GMRV Publications

NeuroScheme: Efficient multiscale representation for the visual exploration of morphological data in the human brain neocortex

Luis Pastor, Susana Mata, Pablo Toharia, Sofia Bayona, Juan P. Brito, Juan J. Garcia-Cantero
Proc. of Congreso Español de Informática Gráfica, page 117-125 - 2015
Download the publication : 117-125.pdf [21.7Mo]  
The analysis of data sets from such a complex domain as neuroscience is severely hampered by the brain complexity and by the huge amount of data originated in current experiments and simulations. The complexity of the brain stems not only from the enormous number of relevant entities that integrate it (for example, around 10<sup>15 </sup>synapses), but also from the complex interrelations among brain structures, which have to be analyzed at the many different organizational levels of the brain. Visualization techniques have proven to be valuable in order to explore and analyze complex systems from a wide variety of areas. Nevertheless, their application to the field of neuroscience must face some domain specific difficulties. In particular, the overwhelming amount of data poses a challenge to the current storage and processing capabilities, calling for compact data descriptions. Furthermore, the visual analysis of neural scenes requires the design of novel visual representations and techniques that help experts in their data exploration and interpretation activities, assisting them in fighting complexity. In this paper we propose a visual exploratory framework that facilitates the process of knowledge extraction from complex neural scenes. This framework contains a multilevel structure, following the different organizational levels of the brain. Schematic or iconic symbols have been designed to portray the entities at each level, providing graphical representations that emphasize relevant features while hiding less important information. These schematic views, together with a multilevel organization, allow the exploration of the brain at different scales, combining in the same view different levels of abstraction whose entities can be either schematically represented (at different abstraction levels) or geometrically depicted at the finest level of detail. The fact that users can choose high geometric detail for just a fraction of brain structures results in a dramatic reduction of visual clutter, decreasing the difficulty of the visual exploration and interpretation task. The complexity due to sheer data size can be also reduced by generating the visual representations on the fly from compact data representations. This work presents a first implementation of this framework, focused on the representation of the morphological aspects of the brain cortex. The underlying multilevel structure and the design of the graphical schematic representations constitute an approach that provides organized and uncluttered views that facilitate visual analysis, a task much more difficult (even unapproachable) if all brain structures were depicted at full geometric detail.

Images and movies

NeuroScheme.png [2.9Mo]
ceig_2015_video_12.mp4 [51.8Mo]

BibTex references

  author       = "Pastor, Luis and Mata, Susana and Toharia, Pablo and Bayona, Sofia and Brito, Juan P. and Garcia-Cantero, Juan J.",
  title        = "NeuroScheme: Efficient multiscale representation for the visual exploration of morphological data in the human brain neocortex",
  booktitle    = "Proc. of Congreso Español de Informática Gráfica",
  pages        = "117-125",
  year         = "2015",
  publisher    = "Eurographics Association",
  note         = "DOI: 10.2312/ceig.20151208",
  url          = ""

Other publications in the database

» Luis Pastor
» Susana Mata
» Pablo Toharia
» Sofia Bayona
» Juan P. Brito
» Juan J. Garcia-Cantero