Figure 1
From: Deep learning of material transport in complex neurite networks
An overview of the GNN model to learn and predict neuron material transport process. The input neurite network is first decomposed into pipes and bifurcations to create the graph representation of the neurite network. Next, the input features \(x_i\) of each pipe or bifurcation are processed by the corresponding GNN simulator (\(G^p_S\) or \(G^b_S\)) to generate intermediate concentration result \(x_{mid}\). Then, the GNN assembly model (\(G_A\)) takes \(x_{mid}\) as input and computes the interaction between simulators to predict concentration result \(x_o\).
