Improved Electrical Impedance Tomography Reconstruction via a Bayesian Approach With an Anatomical Statistical Shape...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computational Anatomy

Background:

• Electrical Impedance Tomography (EIT) offers rapid, continuous lung function monitoring.
• Accurate EIT ventilation reconstruction necessitates patient-specific anatomical data, which is often unavailable.
• Current EIT methods face limitations in spatial fidelity.

Purpose of the Study:

• Develop a statistical shape model (SSM) for torso and lung geometry.
• Evaluate if patient-specific shape predictions from the SSM can improve EIT reconstructions.
• Enhance EIT accuracy and reliability in a Bayesian framework.

Main Methods:

• Generated an SSM from CT data of 81 participants using principal component analysis and regression.
• Integrated SSM-derived shapes into a Bayesian EIT framework.
• Quantitatively compared SSM-enhanced reconstructions against generic methods.

Main Results:

• Five principal shape modes captured 38% of lung and torso geometry variance.
• Nine anthropometric and pulmonary function metrics significantly predicted shape modes.
• SSM integration improved EIT reconstruction accuracy and reliability, reducing errors.

Conclusions:

• Bayesian EIT with SSMs provides more reliable quantitative and visual ventilation data.
• Patient-specific structural information did not conclusively outperform the mean SSM shape.
• The Bayesian framework advances accurate and reliable EIT-based ventilation monitoring.