Signal extraction from movies of honeybee brain activity : the ImageBee plugin for KNIME

Abstract

BackgroundIn the antennal lobe, a dedicated olfactory center of the honeybee brain, odours are encoded as activity patterns of coding units, the so-called glomeruli. Optical imaging with calcium-sensitive dyes allows us to record these activity patterns and to gain insight into olfactory information processing in the brain.MethodWe introduce ImageBee, a plugin for the data analysis platform KNIME. ImageBee provides a variety of tools for processing optical imaging data. The main algorithm behind ImageBee is a matrix factorisation approach. Motivated by a data-specific, non-negative mixture model, the algorithm aims to select the generating extreme vectors of a convex cone that contains the data. It approximates the movie matrix by non-negative combinations of the extreme vectors. These correspond to pure glomerular signals that are not mixed with neighbour signals.ResultsEvaluation shows that the proposed algorithm can identify the relevant biological signals on imaging data from the honeybee AL, as well as it can recover implanted source signals from artificial data.ConclusionsImageBee enables automated data processing and visualisation for optical imaging data from the insect AL. The modular implementation for KNIME offers a flexible platform for data analysis projects, where modules can be rearranged or added depending on the particular application.