Abstract
Network represents adjacent relationships, connections, and interactions among constituent elements in complex systems but often loses critical information about spatial configurations. However, structure-function relationships in biological systems, e.g., the human heart, are highly dependent on both connectivity relationships and geometric details. Therefore, this paper presents a new self-organizing approach to derive the geometric structure from a network representation of the heart. We propose to simulate the network as a physical system, where nodes are treated as charged particles and edges as springs and then let these nodes self-organize to reconstruct geometric details. Despite random initiations, this network evolves into a steady topology when its energy is minimized. This study addresses the open question, i.e., "whether a network representation can effectively resemble spatial geometry of a biological system," thereby paving a stepstone to leverage network theory to investigate disease-altered biological functions.