Image-based quantification of Fusarium wilt severity in banana Ryan Orr 1 & Anthony Pattison 2 & David East 2 & Noeleen Warman 3 & Wayne O’Neill 3 & Elizabeth Czislowski 4 & Paul N. Nelson 1 Received: 4 December 2018 /Accepted: 1 May 2019 # Australasian Plant Pathology Society Inc. 2019 Abstract The severity of Fusarium wilt of bananas has long been classified based on visual assessment of necrosis in rhizome or pseudostem cross-sections. The improved method proposed here uses digital image analysis to quantify the proportion of rhizome tissue that is necrotic. It agrees well with visual classification, but provides greater reproducibility, precision and statistical power. Keywords Fusarium oxysporum f. sp. cubense . Panama disease . Disease severity . Image analysis . Method Fusarium wilt of banana, or Panama disease, is rapidly spread- ing throughout the world. As the impact of this disease in- creases, so too does the need to quantify disease severity in research projects assessing prevention and treatment options. The method proposed here has been designed for Fusarium wilt of banana, but it could also be used to assess Fusarium wilt of other crops, or other plant diseases resulting in tissue discoloration. Existing methods for quantifying Fusarium wilt severity visually categorize cross sections based on the level of vascu- lar tissue discoloration of rhizome and sometimes pseudostem or root tissue (Carlier et al. 2003; Orjeda 1998; Peng et al. 1999; Smith et al. 2008; Viljoen et al. 2016; Zadoks and Schein 1979; Zuo et al. 2018). Fusarium oxysporum f. sp. cubense (Foc), the causal organism of Fusarium wilt, initially infects the roots of bananas, moves through the rhizome, caus- ing rhizome necrosis and discoloration, before moving into the pseudostem and leaves (Warman and Aitken 2018). Therefore, rhizome tissue colour or necrosis is a suitable proxy for disease severity (Ploetz 2015), however the causa- tive organism should still be confirmed using traditional methods (Puhalla 1985). The method presented here has two advantages over current methods: 1) precision and thus statis- tical power are increased as the variable measured is continu- ous rather than discrete 2) subjectivity and human error are reduced by using computerised image analysis rather than the human eye to determine the proportion of tissue that is discolored. Existing methods typically use a 6–class scale with the first class for no disease presence. Each rhizome cross section is scored and then scores are averaged across all sections from an individual plant. The classification system from Viljoen et al. (2016) is: 1 = No symptoms; 2 = Few internal spots; 3 = <1/3 Discoloured tissue; 4 = 1/ 3–2/3 Discoloured tissue; 5 = >2/3 Discoloured tissue; 6 = Entire inner rhizome discoloured. Thus, samples with slightly different discoloration can have severity scores differing by 33% and samples with the same score may differ by up to 32%. By using a continuous scale that has high reproducibility the precision can be improved. The method proposed here quantifies discoloration of the central region of vascular tissue in the rhizome as follows. Three transverse sections of the rhizome are taken, one quarter, half, and three quarters up, after the removal of all soil and roots. A photograph is taken of the sections, including a gray scale reference, and * Ryan Orr ryan.orr@jcu.edu.au 1 College of Science and Engineering, James Cook University, Cairns, Qld 4878, Australia 2 QLD Department of Agriculture and Fisheries, South Johnstone, Qld 4859, Australia 3 QLD Department of Agriculture and Fisheries, Dutton Park, Qld 4102, Australia 4 School of Agriculture and Food Sciences, The University of Queensland, St. Lucia, Qld 4072, Australia Australasian Plant Disease Notes (2019) 14:14 https://doi.org/10.1007/s13314-019-0344-7