Detecting local processing unit in drosophila brain by using network theory

TL;DR

This paper applies network community detection to Drosophila neuron networks, successfully identifying local processing units and subdivisions, and revealing layer structures consistent with optical imaging data.

Contribution

Introduces a network theory-based method for automatic detection of LPUs in Drosophila brain, validated by known structures and optical imaging.

Findings

26 communities matched known LPUs

13 subdivisions identified within LPUs

Layer structures in fan-shaped body confirmed

Abstract

Community detection method in network theory was applied to the neuron network constructed from the image overlapping between neuron pairs to detect the Local Processing Unit (LPU) automatically in Drosophila brain. 26 communities consistent with the known LPUs, and 13 subdivisions were found. Besides, 45 tracts were detected and could be discriminated from the LPUs by analyzing the distribution of participation coefficient P. Furthermore, layer structures in fan-shaped body (FB) were observed which coincided with the images shot by the optical devices, and a total of 13 communities were proven closely related to FB. The method proposed in this work was proven effective to identify the LPU structure in Drosophila brain irrespectively of any subjective aspect, and could be applied to the relevant areas extensively.