Abstract:Graph theoretical approaches have been proven to be effective in the characterization of connected systems, as well as in quantifying their dysfunction due to perturbation. In this paper, we show the advantage of a non-Euclidean (hyperbolic) representation of networks to identify local connectivity perturbations and to characterize the induced effects on a large scale. We propose two perturbation scores based on representations of the networks in a latent geometric space, obtained through an embedding onto the hyperbolic Poincaré disk. We numerically demonstrate that these methods are able to localize perturbations in networks with homogeneous or heterogeneous degree connectivity. We apply this framework to identify the most perturbed brain areas in epileptic patients following surgery. This study is conceived in the effort of developing more powerful tools to represent and analyze brain networks, and it is the first to apply geometric network embedding techniques to the case of epilepsy.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to detect local perturbations in the network in the potential hyperbolic embedding space and characterize the impact of these perturbations on large - scale networks. Specifically, the authors propose a method based on non - Euclidean (hyperbolic) representation to identify and characterize local connectivity perturbations in the network. This method embeds the network into the Poincaré disk and uses two perturbation scores to quantify and locate local changes in the network. These scores can effectively locate perturbations in networks with uniform or heterogeneous degree distributions. ### Main research questions: 1. **Detecting local perturbations**: How to effectively detect changes in local connectivity in complex networks? 2. **Characterizing the impact of perturbations**: How to quantify and characterize the impact of these local perturbations on the entire network? 3. **Applying to brain networks**: How to apply these methods to the brain networks of epilepsy patients to identify the most affected brain regions after surgery? ### Method overview: - **Hyperbolic embedding**: Embed the network into hyperbolic space, using the Poincaré disk model. Hyperbolic embedding can better capture the hierarchical structure of the network and the similarity between nodes. - **Perturbation scores**: - **HypDisp score**: Calculate the displacement of nodes in hyperbolic space, which is used to quantify the position change of nodes due to perturbation. - **kNHypD score**: Calculate the distance change between a node and its nearest neighbor nodes in hyperbolic space, which is used to quantify the impact of local perturbations on the node and its neighbors. ### Experimental verification: - **Synthetic networks**: Conduct experiments on networks generated by the Watts - Strogatz and Barabási - Albert models to verify the effectiveness of the method. - **Clinical data**: Apply these scores to the brain network data of 51 epilepsy patients who have undergone surgery, and compare the changes in the brain network before and after surgery. ### Results: - **Synthetic networks**: The HypDisp and kNHypD scores can effectively detect and quantify local perturbations, and as the degree of perturbation increases, the score values increase significantly. - **Clinical data**: In the brain networks of epilepsy patients, these scores can accurately locate the most affected brain regions after surgery, especially in patients after left temporal lobe and right temporal lobe surgeries. ### Conclusion: This study shows the advantages of hyperbolic embedding and perturbation scores in detecting and characterizing local perturbations in networks, especially the application potential in brain network analysis. These methods provide new ideas for developing more powerful tools to represent and analyze brain networks.

Detecting local perturbations of networks in a latent hyperbolic embedding space

Hyperbolic embedding of brain networks detects regions disrupted by neurodegeneration in Alzheimer's disease

Hyperbolic embedding of brain networks as a tool for epileptic seizures forecasting

Hyperbolic embedding of brain networks can predict the surgery outcome in temporal lobe epilepsy

Unified Topological Inference for Brain Networks in Temporal Lobe Epilepsy Using the Wasserstein Distance

Neighbourhood topology unveils pathological hubs in the brain networks of epilepsy-surgery patients

Topological Embedding of Human Brain Networks with Applications to Dynamics of Temporal Lobe Epilepsy

Identifying the latent space geometry of network models through analysis of curvature

Joint Detection of Community and Structural Hole Spanner of Networks in Hyperbolic Space

Spectral embedding and the latent geometry of multipartite networks

Exploring Brain Connectivity with Spatial-Temporal Graph Neural Networks for Improved EEG Seizure Analysis.

Geometrical congruence, greedy navigability and myopic transfer in complex networks and brain connectomes

Networking property during epileptic seizure with multi-channel EEG recordings

Localization of epileptogenic zone based on time-varying effective networks

Enhancing Non-Invasive Pre-Surgical Evaluation through Functional Connectivity and Graph Theory in Drug-Resistant Focal Epilepsy

Towards a Quantitative Theory of Digraph-Based Complexes and its Applications in Brain Network Analysis

Latent geometry of bipartite networks

Network phenotypes and their clinical significance in temporal lobe epilepsy using machine learning applications to morphological and functional graph theory metrics

Neural Embeddings of Graphs in Hyperbolic Space

How the latent geometry of a biological network provides information on its dynamics: the case of the gene network of chronic myeloid leukaemia

Localization of Epileptogenic Zone Based on Cortico-Cortical Evoked Potential (CCEP): A Feature Extraction and Graph Theory Approach.