ARtificial Intelligence Applied to collageN imagiNg datA in physiologic, pathologic and tissue-engineered conditions
Progetto Living tissues are active, multifunctional materials capable of generating,
sensing, withstanding and responding to mechanical stress. Mechanical
stimuli are regulators not only in cell but also of the extracellular matrix
(ECM) activity, with special reference to collagen bundles; thus, sustained
mechanical stimulation may lead to modifications of the collagen
composition, amount and distribution. The mechanical aspects of these
interactions can determine pathophysiological processes, including
developmental defects, fibrosis, inflammatory diseases, tumor growth and
metastasis. Mechanical stimuli of the ECM can also support the body’s
reaction to a therapy and the long-term therapeutic outcome: in fact, the
collagen deposition rate and distribution was shown to determine the
quantity of macroscopic contraction and tensioning of regenerating tissues,
which is important to restore their function.
The previous considerations suggest that the maintaining or re-establishment
of tissue tension, modulating external forces, is key to successful and
regulated tissues remodeling/repairing and wound healing. This is one of the
main focus of an emerging multidisciplinary research field named
mechanobiology that studies how physical forces and cell/tissue mechanics
regulate cell behavior, development, differentiation, physiology and disease.
In this context the ARIANNA project has two main objectives: (1) the
identification of three-dimensional morphometric parameters deriving from
the analysis of pathological (fibrotic or cancerous) and regenerated collagenbased tissues (with comparison with healthy twin contexts); (2) to
reconstruct volume forces and contact forces acting locally in these contexts.
The morphometric parameters are those that will be extracted by the big
amount of high-resolution phase-contrast synchrotron imaging data,
contained in the archive of the PI research group and acquired in four
anatomic sites of interest: (1) human dental peri-implant connective tissue
(regenerated site/wound healing model); (2) stroma in human oral squamous
cell carcinoma (pathologic site/ cancerous model); (3) pulmonary fibrosis in
transgenic mice affected by scleroderma (pathologic site/1st fibrotic model);
(4) human uterine leiomyoma (pathologic site/ 2nd fibrotic model).
Segmentations guided by artificial intelligence (machine learning and deep
learning procedures) will be implemented and followed by data mining.