AymanAmeen/SimEIT-dataset

SimEIT: A Scalable Simulation Framework for Generating Large-Scale Electrical Impedance Tomography Datasets

Authors

• Ayman A. Ameen¹, Franziska Mathis-Ullrich¹, Bernhard Kainz^1,2

¹Friedrich-Alexander University Erlangen-Nürnberg
²Imperial College London

This repository contains two large-scale synthetic EIT (Electrical Impedance Tomography) datasets generated using the SimEIT framework, a scalable simulation platform for creating physically consistent, AI-ready training data.

Abstract

Electrical Impedance Tomography (EIT) offers advantages over conventional imaging methods, such as X-ray and MRI, but suffers from an ill-posed inverse problem. Deep learning can alleviate this challenge, yet progress is limited by the lack of large, diverse, and reproducible datasets. We present SimEIT, a scalable framework for deterministic simulation and generation of synthetic EIT data. SimEIT enables high-throughput creation of diverse geometries and conductivity maps using parallelized finite element simulations, reproducible seeding, and automated validation. The framework provides multi-resolution, AI-ready HDF5 outputs with PyTorch integration. Demonstrated on two datasets exceeding 100,000 samples, SimEIT bridges the gap between physical simulation and AI training, supporting reliable benchmarking and development of advanced reconstruction algorithms.

Overview

Electrical Impedance Tomography (EIT) offers advantages over conventional imaging methods, such as X-ray and MRI, but suffers from an ill-posed inverse problem. Deep learning can alleviate this challenge, yet progress is limited by the lack of large, diverse, and reproducible datasets.

To address this critical gap, we introduce SimEIT: an open-source, parallelized framework for generating large-scale, physically consistent EIT datasets. Built on the validated EIDORS engine, SimEIT integrates flexibility, reproducibility, and scalability through several key innovations:

• Modular Framework & Parallel Processing: Flexible architecture with interchangeable components enables parallel execution in geometry generation and simulation stages, overcoming scalability bottlenecks.
• Geometry-Boundary Flexibility: Parametric customization of inclusion shapes (e.g., circles, ellipses, triangles), conductivity distributions, electrode placements, and domain boundaries (e.g., spherical substrates).
• Reproducible High-Throughput Synthesis: Deterministic seed control ensures batch-wise traceability for large-scale, physically accurate data generation.
• AI-Ready Data Optimization: Multi-resolution ground-truth maps (e.g., 256x256 to 32x32), differential outputs, and metadata-linked HDF5 storage with PyTorch integration.
• EIDORS-Based Physical Fidelity: Maintains physical consistency while supporting MATLAB and open-source Octave environments.
• Open Ecosystem & Visualization: Public codebase, Hugging Face demos, configurable noise models, and visualization tools enable community-driven expansion and validation.

By democratizing large-scale EIT data synthesis, SimEIT accelerates inverse solver development, enables systematic study of ill-posedness origins, and establishes a foundation for reproducible AI advancements in the field, launching the realization of the EIT promise across medical, industrial, and scientific domains.

Datasets

1. FourObjects Dataset

This dataset contains EIT simulations with up to four different object types and geometries.

Structure:

• dataset.h5 - HDF5 dataset file containing voltage measurements and ground-truth conductivity maps.
• dataset.zip - Compressed archive containing dataset.h5
• csv_dataset.csv - Main dataset metadata and parameters. The columns include:
  • index1: Sample index.
  • number_of_objects: Number of inclusions in the sample (from 1 to 4).
  • type_[1-4]: Geometric type of each inclusion (e.g., circle, ellipse, triangle, rectangle).
  • conductivity_[1-4]: Conductivity value for each inclusion.
  • features_[1-28]: Geometric parameters defining the position, size, and orientation of each inclusion.
  • coverage_area: The proportion of the domain area covered by the inclusions.
• metadata/ - Auxiliary data and reference files
  • masks_and_nodes/ - Multi-resolution masks and node information (32×32, 64×64, 128×128, 256×256)
  • volt/16/ - Voltage measurements and homogeneous reference data
• parameters/ - Train/validation/test split definitions

2. CirclesOnly Dataset

This dataset contains EIT simulations with circular inclusion geometries.

Structure:

• dataset.h5 - HDF5 dataset file containing voltage measurements and ground-truth conductivity maps.
• dataset.zip - Compressed archive containing dataset.h5
• csv_dataset.csv - Main dataset metadata and parameters. The columns include:
  • index1: Sample index.
  • number_of_objects: Number of circular inclusions in the sample.
  • type_[1-4]: Geometric type of each inclusion (always circle for this dataset).
  • conductivity_[1-4]: Conductivity value for each inclusion.
  • features_[1-28]: Geometric parameters defining the position and size of each circular inclusion.
  • coverage_area: The proportion of the domain area covered by the inclusions.
• metadata/ - Auxiliary data and reference files
  • masks_and_nodes/ - Multi-resolution masks and node information (32×32, 64×64, 128×128, 256×256)
  • volt/16/ - Voltage measurements and homogeneous reference data
• parameters/ - Train/validation/test split definitions

Data Format

Multi-Resolution Support

Each sample includes ground-truth conductivity maps at multiple resolutions:

• 256×256 pixels (high resolution)
• 128×128 pixels
• 64×64 pixels
• 32×32 pixels (low resolution)

File Formats

• HDF5: AI-ready format with PyTorch integration
• CSV files: Metadata and sample parameters
• MAT files: MATLAB-compatible data structures for masks and voltage measurements
• XLS files: Mask definitions at various resolutions

Dataset Statistics and Samples

The statistical properties of the generated dataset, summarized in the figures below, confirm its controlled diversity. The number of inclusions per sample is uniformly distributed from one to four, ensuring balanced complexity. The distribution of shapes is varied, with rectangles and circles being most frequent. Inclusion conductivities follow a near-uniform logarithmic distribution, offering a wider range than typical phantoms. The fractional area covered by objects is right-skewed, prioritizing samples with lower object density while still including high-density cases.

Figure: Statistical analysis of the dataset showing (a) number of inclusions distribution, (b) shape type frequencies, (c) conductivity value distributions, and (d) fractional area coverage.

Sample Geometries and Shapes

Our framework procedurally generates EIT datasets with morphological diversity. Each sample represents a circular domain containing randomized inclusions of circles, triangles, rectangles, and ovals with heterogeneous sizes, orientations, and spatial distributions. Conductivity values σ follow a logarithmic distribution spanning multiple orders of magnitude, creating challenging physical scenarios.

Figure: Example inclusion geometries and conductivity distributions. Each domain shows unique configurations of shapes with varying sizes, orientations, arrangements, and conductivity values (color-coded).

Multi-Resolution Dataset Structure

To accommodate different computational requirements, conductivity maps are generated at 256×256 pixel resolution and downsampled to 128×128, 64×64, and 32×32 resolutions. A PyTorch DataLoader integrates these datasets into deep learning workflows, while built-in circular masks exclude extraneous regions outside the EIT domain boundary, ensuring only relevant pixels are processed.

Figure: Multi-resolution conductivity maps showing the same sample at different pixel resolutions (256×256, 128×128, 64×64, and 32×32).

Dataset Statistics

The FourObjects dataset includes comprehensive statistical analysis:

• Conductivity distributions across multiple objects
• Coverage area statistics
• Number and types of objects per sample
• Detailed visualizations available in FourObjects/metadata/plot/

Data Splits

Pre-defined train/validation/test splits are provided in the parameters/ directory:

• train.txt - Training set indices
• val.txt - Validation set indices
• test.txt - Test set indices

---

BibTeX

@article{ameen2025simeit,
  title={SimEIT: A Scalable Simulation Framework for Generating Large-Scale Electrical Impedance Tomography Datasets},
  author={Ameen, Ayman A. and Mathis-Ullrich, Franziska and Kainz, Bernhard},
  year={2025},
}