Mechanical characterization of soft tissue in vivo by microstructural imaging and physics-informed neural networks: bridging the gap between biomechanics and clinical practice

Computational biomechanics is a fast-growing area of science. However, many of its fruits have not yet been translated into clinical practice.

The main barrier to the translation of computational biomechanics into clinics is the lack of information about patient-specific mechanical properties of biological tissues. There are some methods for partially assessing these properties, but they have substantial limitations.

A major cause of these limitations is the operating principle of current approaches, which rely on the analysis of the response of a tissue to some form of mechanical loading in vivo. To overcome this bottleneck, we propose a paradigm change.

We will develop a method to infer mechanical properties of soft biological tissues in vivo based on a fundamentally new operating principle.

To this end, we will leverage the synergy of three scientific areas covered by the three principal investigators of this project: (i) magnetic resonance imaging (MRI), (ii) experimental biomechanics, and (iii) physics-informed machine learning.

Specifically, we will develop a new type of subvoxel MRI relaxometry to probe tissue microstructure non-invasively and establish a combined experimental and computational framework that will uncover for the first time the mechanistic link between transcriptomics, microstructure, and mechanical properties of soft biological tissues in a detailed manner.

By leveraging this information with novel physics-informed machine learning techniques, we will gain the ability to determine the mechanical properties of soft tissues from clinical MRI data and blood samples with unprecedented accuracy and completeness.

Our approach will be a crucial steppingstone to translate biomechanical computational models into clinical practice at a large scale.

As a proof of concept, we will demonstrate how our new method can support the diagnosis of heart failure with preserved ejection fraction (HFpEF), one of the most common causes of mortality and morbidity.