3D image analysis of plant roots for genotype-phenotype associations

Abstract

The architecture of a plant root system (phenotype) can indicate its tolerance to acid soils and capacity to absorb nutrients, such as water and phosphorus. These properties are related to certain genes (genotype) and the identification of the candidate genes by the phenotype analysis can improve the genetics of the plants. In this sense, plant root imaging systems and shape analysis of plant root architectures have been actively investigated in the last years. This project aims to discover the traits (shape features) in 3D images of plant roots that best discriminate genotypes. Our emphasis will be on varieties of rice and the work will be developed in cooperation with Prof. Leon Kochian (Center for Agriculture and Health & Department of Plant Biology, Cornell University) and his associates (biologists, statisticians, computer scientists, geneticists, etc). This group of researchers has been developing a system for 3D image acquisition, reconstruction, and analysis of plant roots in view of improving the most popular crops in the world, such as maize, rice, sorghum, and wheat. In this project, I will be studying and developing techniques for multiscale skeletonization of 3D plant root images, methods for feature extraction from the 3D skeletons, and optimization and pattern recognition algorithms to select shape features that best discriminate genotypes. The project will be executed in one year at the laboratory of Prof. Leon Kochian, from August 1st 2011 to July 31st 2012, and represents the beginning of a long term cooperation among Cornell University, UNICAMP, and EMBRAPA, with benefits to the Brazilian Agriculture. (AU)