Testing Data for Effects of Hydrogel Compressibility in 3D Traction Force Microscopy

Based on observed compressibility in cellularized hydrogels implemented in 3D traction force microscopy (TFM), we created synthetic test cases to evaluate the effect of this compressibility on computational TFM analysis. In particular, we evaluated TFM analysis to determine the heterogeneous modulus around cells associated with extracellular matrix secretion (ECM), MMP degradation, and other cellular metabolic activity. A prescribed target modulus field was created to reflect expected heterogeneity and ranges of real ECM-infused hydrogel modulus. Realistic cellular geometry and boundary conditions were obtained from an actual TFM experiment. A forward model produced ground-truth target displacements from the prescribed modulus field and compressibility similar to the observed level. For most tests, noise was added to the target displacements at experimental levels. The inverse-modeling approach was validated by achieving accurate recovery of modulus in the case where simulated compressibility matched ground-truth. The necessity of accounting for heterogeneous hydrogel modulus was demonstrated by findings of large differences in predicted mechanical behavior. The effects of assuming incompressibility, common in large-deformation TFM analysis, was assessed by the evaluating the inverse model using a nearly-incompressible formulation. Resulting errors in modulus exceeded 415% and errors in predicted traction forces were amplified 10-fold.

All data are contained in sub-directories of the .zip archive. This dataset first contains the test problem geometry and boundary conditions with the coarse mesh used by the inverse model (cell_data_A) and the fine mesh used by ground-truth simulations to avoid the inverse crime (cell_data_A_finer). The ground-truth displacements and recovered modulus fields for each inverse-model stage in the test case with perfect displacement information are contained in exact_target_test. Displacements (ground-truth and noisy target), recovered modulus fields for each case of inverse-model compressibility, corresponding cellular-surface traction forces, and predicted J are in noisy_target_test. For comparisons with assuming hydrogel homogeneity, real_mvic_data/A_homogeneous contains forward-model displacement, traction, and strain-energy predictions from assuming that modulus was unchanged.

Human MVICs were isolated from consented patients undergoing mitral valve replacement by the Columbia Biobank for Translational Science (IRB-AAAR6796). The deidentified MVICs were then shipped to the University of Texas where culturing and imaging occurred (IBC-2023-00293).