Skip to content

README.md

Latest commit

 

History

History
99 lines (72 loc) · 4.77 KB

README.md

File metadata and controls

99 lines (72 loc) · 4.77 KB
Raw

image2iga: IGA-DigitalTwin-Pipeline

An integrated ecosystem for creating Analysis-Ready Digital Twins from volumetric images (CT-scans). This repository serves as the central hub and orchestrator for a pipeline bridging Computer-Aided Design (CAD), Virtual Image Correlation (VIC), and IsoGeometric Analysis (IGA).

🏗 Modular Architecture

The framework consists of three independent libraries. This repository contains the integration example, but the core libraries must be installed separately:

  • bsplyne: The geometric core. Handles multivariate tensor-product B-splines and multi-patch topology.
    📖 Documentation | 📂 Examples
  • IGA_for_bsplyne: The mechanical solver. A high-performance IGA linear elasticity engine.
    📖 Documentation | 📂 Examples
  • volVIC: The vision layer. Performs Virtual Image Correlation to fit B-spline surfaces onto 3D image features.
    📖 Documentation | 📂 Examples

Note: Both bsplyne and IGA_for_bsplyne include a dedicated examples/tutorial folder containing a pedagogical PDF guide and ordered Python scripts (01_..., 02_...) for a smooth onboarding.

🛠 Installation

We recommend using Conda to manage optional but highly recommended high-performance dependencies (like SuiteSparse).

1. Environment Setup

conda create -n iga_pipeline python=3.9
conda activate iga_pipeline

2. Install Optional Dependencies

conda install scikit-sparse  # Highly recommended for large problems
pip install sparseqr         # Recommended for fast C1 constraint imposition

3. Install the Libraries

To ensure a consistent environment, it is recommended to install the exact versions from the requirements.txt file:

pip install -r requirements.txt

Alternatively, since volVIC depends on both bsplyne and IGA_for_bsplyne, you can install the full stack (potentially with more recent versions) via volVIC:

pip install volVIC

📖 Key Features

  • Unified B-spline Representation: Continuous pipeline from image registration to simulation without remeshing.
  • High Performance: Critical numerical kernels are either JIT-compiled with numba or fully vectorized with numpy and sparse with scipy.sparse.
  • Advanced BCs: Complex kinematic constraints handled via a QR-based condensation layer.

🧪 Illustrative Example: BCC Lattice Cell

This repository provides a complete demonstration of an as-manufactured digital twin reconstruction. The scripts are located in examples/geo_to_twin_one_cell/:

  1. make_geo.py: Generation of the as-designed 48-patch BCC B-spline assembly.
  2. fit_tomography.py: Rigid ICP pre-alignment and volVIC non-rigid surface fitting onto the cropped_CT_scan.tiff.
  3. compute_response.py: Linear elastic IGA simulation (compression-torsion) on the deformed geometry.

🎓 Citation & Authors

Authors: D. Bichet, J.C. Passieux, J.N. Périé, R. Bouclier

If you use this framework in your research, please cite:

  • The method:
@article{Bichet2025_method,
  title={Isogeometric multipatch surface fitting in tomographic images: Application to lattice structures},
  author={Bichet, D. and Passieux, J.C. and Périé, J.N. and Bouclier, R.},
  journal={Computer Methods in Applied Mechanics and Engineering},
  volume={436},
  pages={117729},
  year={2025},
  issn={0045-7825},
  doi={10.1016/j.cma.2025.117729},
  url={[https://doi.org/10.1016/j.cma.2025.117729](https://doi.org/10.1016/j.cma.2025.117729)}
}
  • The software:
@article{Bichet2026_software,
  title={image2iga: A JIT-Compiled Python Framework for Image-to-Isogeometric Analysis via Spline Template Fitting},
  author={Bichet, Dorian and Passieux, J-C. and Périé, J-N. and Bouclier, R.},
  year={2026},
  journal={Submitted}
}

📄 License

This software is governed by the CeCILL v2.1 license under French law and abiding by the rules of distribution of free software. You can use, modify and/or redistribute the software under the terms of the CeCILL license as circulated by CEA, CNRS and INRIA at the following URL: http://www.cecill.info.