An automatic animal pain analysis system based on video timing analysis

Abstract

This invention relates to an automated animal pain analysis system based on video temporal analysis, specifically designed for formalin-induced pain experiments, belonging to the interdisciplinary field of pharmacology and computer vision. The system includes: a multi-channel video acquisition and spatiotemporal feature extraction unit, a pain behavior recognition and quantification unit, and a systematic data management and control unit. The acquisition unit extracts motion and contact features from simultaneously acquired multi-channel foot videos using a spatiotemporal attention convolutional architecture; the recognition and quantification unit uses a temporal behavior analysis model to automatically identify and quantify licking, biting, and foot-lifting events based on the temporal evolution of features, and divides the pain response into acute and delayed phases based on event statistical characteristics; the data management and control unit is responsible for associating and storing data at all levels, executing standardized processes, outputting structured reports, and supporting remote access. This invention achieves full-process automation, objectification, and high-throughput quantification of pain behavior analysis.

Images