An In-DEPTH Mechanical Characterization of Articular Cartilage: Machine Learning Optimization of Stereo DIC Data

Abstract

Articular cartilage is a complex, spatially graded material. Understanding how the tissue’s mechanical behavior changes throughout its depth is important for understanding musculoskeletal pathologies and their possible treatments. This study aimed to classify the fibril reinforced poroviscoelastic behavior of porcine cartilage in a depthwise manner. To do so, unconfined compression testing was paired with stereo-DIC to generate depth dependent data. Data were used to make sample specific finite element models. These models were optimized through a machine learning pipeline that updated the material parameters of the governing constitutive models to best match the experimental results. Optimized material property results show minor variation in depth, contrasting much of prior literature that showed greater variation with depth. This work provides a framework for simultaneously characterizing multiple aspects of material behavior in spatially graded materials.