Tracking and Analyzing Locomotor Changes in Zebrafish

Zebrafish is one of the most widely used model organisms for behavior research in biomedical and pharmaceutical field. Many zebrafish studies used drugs to test its responses, then tracked its movement and analyzed the locomotor features. Such tracking analysis is a challenging task due to the complex body deformation, occasional occlusions, and its “burst” movements. In this study an object detection model YOLOv7 and a multi-object tracking method StrongSORT were integrated to develop an automated zebrafish tracking system and generate relevant locomotor features. Several analyses can be performed through the system. First, we proposed to use approximate entropy to quantify a series of locomotor feature change to evaluate the regularity and unpredictability of movement. Second, through the tracking function we can establish the locomotor trajectory data and collect the time series of several locomotor features including distance, velocity, and different types of body angles when a zebrafish moving in a camera-monitored tank. These analyses help us further understand the impact of drugs through zebrafish’s movement change. The experimental results showed the capabilities of the proposed system and demonstrated that the extracted motion features can be used to distinguish healthy versus diseased groups of zebrafish. The proposed system provides a useful and friendly tool for zebrafish research.