Unraveling the biomechanical properties of collagenous tissues pathologies using synchrotron-based phase-contrast microtomography with deep learning

Mechanical stimuli are regulators not only in cells but also of the extracellularmatrix activity, with special reference to collagen bundles composition, amountand distribution. Synchrotron-based phase-contrast computed tomography waswidely demonstrated to resolve collagen bundles in 3D in several body districtsand in both pre-clinical and clinical contexts. In this perspective study wehypothesized, supporting the rationale with synchrotron imaging experimentalexamples, that deep learning semantic image segmentation can better identifyand classify collagen bundles compared to common thresholding segmentationtechniques. Indeed, with the support of neural networks and deep learning, it ispossible to quantify structures in synchrotron phase-contrast images that werenot distinguishable before. In particular, collagen bundles can be identified by theirorientation and not only by their physical densities, as was made possible usingconventional thresholding segmentation techniques. Indeed, localised changes infiber orientation, curvature and strain may involve changes in regional straintransfer and mechanical function (e.g., tissue compliance), with consequentpathophysiological implications, including developmental of defects, fibrosis,inflammatory diseases, tumor growth and metastasis. Thus, the comprehensionof these kinetics processes can foster and accelerate the discovery of therapeuticapproaches for the maintaining or re-establishment of correct tissue tensions, as akey to successful and regulated tissues remodeling/repairing and wound healing.