MorphoGNN: Morphological Embedding for Single Neuron with Graph Neural Networks

Abstract

With the development of optical imaging systems, neuroscientists can now obtain large datasets of morphological structure at a single neuron scale positioned across the whole mouse brain. However, the enormous amount of morphological data challenges the classic approach of neuron classification, indexing and other analysis tasks. In this paper, we propose MorphoGNN, a single neuron morphological embedding based on the graph neural networks (GNN). This method learns the spatial structure information between the nodes of reconstructed neuron fibers by its nearest neighbors on each layer and captures the lower-dimensional representation of a single neuron through an end-to-end model. This model is composed of densely connected edge convolution (EdgeConv) layers and a double pooling operator, regularized with joint cross-entropy loss and triplet loss. An increasing population of the neighbor nodes meets the need of learning more information with features expanding at the deep layer. We tested the proposed embeddings on the neuron classification and retrieval tasks. Our method achieves competitive performance both on the general point cloud dataset and the neuron morphology dataset.

Joint Learning Neuronal Skeleton and Brain Circuit Topology with Permutation Invariant Encoders for Neuron Classification

“…The morphology, or shape, of a neuron can provide significant insights into its function within the nervous system. Traditional methods based on feature statistics (Gillette and Ascoli 2015;Kanari et al 2019;Zhu et al 2022) usually conduct analysis on the optical data. However, the resolution of neuronal morphological data obtained from optical microscopes is relatively low, which cannot provide neuronal connectivity information.…”

“…Taking advantage of these innovations, the brain circuit reconstruction have been made to map the connectivity in Drosophila optic and antennal lobes (Berck et al 2016;Takemura et al 2013), the mouse retina, thalamus, and cortex (Greene, Kim, and Seung 2016;Lee et al 2016;Morgan et al 2016). However, there are relatively few works (Costa et al 2016b;Schubert et al 2019;Zhu et al 2022) that use neural networks for feature extraction and classification of the above-described neuron data, and none of which take into account the topological information of brain circuit.…”