Machine vision based detection of omega bends and reversals in C. elegans.

The behavior of the nematode Caenorhabditis elegans has proven increasingly useful for the genetic dissection of neurobiological signaling pathways and for investigating the neural and molecular basis of nervous system function. Locomotion is among the most complex aspects of C. elegans behavior, and involves a number of discrete motor activities such as omega bends (deep bends typically on the ventral side of the body which reorient the direction of forward locomotion) and reversals (changes in the direction of the locomotion wave that cause a switch from forward to backward crawling). Reliable methods for detecting and quantifying these movements are critical for escape reflexes and navigation behaviors. Here we describe a novel algorithm to automatically detect omega bends, which relies in part on a new method for obtaining a morphological skeleton describing the body posture of coiled worms. We also present an optimized algorithm to detect reversals, which showed improved performance over previously described methods. Together, these new algorithms have made it possible to reliably detect events that are time-consuming and laborious to detect by real-time observation or human video analysis. They have also made it possible to identify mutants with subtle behavioral abnormalities, such as those in which omega bends are dorsoventrally unbiased or uncorrelated with reversals. These methods should therefore facilitate quantitative analysis of a wide range of locomotion-related behaviors in this important neurobiological model organism.