Contrast-based detectability of structural gaps and hollow defects using 2D and 3D electrical impedance tomography: a finite element evaluation

Electrical Impedance Tomography (EIT) is a non-invasive and low-cost imaging modality with increasing potential for nondestructive evaluation (NDE) of structural discontinuities in conductive media. However, due to the inherently diffusive nature of the inverse problem, realistic assessment of defect detectability requires physics-consistent forward modeling and quantitative validation. In this study, a finite element method (FEM)-based framework under the Complete Electrode Model (CEM) is employed to investigate the detectability of narrow structural discontinuities and hollow inclusion using both two-dimensional (2D) and three-dimensional (3D) EIT reconstructions. A 16-electrode configuration was implemented, yielding 120 independent voltage measurements per frame in a conductive saline phantom. Conductivity distributions were reconstructed using Gauss-Newton optimization with Tikhonov Regularization. Performance was quantitatively evaluated using contrast-to-noise ratio (CNR), peak signal-to-noise ratio (PSNR). structural similarity index (SSIM), area deviation, and computational time. Results indicate that defect detectability strongly depends on conductivity contrast and geometric size. Three-dimensional reconstructions demonstrate improved volumetric localization and reduced boundary artifacts compared to 2D models, while 2D reconstruction offers significantly lower computational time & cost suitable for rapid screening applications. The results also indicate that three-dimensional EIT reconstructions significantly improve the localization of narrow structural gaps and hollow inclusions compared to two-dimensional reconstructions, particularly in conductive media such as saline considering both error margins and the visibility quality. Hollow inclusion was successfully identified in both 2D and 3D models, with 3D reconstructions providing improved internal cavity delineation. These findings provide a quantitative assessment of the trade-offs between reconstruction fidelity and computational efficiency in EIT-based structural inspection, supporting its potential application in structural health monitoring, material characterization, and conductive-environment NDE systems.