A new method for characterizing the long-term behavior of zebrafish from the trajectory

Image-based tracking technologies are popular in the behavioral research of zebrafish. The moving trajectory of a zebrafish is derived from the raw videos and the behavior of the zebrafish can be characterized from the moving trajectory. The speed is a common index in characterizing the trajectory of a zebrafish. However, if the trajectory is directly differentiated over time to calculate the instantaneous speed, the noise of the system will be amplified by the differentiation, significantly decreasing the Signal-Noise Ratio. As a result, it will be difficult to separate the useful information from the noise. This paper proposes two parameters for characterizing the linear moving and turning behaviors based on the zebrafish´s trajectory. Different from the conventional instantaneous speed calculation, the proposed two parameters focus on the long-term observation. As it doesn´t differentiate the trajectory, the noise of the system will not be amplified and the resultant plots are very smooth. Further, the proposed parameters are designed to obtain the features of the zebrafish´s movement, not the exact values of the linear and angular speeds. Experimental results show that the characteristics of zebrafish´s movement, e.g., the rest/wake behaviors and almost constant linear and angular speeds, can be clearly observed via the proposed parameters.