positive
Test outcome negative
F580
Test outcome positive
Test outcome negative
F690
Test outcome positive
Test outcome negative
Excitation at 470 nm
F530
Test outcome positive
Test outcome negative
F550
Test outcome positive
Test outcome negative
F690
Test outcome positive
Test outcome negative
Table 6.
ANN Classifier Sensitivity, Specificity, and Accuracy Parameters
Excitation at 405 nm
Excitation at 470 nm
F560
F580
F690
F530
F550
F690
Sensitivity
72.6%
96.1%
43.1%
67.7%
74.5%
73.5%
Specificity
56.4%
15.4%
69.2%
76.9%
71.8%
51.3%
Accuracy
65.6%
61.1%
54.4%
71.7%
73.3%
63.9%
Table 7.
Sensitivity, Specificity, and Accuracy Parameters Considering the Combination of Fluorescence Bands to Discriminate HLB from Zinc Deficiency Using the ANN Classifier
Excitation at 405 nm
Excitation at 470 nm
Combination of the Two Excitations
All Fluorescence Bands
Sensitivity
88.2%
87.3%
82.4%
89.2%
90.2%
Specificity
55.1%
89.7%
82.1%
89.7%
97.4%
Accuracy
73.9%
88.3%
82.2%
89.4%
92.2%
Table 8.
Performance of the Classifiers for Florida Samples according to the Combinations of the Spectral Bands from Brazil