Synthetic FEM Dataset for Inverse Identification of Breast Tissue Elastic Properties

This dataset contains synthetic finite element simulation data generated for the study "Finite Element–Machine Learning Surrogates for Inverse Identification of Breast Tissue Elastic Properties: Robustness, Sparse Observations, and Extrapolation Analysis". The dataset comprises 1,000 finite element simulations of breast compression generated using FEBio. The simulations vary the Young's modulus of skin, adipose tissue, and glandular tissue within predefined parameter ranges, together with the prescribed compression displacement. The resulting displacement fields and corresponding simulation parameters were used to train and evaluate machine-learning surrogate models for inverse identification of tissue elastic properties. The main dataset, datasetEsD_1000_v2.csv, contains 48,666 displacement features corresponding to three displacement components (ux, uy, uz) at 16,222 finite element nodes, together with the material parameters Es, Ef, Eg and the prescribed displacement D. Version 1.1.0 extends the original dataset deposit with sampled_parameters.tsv, containing the exact parameter combinations used for the 1,000 FEM simulations, and README_dataset.md, providing detailed documentation of the dataset structure and its relationship to the computational workflow. The numerical FEM dataset itself is unchanged from version 1.0.0. The complete computational workflow, including the FEBio model/template, parameter-sampling script, simulation-execution and displacement-extraction scripts, machine-learning code, cross-validation assignments, random seeds, software environment, and validation analyses, is permanently archived in the associated software release v1.1.0: DOI: 10.5281/zenodo.21959135