Thesis
Digital RGB and hyperspectral imaging for quantitative evaluation of aphanomyces root rot disease resistance in lentil
Washington State University
Master of Science (MS), Washington State University
2018
Handle:
https://hdl.handle.net/2376/102427
Abstract
Aphanomyces root rot, caused by Aphanomyces euteiches Drechs, is a severe soil-borne disease affecting legumes worldwide. The development of resistant cultivars is one of the best strategies to control this pathogen. However, the evaluation of disease severity is limited to visual scores that are subjectively assigned. High-throughput phenotyping, in contrast, can provide an objective and accurate measurement of crop traits, which can improve the plant selection process. In this research, we developed a digital Red-Green-Blue image-phenotyping pipeline to quantify the severity of Aphanomyces root rot disease in 350 lentil (Lens culinaris Medik) accessions and 200 lines from a lentil recombinant inbred line population in 2017 and 2018, respectively. We employed hyperspectral imaging system (550-1700 nm) to detect severity classes in 79 and 21 lines of lentils from 2017 and 2018, respectively. Geometric features extracted from RGB images were strongly correlated with shoot and root dry biomass with correlation coefficients (r) of 0.90 and 0.89, respectively (P < 0.0001). Color and texture features showed significant correlation with visual disease scores (0.22 < |r| < 0.89, P < 0.0001) among both datasets. Features extracted from shoot and root images were employed to evaluate the performance of support vector machine models in classifying resistant, partially resistant, and susceptible classes and resulted in an accuracy of 85% in 2017 data, whereas features extracted from root section only resulted in an accuracy of 74% in 2018 data. Hierarchical clustering analysis of inoculated root spectra--extracted from hyperspectral images--revealed three distinct groups that did not reflect the level of disease severity based on visual scores. A total of 29 normalized difference spectral indices (NDSI) were selected as best ratios reflecting visual scores (0.43 ≤ |r| ≤0.51 and P < 0.0001). Classification results using these NDSIs showed higher accuracies in comparison with using all wavelengths spectra or their principal components. This model was able to classify inoculated roots into three classes with an accuracy of 67% for 2018 data, whereas this same approach failed in capturing the differences between these classes for 2017 data.
Metrics
21 File views/ downloads
53 Record Views
Details
- Title
- Digital RGB and hyperspectral imaging for quantitative evaluation of aphanomyces root rot disease resistance in lentil
- Creators
- Afef Marzougui
- Contributors
- Sindhuja Sankaran (Degree Supervisor)
- Awarding Institution
- Washington State University
- Academic Unit
- Biological Systems Engineering, Department of
- Theses and Dissertations
- Master of Science (MS), Washington State University
- Publisher
- Washington State University; [Pullman, Washington] :
- Identifiers
- 99900525124101842
- Language
- English
- Resource Type
- Thesis