Computer-driven automatic identification of locomotion states in Caenorhabditis elegans.

We developed a computer-driven tracking system for the automated analysis of the locomotion of Caenorhabditis elegans. The algorithm for the identification of locomotion states on agar plates (forward movement, backward movement, rest, and curl) includes the identification of the worm's head and tail. The head and tail are first assigned, by using three criteria, based on time-sequential binary images of the worm, and the determination is made based on the majority of the three criteria. By using the majority of the criteria, the robustness was improved. The system allowed us to identify locomotion states and to reconstruct the path of a worm using more than 1h data. Based on 5-min image sequences from a total of 230 individual wild-type worms and 22 mutants, the average error of identification of the head/tail for all strains was 0.20%. The system was used to analyze 70 min of locomotion for wild-type and two mutant strains after a worm was transferred from a seeded plate to a bacteria-free assay plate. The error of identifying the state was less than 1%, which is sufficiently accurate for locomotion studies.