Neural Network in VR

Exploring large-scale spiking neural network simulations in virtual reality

In collaboration with Tobias Rick, Torsten Kuhlen, Tobias Potjas and Markus Diesmann from the institute of neuroscience and medicine (INM-6) at the Forschungszentrum Jülich.

This prototype addresses the visualization of the microscopic activity structure in the mammalian brain. A layered cortical network model represents the local microcircuit below 1mm² of cortical surface. The model includes eight cell types, four excitatory and four inhibitory, distributed in the cortical layers 2/3, 4, 5 and 6. Overall, the model consists of 80,000 neurons and 0.3 billion synapses. Based on the cell-type specific connectivity, the spontaneous and evoked activity of the model reflects in vivo recordings for a time span of 1ms.

On the basis of the gathered data a prototypical 3D visualization of the neuron layers has been created. The visualization reflects the neuronal activity, including e.g. the membrane potential and spike events of the neurons over time. This system is built on top of a virtual reality application in order to provide direct interaction metaphors and increase the depth perception.

Through this project we submitted a subtask proposal within the Human Brain Project which will be funded through the European FET Flagship Programme.

Publications:

  • Anette von Kapri, Tobias C. Potjans, Torsten Kuhlen, Markus Diesmann
    A virtual reality exploration tool for multi-scale data from brain-scale simulations [pdf] [doi] [bibtex]
    Neuroinformatics, 2011, Article in Press
  • Anette von Kapri, Tobias Rick, Tobias C. Potjans, Markus Diesmann, Torsten Kuhlen
    Poster: Towards the Visualization of Spiking Neurons in Virtual Reality [pdf] [doi] [bibtex]
    Medicine Meets Virtual Reality 2011