James, C. J., Andrianakis, Ioannis, Dillon, James, Mitchell, Philippa, Mould, Richard, Glautier, Steve, O'Connor, Vincent and Holden-Dye, Lindy (2008) Body shape analysis and 2-dimensional cluster representation of C. elegans behaviour: a route to defining complexity in ethanol's modes of action. In: Synaptic Function and Behaviour of C. Elegans Topic Meeting #2, University of Wisconsin, Madison, WI, U.S.A., July, 2008. Published in: Synaptic Function and Behaviour of C. Elegans Topic Meeting #2, Poster Abstracts p. 77.

Abstract

We present a multi-stage system for analysis of C. elegans behaviour through digital capture of magnified images. Images were acquired using a Hamamatsu digital camera attached to a Nikon SMZ800 stereomicroscope. The system identifies C. elegans on the image then parameterises the worm shape on a frame-by-frame basis. This allows for general features such as heading and speed to be obtained at the same time as intricate details of body shape on a frame-by-frame basis. The system consists of 4 stages: (i) An image de-noising and segmentation / thresholding stage that identifies the worm on each video frame and subtracts the background and segments the worm; (ii) An energy minimisation technique that uses the expectation-maximisation (EM) algorithm to learn a fixed number of Gaussian nodes in a Gaussian mixture model (GMM). The nodes of the GMM distribute themselves along the worm shape derived from the first stage uniformly; (iii) The node centers now represent the parameterised worm shape and are stored on a per frame basis. These parameters are used to provide information regarding instantaneous worm shape and frame-by-frame worm activity; (iv) The worm parameters are clustered using a generative clustering algorithm called Neuroscale, this step allows different worm body shapes to be compared and contrasted in a 2-dimensional visualisation space which displays topographic ordering i.e. similar worm shapes cluster together. Finally, as Neuroscale provides a generative mapping, it is possible to track the trajectory of worm body shapes frame-by-frame over the visualisation space. This means that it now becomes possible to track worm heading, speed and shape over time and to observe and quantify worm behaviour with very fine temporal resolution. The system is designed as a general purpose tool with the possibility of application in a wide variety of behavioural preparations. This approach is currently being used to delineate the effects of ethanol on C. elegans locomotor patterns with the aim of resolving discrete effects at concentrations that are relevant to intoxication rather than sedation (see Dillon et al. this meeting).

Request changes to a record

Actions (login required)

View Item