Generalized box-plot for root growth ensembles

Information

• Publication Type: Journal Paper with Conference Talk
• Workgroup(s)/Project(s):
  • Visual Computing: Illustrative Visualization
• Date: 2016
• Journal: BMC Bioinformatics
• Lecturer: Viktor Vad

Abstract

Background In the field of root biology there has been a remarkable progress in root phenotyping, which is the efficient acquisition and quantitative description of root morphology. What is currently missing are means to efficiently explore, exchange and present the massive amount of acquired, and often time dependent root phenotypes. Results In this work, we present visual summaries of root ensembles by aggregating root images with identical genetic characteristics. We use the generalized box plot concept with a new formulation of data depth. In addition to spatial distributions, we created a visual representation to encode temporal distributions associated with the development of root individuals. Conclusions The new formulation of data depth allows for much faster implementation close to interactive frame rates. This allows us to present the statistics from bootstrapping that characterize the root sample set quality. As a positive side effect of the new data-depth formulation we are able to define the geometric median for the curve ensemble, which was well received by the domain experts.

Additional Files and Images

Weblinks

• https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5333180/

BibTeX

@article{vad-2016-bre,
  title =      "Generalized box-plot for root growth ensembles",
  author =     "Viktor Vad and Douglas Cedrim and Wolfgang Busch and Peter
               Filzmoser and Ivan Viola",
  year =       "2016",
  abstract =   "Background In the field of root biology there has been a
               remarkable progress in root phenotyping, which is the
               efficient acquisition and quantitative description of root
               morphology. What is currently missing are means to
               efficiently explore, exchange and present the massive amount
               of acquired, and often time dependent root phenotypes. 
               Results In this work, we present visual summaries of root
               ensembles by aggregating root images with identical genetic
               characteristics. We use the generalized box plot concept
               with a new formulation of data depth. In addition to spatial
               distributions, we created a visual representation to encode
               temporal distributions associated with the development of
               root individuals. Conclusions The new formulation of data
               depth allows for much faster implementation close to
               interactive frame rates. This allows us to present the
               statistics from bootstrapping that characterize the root
               sample set quality. As a positive side effect of the new
               data-depth formulation we are able to define the geometric
               median for the curve ensemble, which was well received by
               the domain experts.",
  journal =    "BMC Bioinformatics",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2016/vad-2016-bre/",
}