Skip to main content
ARS Home » Southeast Area » Fort Pierce, Florida » U.S. Horticultural Research Laboratory » Subtropical Plant Pathology Research » Research » Publications at this Location » Publication #187249
Title: REPEATABILITY AND COMPARISION OF IMAGE ANALYSIS AND VISUAL ASSESSMENT FOR DISEASE ASSESSMENT OF CITRUS CANKER

Author
item BOCK, C. H. - UNIV. OF FLORIDA
item PARKER, P. E. - USDA-APHIS
item COOK, A. Z. - USDA-APHIS
item Gottwald, Timothy

Submitted to: Phytopathology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/1/2006
Publication Date: 7/1/2006
Citation: Bock, C., Parker, P., Cook, A., Gottwald, T.R. 2006. Repeatability and comparision of image analysis and visual assessment for disease assessment of citrus canker. Phytopathology. 96(6): (Supplement) S13-14.

Interpretive Summary: Citrus canker is caused by the bacterial pathogen Xanthomonas axonopodis pv citri (Xac). The disease is of concern in wet tropical and subtropical citrus growing regions as infection results in yield loss and severely blemished fruit unsuitable for the fresh market. Ways of managing the disease are being sought, and accurate, precise and reliable disease assessment is needed for monitoring epidemics. Visual assessment (VA) is the only reliable means estimating disease. The objective of this study was to compare image analysis (IA) to VA for assessing symptoms of citrus canker. Digital images of citrus leaves with a range of incidence and severity of citrus canker were assessed by three plant pathologists (VA1-3) and by image analysis (IA), and the results analyzed and compared. All four assessments showed only a moderate relationship between lesion number and area infected, suggesting appropriate assessments types should be chosen to suit the objectives of the experiment. There was no differences between IA and VA for lesion number although there was some difference in severity which may be due to VA being less capable at differentiating necrotic areas but being better at assessing the whole infected area. IA appears to provide a reliable way to assess canker infected leaves for disease, but symptom characters (heterogeneity, coalescence of lesions) combined with assessors loss of concentration, can lead to discrepancy in results.

Technical Abstract: Citrus canker, a disease of several citrus species, is caused by the bacterial pathogen Xanthomonas axonopodis pv citri (Xac). The disease is of concern in several wet tropical and subtropical citrus growing regions as infection results in yield loss and severely blemished fruit unsuitable for the fresh market (4). Ways of managing the disease are being sought, and accurate, precise and reliable disease assessment is needed for monitoring epidemics. Visual assessment (VA) is the only reliable means estimating disease. The objective of this study was to compare image analysis (IA) to VA for assessing symptoms of citrus canker (1,3). Digital images of 214 citrus leaves with a range of incidence and severity of citrus canker were assessed by three plant pathologists (VA1-3) and by image analysis (IA), and the results analyzed using general linear modeling (GLM) and regression analysis. All four assessments showed only a moderate relationship between lesion number and % area infected (R2=0.55-0.68, Figure 1), suggesting that appropriate assessments (lesion counts or % area infected) should be chosen to suit the objectives of the experiment. GLM analysis showed no significant differences between IA and VA for lesion number (mean 16.3-18.7 lesions per leaf), although there were significant differences in severity using % area necrotic (3.3-5.4%), but not % area chlorotic+necrotic (8.9-11.0% mean area). This may be due to VA being less capable at differentiating necrotic areas but being better at assessing the whole infected area. Using IA as a standard, variance of the VA assessment increased as lesion number increased and as severity increased to 50% area (Figure 2), agreeing with previous observations on perception of disease severity (2): the more disease there is (at least up to 50%), the less accurate the assessment. A sample of 53 leaf images was converted to black/white (b/w) and IA/VA performed. Both image sets were assessed for incidence and severity with comparable accuracy by VA1-3. However, variance tended to be greater for color images suggesting that the greater degree of subjectivity in identifying and delimiting disease where color is involved (although difference between image type may also be due in part to the b/w images standardized by IA). IA appears to provide a reliable way to assess canker infected leaves for disease, but symptom characters (heterogeneity, coalescence of lesions) combined with assessors loss of concentration, can lead to discrepancy in results.