Knowledge Base

Plot Heatmaps

Notes on Image Resolution and Formats

The available image resolutions depend on the aspect ratio of the original background pictures. The zoom level chosen for the plots affects the default resolution.

Available Image Formats

• Portable Network Graphic (*.png) (default). A picture format using lossless compression.
• JPEG file (*.jpg). A compression format for color images, particularly photographs. It can reduce file sizes to about 5% of their original size, but some detail might be lost in the compression.
• Windows Bitmap (*.bmp). A non-compressed format generally resulting in large files.
• Graphics Interchange Format (*.gif). A lossless format limited to 256 colors. This makes the GIF format suitable for storing graphics with relatively few colors such as simple diagrams.

Heatmaps in EthoVision XT

What Are Heatmaps?

A heatmap is a graphical representation of the subject's position in the 2D space where the density of utilization of a specific place is represented as a color. Heatmaps facilitate identification of "hotspots" and clustering of data points.

Choose Heatmap Visualization to generate heatmaps. You can generate a heatmap for single tracks or groups of tracks, and sort heatmaps according to various criteria. You can combine heatmaps with data selection to visualize, for example, the subject's position while it shows a specific behavior. Finally, you can export heatmaps to a high-resolution graphics file.

How Are Heatmaps Created?

In EthoVision XT, a track is a group of sample points, each with x and y coordinates. If one sample is found at specific coordinates, the frequency (= number of samples) for the corresponding pixel in the video image is 1. If more samples are found exactly on that pixel, for example when the subject returns to the same location, or when it sits still for some time in that location, then the frequency for that pixel is higher: 2, 3, 4, etc.

Important: Here by frequency we mean the number of samples that are found on a specific pixel, not the number of entries in a zone or the like. The number of samples can be plotted in a 2D histogram, where the length of each column indicates how much time the body point of the subject was found exactly at that location.

The heatmap is created from that histogram, based on the concept of kernel density estimation. At each pixel, the probability is calculated to find the subject at that specific location. The density function states that there is some (lower) probability to find the subject also in the surroundings of the focal pixel. This can be represented with a bell-shaped three-dimensional curve.

The kernel density function is calculated at each location. The single 3D bell curves for neighboring locations are merged together, resulting in higher and wider curves.

However, heatmaps in EthoVision XT are a 2D representation of the hilly landscape that results from the kernel density estimation. It is like viewing the hills and valleys from the top. The colors represent the altitude of the landscape.

In EthoVision XT, the colors may represent the cumulative time in a particular pixel, or the fraction of the track that is found at that pixel. This depends on what you choose under Color Level.

Examples of Heatmaps

Morris Water Maze Test

In this example, we compare the swim pattern between two groups of rats in probe trials. The young rats (right) display more focused search than the old ones (left).

Zebrafish in a Novel Tank Diving Test

In this example we want to show the bias toward top-swimming after the fish was treated with an anxiolytic drug. The two zones, Top and Bottom, are outlined in white.

Novel Object Recognition Test

In this test the subject explores two objects, one familiar and one novel. The heatmap clearly shows the difference in the "density" of location points around the two objects. The two object zones are outlined in white.

Mosquito in a Wind Tunnel

A mosquito tends to reach the odour source located behind the upwind screen of the flight chamber. The middle of the screen is indicated with a cross.

Notes

• Be careful when interpreting color heatmaps. First, the colors lack the natural perceptual ordering found, for example, in gray scale color maps. The changes between colors lead to perception of gradients that are not actually present, making actual gradients less prominent.