The Principles (revised)

Apart from the fact, that virtual reality for mice sounds pretty fun to do, one might ask the question what the scientific benefit is.

While this setup allows us to do many different experiments, we are primarily interested in the spatial navigation system, i.e. the hippocampal formation and all its interesting cells (for background, read: Place cells, Grid cells, Head direction cells and boundary vector cells). All of these cell types have been discovered using extracellular recordings. This method allows us to see firing patterns of individual cells, but to get a better insight as to how the specific firing patterns of those cells are generated, we need single unit recordings.

The key problem is: to investigate spatial navigation, the rodent needs to run through a maze or an open field, while doing so however, we can't record from single neurones because of the movement. The solution: keep the head fixed in one place and generate a virtual reality around it. This is what this blog is all about.

Virtual reality for rats has been pioneered by Hölscher et. al.(2005) and adapted by Harvey et.al. (2009) for mice. The setup developed here is based on the latter study. In order to visualise the design and make sure everything fits together, a 3d model has been created in Autodesk (r) Inventor (r) 2011.

The model of the setup from the side. Note that the screen is not showen, only the skeleton of copper tubes.

Components:

Treadmill
We want the animal to be able to run freely into any direction. To allow this we use a polystyrene sphere sitting in a cup of slightly bigger size. An air cushion is produced between cup and sphere allowing virtually frictionless movement of the sphere. More information about the treadmill can be found on the corresponding page.

The ball in the centre is the spherical treadmill. The air supply (not modelled) comes in from the bottom and creates an air cushion for the ball to run on (you can see a video of this here). The two poles reaching for the top of the ball will hold the headplate to fix the head of the mouse in place. The semi-circular tubes serve as a skeleton for the screen and the projector and mirror system will project the virtual reality onto the screen.

 The treadmill. More information here.
Frame
The entire setup is placed on an airtable to minimise vibrations wich could interfere with electrophysiological experiments. Further, our setup requries things to be mounted at about 80cm over the surface of the airtable, making the use of posts impractical. Therefore we built a simple cube-frame with a breadboard mounted at a convenient height. This allows us to mount things from the breadboard in the same way we can mount something on the airtable.

Frame and breadboard. Mirrorsystem and screen are mounted to it.

Mirror System and Projector 
We based our design on that of the previous two studies mentioned earlier. A projector outputs onto a flat mirror which in turn deflects onto a convex mirror of a specific profile (Chahl and Srinivasan, 1997). From there the picture is spread onto the screen. It is a suboptimal solution as we have to focus the projector onto the convex mirror instead of the screen which means unfocused light is scattered around, decreasing contrast and sharpness.
 
The profile of the convex mirror is designed to deflect incident rays by a certain factor. This is easier explained by an example: if the profile is calculated to have an 'angular amplification' of 10, then a ray with an incident angle of 1 will be deflected at 10 degrees. A ray at 2 degrees at 20, 3 at 30 and so on. Having this profile simplifies calculating the dimensions of screen and mirror as well as the image transformation required to display the virtual reality correctly.

Mirror system (the projector is not modelled). The round disc is the flat mirror deflecting the projected picture so the centre of the projector output hits the apex of the convex mirror.




This picture shows the raytracing from the origin (projector, top left) until the rays hit the screen. This is of course a very simplified raytracing diagram but it shows where the rays hit the screen.
Screen
The projection screen for the virtual reality has the shape of a torus centered around the mouse. It consists of two rings, one at the top, one at the bottom, copper tubes mounted in between them and canvas paper strips which provide the actual surface. The toroidal shape is difficult to construct but has a number of geometric properties very beneficial for the virtual reality.


If one would complete the semi-circular copper tubes into full circles they would all interesect in two points on a vertical axis. The mouse is located at the lower one of these points and the convex mirror on the upper one. This has the effect that the sight line of the projection and the sight line of the mouse to any point on the screen always have a constant angle, making the correct geometric presentation of a virtual reality much easier.

The screen without paper strips.

Reward System
A reward system will dispense a liquid reward such as sugar water whenver the animal has successfully exectuted a task. A drop will be produced just in front of the animal so it can reach it with its tongue without moving its head.

Head Fixation
To keep the head of the mouse in place (which prevents it from falling off and allows us to do recordings) a headplate will be implanted on the mouse. This headplate is then fixed to a post for the duration of the experiment, rendering the head immovable.


Interplay 
Movement of the treadmill will be recorded with computer mice. The movement information is fed back into the computer that runs the virtual reality and updates the location within the virtual reality accordingly. If the mouse finishes a task correctly the computer will send a signal to the reward system to release a pre-defined amount of reward.

If you've got any questions relating to this system, feel free to leave a comment or email me.

References:

Chahl JS, Srinivasan MV. Reflective surfaces for panoramic imaging. Applied optics. 1997;36(31):8275-85. Available at: http://www.ncbi.nlm.nih.gov/pubmed/18264368.

Harvey, C. D., Collman, F., Dombeck, D. A., & Tank, D. W. (2009). Intracellular dynamics of hippocampal place cells during virtual navigation. Nature, 461(7266), 941-946. Nature Publishing Group. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/19829374

Hölscher, C., Schnee, A., Dahmen, H., Setia, L., & Mallot, H. A. (2005). Rats are able to navigate in virtual environments. Journal of Experimental Biology, 208(Pt 3), 561-569. Co Biol. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/15671344

1 comment: