Extraction of Quantitative Traits from 2D Images of Mature Arabidopsis Plants (bibtex)
by Marco Augustin
Abstract:
The functional analysis of genes is a popular and interesting challenge in natural sciences. The understanding about genes causing pathologies in humans and animals or genes causing an increasing crop yield are only two important and relevant applications. High-throughput phenotyping studies are seeking to increase the understanding about the impact of the genotype of an organism on its appearance- the phenotype. To find this correlation, genetic sequenced data as well as the phenotypic characteristics, so called traits, have to be determined. The bottleneck in these large-scale studies is the manual manipulation of samples and the subsequent determination of traits. Arabidopsis thaliana is a widespread, small, flowering plant and a popular model in functional genomics. In this work a framework is presented to extract geometrical and topological traits from 2D images of mature Arabidopsis (e.g. length of a stem, number of branches). Due to logistical reasons the plants were dried and pressed before the images were acquired. Therefore some parts of the plants are overlapping and the 'realistic' architecture has to be reconstructed from the 2D images before the traits can be extracted. The reconstruction of the plants architecture is done in two steps. In the first step, a tracing approach is used for the extraction of the centerline of the plant. In the second step, continuity principles are used to group centerline segments and reconstruct the plants' realistic architecture. The need for supervision along the pipeline is tried to be brought towards a minimum. Nevertheless, methods for minor corrective interventions are provided to rise the throughput rate. The accuracy and the grade of automation during the plant reconstruction is depending on the morphological complexity of the plant structure. Unsupervised trait extraction using the framework is reserved to plants with a limited morphological complexity and images with a uniformly high contrast.
Reference:
Extraction of Quantitative Traits from 2D Images of Mature Arabidopsis Plants (Marco Augustin), Technical report, PRIP, TU Wien, 2014.
Bibtex Entry:
@TechReport{TR132,
  author =	 "Marco Augustin",
  title =	 "Extraction of Quantitative Traits from 2D Images of Mature Arabidopsis Plants",
  institution =	 "PRIP, TU Wien",
  number =	 "PRIP-TR-132",
  year =	 "2014",
  url =		 "ftp://ftp.prip.tuwien.ac.at/pub/publications/trs/tr132.pdf",
  abstract =	 "The functional analysis of genes is a popular and interesting challenge in natural sciences. The understanding about genes
causing pathologies in humans and animals or genes causing an increasing crop yield are only two important and relevant
applications. High-throughput phenotyping studies are seeking to increase the understanding about the impact of the genotype of an
organism on its appearance- the phenotype. To find this correlation, genetic sequenced data as well as the phenotypic
characteristics, so called traits, have to be determined. The bottleneck in these large-scale studies is the manual manipulation
of samples and the subsequent determination of traits. Arabidopsis thaliana is a widespread, small, flowering plant and a popular model in functional genomics. In this work a framework is presented to extract geometrical and topological traits from 2D images of mature Arabidopsis (e.g. length of
a stem, number of branches). Due to logistical reasons the plants were dried and pressed before the images were acquired.
Therefore some parts of the plants are overlapping and the 'realistic' architecture has to be reconstructed from the 2D images
before the traits can be extracted. The reconstruction of the plants architecture is done in two steps. In the first step, a
tracing approach is used for the extraction of the centerline of the plant. In the second step, continuity principles are used
to group centerline segments and reconstruct the plants' realistic architecture. The need for supervision along the pipeline
is tried to be brought towards a minimum. Nevertheless, methods for minor corrective interventions are provided to rise the
throughput rate. The accuracy and the grade of automation during the plant reconstruction is depending on the morphological complexity of
the plant structure. Unsupervised trait extraction using the framework is reserved to plants with a limited morphological
complexity and images with a uniformly high contrast.",
}
Powered by bibtexbrowser