Knowledge Base

Physical Setup

The following suggestions are specifically to optimize video tracking:

• Place the plus maze in such a way that its apparent size is maximized. You can obtain this by rotating the plus maze (or the camera) until the closed arms lie along one of the diagonals of your video window.
• The lighting should be as even as possible throughout the arena. Bright lights are not necessary but should you need to supplement the light, consider red (or infra-red) light, as the animals will be less sleepy.
• The camera should have a good view of the entire region the animal can be in. Center the plus maze in order to have the image of the arms as symmetrical as possible. This minimizes inaccuracy of calibration due to perspective.
• The lighting should be diffuse, so as not to cast strong shadows (which might be tracked instead of the animal). Even, diffuse lighting is important to optimize tracking of the nose.
• A plus maze with transparent walls may help reducing shadows and make it easier to observe closed arm rears. There is little practical difference between the two mazes in terms of their ability to detect differences in anxiety-related behavior. However, the transparent design may reduce the sensitivity for the detection of anxiolytic drug effects because it decreases the anxiogenic potential of the open arm, therefore leading to the reduction of anxiety-related behavioral baseline (Violle et al., 2009 and references therein).
• The background should contrast with the animal. If necessary, use a different setup for light and dark colored animals. In the closed arms, the contrast between animal and background is generally lower. Take this into account when positioning lights and choosing the background color.
• Make sure that the color of the floor is in good contrast with the color of the animal, and in minimal contrast with the arms. This makes it possible to track the animal's nose off the edge of the open arms.
• Both the plus maze arms and the floor should not be reflective. If necessary, place an opaque rubber mat on the floor.
• If there are barriers above the closed arms, remove them. If that is not possible, in the Detection Settings use the Dilation-Erosion filter (see page 97) to prevent EthoVision from "seeing" the animal cut in two. See also Contour Settings in the EthoVision XT Help.
• Depending on circumstances, pools of urine can cause problems with the tracking. An absorbent base can help.
• Always place the apparatus in the same position in the room, as cues from overhead might influence the behavior of the animal. For the same reason, the experimenter should not be visible to the animal during the trial (automated tracking makes this possible).
• If you want to use the Deep Learning technique to track the subject's nose, note that there are additional requirements and limitations. Among other things, you need a powerful graphics card. See Deep Learning: Requirements and Limitations in the EthoVision XT Help.

Source: EthoVision XT 18 - Application Manual, Noldus Information Technology