PyEIDORS v1.1.0 - SoftwareX Paper Snapshot

PyEIDORS v1.1.0 — Paper Snapshot Release

This release corresponds to the fixed code version accompanying the SoftwareX manuscript:

PyEIDORS: An open-source Python toolkit for EIT forward modelling and reconstruction of simulated and experimental data

Paper snapshot commit: dde7124e8500223e919e5ba27f8bee205a52d3db

What's included

• Complete Electrode Model (CEM) forward solver based on FEniCS with P1/DG0 discretization
• Inverse solvers: difference reconstruction, absolute Gauss–Newton, sparse Bayesian (CUQI), and native BSBL
• Regularization: NOSER, Tikhonov, smoothness, and TV helper class
• GPU acceleration via PyTorch for inverse/tensor operations
• EIDORS interoperability: bidirectional mesh and data exchange with MATLAB/EIDORS
• Benchmark suite: reproducible runtime, fairness, and cross-framework comparison scripts
• Experimental data: saline-tank measurements (single and multi-condition) and maize-stalk feasibility case

Supplementary material

The attached supplementary_material.pdf contains:

• S1: Runtime benchmarks (measurement-space speedup, CPU/GPU comparisons)
• S2: Fairness controls (same-geometry cross-validation, tri-framework comparisons, thorax cross-section validation with Borsic et al. 2001 mesh, public Montreal DEMO3 tank data validation)
• S3: Multi-condition saline-tank evidence and generalization controls (mesh resolution, domain geometry, electrode configurations, contact impedance)
• S4: Implementation scope and CEM numerical details
• S5: Fixed-version information, data availability, and 3D prototype demonstration

Hardware environment

All benchmarks were conducted on: Intel Core i7-12700 (12 cores), 32 GB RAM, NVIDIA GeForce RTX 4070 (12 GB VRAM), CUDA 12.8, PyTorch 2.7.1. All computations in float64.

License

MIT License