Virtual Retina
retina logo
Virtual Retina
Large scale simulator of biological retina
INRIA CeCILL C open-source license (2007)
Partially supported by the EC IP project FP6-015879 FACETS
Neuromathcomp research-team, INRIA Sophia Antipolis - Méditerranée

Overview
Web-service Results
Customization
Download Publications Contact


selection



Output of the software

What you will get from Virtual Retina is a list of spikes in an ASCII file, as decribed in the sample result below.

9333 0.0153373
12859 0.0153388
9273 0.0153409
9539 0.0153444
9127 0.0153548
9334 0.0153571
7899 0.0153612
9540 0.0153651
6211 0.015371
9128 0.0153717
9272 0.015372
7900 0.0153743
9332 0.015375
7692 0.0153755

etc.

The number on the left is the index of the spiking cell, the number on the right is the corresponding time of spike, in seconds after the beginning of the simulation.


Reconstruction

In order to visualize the results, Virtual Retina also offers you the possibility to reconstruct a sequence, where each emitted spike adds a spot to the image, for a duration of 50 ms, and with a size that depends on the distance from the center of the retina (far from center means less cells, but associated to bigger spots in the reconstruction).

walkingsequence.gif
Spike emission of a modeled channel of ON parvocellular cells,
in response to five seconds of video input.


 This reconstruction procedure allows to observe the qualitative effect of the contrast gain control stage. We give below two examples of retinal output on the same input image (left) , with (right) and without (middle) a contrast-gain-control feedback stage.


contrastgaincontrolresult


Physiological Validation

Interestingly, Virtual Retina can reproduce several physiological recordings on real ganglion cells. In particular, we reproduced (see below) the responses of ganglion cells to drifting gratings (Left, Enroth-Cugell and Robson, 66), and the non-linearity of ganglion cells with the mean level of contrast, when the stimulus is a static grating modulated by an approximate temporal white noise (Right, Shapley and Victor, 78).
For both these stimuli, our non-linear contrast-gain-control stage was found mandatory to reproduce the results.

driftinggratings            shapleyvictor