EIT reconstruction parameters’ impact on ventilation monitoring features

Abstract

Electrical Impedance Tomography (EIT) enables continuous, non-invasive monitoring of regional pulmonary ventilation; however, the influence of reconstruction parameters on clinically derived features remains poorly understood. In this study, thoracic EIT recordings from 31 subjects were reconstructed using the GREIT framework in pyEIT; five parameters were modulated at three perturbation levels (±10%, ±25%, ±50%). Nineteen ventilation features were extracted per configuration and their sensitivity was quantified via a normalised influence score. The sigmoid threshold ratio dominated all other parameters, while regularisation strength and background permittivity yielded identical effects, suggesting a shared computational pathway within the reconstruction matrix. These findings highlight the importance of standardising reconstruction settings and explicitly reporting reconstruction configurations in clinical and research applications of EIT-based ventilation monitoring.