Functional validation and comparison framework for EIT lung imaging.

Area of Science:

• Pulmonary physiology
• Medical imaging
• Biomedical engineering

Background:

• Electrical impedance tomography (EIT) is an emerging clinical tool for monitoring ventilation distribution in mechanically ventilated patients.
• Numerous image reconstruction algorithms exist for EIT, necessitating a framework for their assessment.
• Physiological changes in lung air content were experimentally induced in pigs to evaluate algorithm performance.

Purpose of the Study:

• To establish an experimental framework for assessing EIT image reconstruction algorithms.
• To compare the ability of different algorithms to represent clinically relevant ventilation changes.
• To evaluate the performance of twelve 2D EIT reconstruction algorithms.

Main Methods:

• An experimental framework was developed using 8 pigs with controlled ventilator settings (tidal volume, PEEP, FiO2).
• Twelve 2D EIT reconstruction algorithms were compared, including backprojection, GREIT, TSVD, Gauss-Newton variants, and iterative methods.
• Factors such as 3D finite element models, non-uniform conductivity, noise, electrode movement, TV reconstruction, robust norms, smoothing priors, and data types were considered.

Main Results:

• Clinically relevant parameters showed minimal variation among advanced EIT algorithms, despite differences in image appearance.
• Several advanced algorithms performed well, while others demonstrated significantly poorer performance.
• The traditional backprojection algorithm performed surprisingly well, validating previous lung EIT studies.

Conclusions:

• Advanced EIT algorithms exhibit comparable performance for key clinical parameters in ventilation monitoring.
• Algorithm selection impacts performance, with some advanced methods outperforming others.
• Backprojection remains a valid and effective algorithm for lung EIT applications.