Time Course Lesion Tracking for Maize Phenotype Images

The hypersensitive reaction (HR) is one of the most important immune responses in the plant kingdom, and yet it is not fully understood in plants.  Researchers interested in studying this process oftentimes utilize lesion mimic mutants, in which the response is activated in the absence of disease, or other mutants that are known to affect HR.  

While researchers have examined the HR collapse of host cells with regard to a number of conditions and have identified a number of known variables that affect lesion formation, there have not been computational methods developed to facilitate in-depth quantitative study of the temporal behavior of these phenotypes.  As the temporal characteristics of these phenotypes are likely influenced by factors that underlie HR, the development of such an automated method that can generate temporal trait data has the potential to help researchers gain further insight into genetic pathways and networks that trigger and execute the HR response.

 In this poster, we present a novel method, called “2-Tier Floating Approach to Variable Object Pattern Matching” for processing time-course studies of maize leaves that allows tracking of necrotic lesions across frames and models the temporal behavior of those tracked lesions.   This approach is empirically proven to be effective and accurate to handle challenges, such as translation/rotation of leaves and lesion growth differences between images, as well as waviness and distortion of leaves over time.

This project is supported in part by the National Science Foundation DBI-0447794 and DBI-1053024.