Probabilistic classification of disease symptoms caused by salmonella on arabidopsis plants

Abstract

Several reports have linked food poisoning with the consumption of raw vegetables and fruits contaminated by Salmonella. Most studies suggested an extracellular lifestyle of Salmonella on plants. However, more recent studies show that Salmonella are also able to colonize the intracellular compartment of various plant tissues causing chlorosis and eventually death of infected organs. The aim of this work is to present a probabilistic classification algorithm for disease symptoms on Arabidopsis thaliana plant in order to improve the current biological research. The algorithm itself uses images of Arabidopsis thaliana leaves as input and consists of two steps. The first step is the detection of pixels belonging to a leaf. This is done with a globally optimal color segmentation method. The second step is realized with a probabilistic framework to classify each pixel. Finally a morbidity rate is computed based on the classification result.