Application of computer vision in the automatic analysis of feeding behavior in C. elegans

Abstract

The nematode Caenorhabditis elegans has been widely used as a perfect model organism to study the relationship between genes and behavior. The pharyngeal microcircuit of the worm controls a complex feeding behavior. In order to study the molecular basis of this feeding behavior, it is necessary to identify subtle differences in feeding activity of the worm. However, most of the phenotype analyzing of feeding behavior is accomplished by human eyes. And it is a tough and poor efficiency job to analyze the fast pumping muscle of the worm. To help improving this problem, an automated system has been developed based on computer vision for the high-throughput analysis of the feeding behavior by virtue of a simple webcam. Our system enables the consistent and subtle analysis of C. elegans pumping recordings and the accuracy of pumping detection is up to 98%. Under this high accuracy, the time cost of the behavior analysis is cut down by 67% versus human manipulation.