BREIT
Brain Stroke Reconstruction using Multi-Frequency 3D EIT
BREIT: A Framework for Brain Stroke Reconstruction using Multi-Frequency 3D EIT
BREIT is a modular and reproducible framework for 3D Multi-Frequency Electrical Impedance Tomography (MF-EIT) stroke reconstruction. EIT is a non-invasive, low-cost, and portable modality that reconstructs electrical property distributions from boundary voltages — making it a promising tool for continuous, bedside stroke monitoring where CT/MRI are unavailable.
Framework Overview
BREIT unifies three core components into a single configurable pipeline:
- Neuroimaging-to-EIT pipeline — converts CT/MRI scans into frequency-dependent ground-truth admittivity volumes, supporting both ischemic (DWI/ADC/MRI) and hemorrhagic (CT) stroke types, registered to MNI152 space.
- 3D CEM forward solver — a self-contained Python implementation of the 3D Complete Electrode Model (CEM) for simulating MF-EIT voltages across 17 frequencies (5–2000 Hz), validated against EIDORS on the public UCLH stroke dataset.
- 3D D-bar reconstruction — supports non-uniform electrode layouts via spherical Voronoi surface-area weighting, providing classical baselines for benchmarking.
Code
The codebase is currently under review. Code coming soon at BREIT GitHub Repository.