Reconstructing collagen-I fiber network using the stochastic annealing for geometric analysis of the pore network in 3D

Abstract

The extracellular matrix (ECM) fiber alignment is important factor when there is cancer progression as cancer invasion is enhanced through ECM fiber alignment. While there have been great efforts to understand the changes in mechanical properties of ECM using both in vitro and in silico models, the detailed characterization of the pore network of the ECM still remains to be elucidated. The analysis of the pore network is crucial as the network is affected by matrix metalloproteinase, which cancer cells secrete and utilize to migrate. Until now, the investigation on the pore network of ECM structure has been achieved using the pore diameter. However, in this work, we have hypothesized that quantifying the pore throat can better capture the changes of the pore network induced by the alignment of the fiber. First, we have imaged fibrous collagen-I, and applied the watershed algorithm to extract information of the pore network. With the extracted data, ECM is reconstructed by stochastic annealing method, and the watershed algorithm is used again on the reconstructed ECM structures to validate that the reconstructed structure can recover the pore network of the original structure. Finally, by manipulating the fiber alignment of the reconstructed structure, we have shown that the fiber alignment significantly changes the area of the pore throats, while the average size of the pores remained mostly unchanged. Our finding suggests that the pore network analysis including the pore throats can reveal the changes in the pore network that previously was note possible fully characterized using the pore size measurements commonly achieved with the reconstituted collagen-I characterization. At the last part of the paper verifies the findings that the characterization of the pore throat is more suitable in characterizing the changes in the collagen network caused by the fiber alignment experimentally.