A Moment-Based Weighted Block Sparse Bayesian Learning Approach for Simultaneous Dual-Layer Lung Electrical Impedance Tomography

• Published in: Proceedings (International Symposium on Biomedical Imaging) pp. 1 - 5
• Main Authors: Dimas, Christos, Alimisis, Vassilis, Uzunoglu, Nikolaos, Sotiriadis, Paul P.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 27.05.2024
• Subjects: Bayes methods; Complexity theory; Electrical impedance tomography; electrode layer; Lung; moment; Real-time systems; sparse Bayesian learning; Spatial resolution; Three-dimensional displays
• ISSN: 1945-8452
• DOI: 10.1109/ISBI56570.2024.10635547

Electrical impedance tomography (EIT) has emerged as a non-invasive, fast, and safe medical technique for real-time thoracic imaging. However, EIT is typically conducted in a 2D fashion, often lacking crucial structural information associated with the z-axis direction. Furthermore, 3D EIT incurs high computational costs, rendering it impractical for real-time monitoring, while EIT reconstruction is an ill-posed and nonlinear problem. In this paper, a weighted block sparse Bayesian learning and an efficient method-of-moment approach are combined to simultaneously conduct lung EIT in 2 distinct z-planes, each one defined by 16 electrodes. Therefore, the inverse problem's non-linearity is reduced, robustness to noise and modeling errors is improved, while z-axis information is obtained, avoiding the 3D case complexity. Reconstructions based on simulated human thoracic structures and on horse subject thoracic data demonstrate improvement spatial resolution with limited presence of artefacts compared to the linear method-of-moment reconstruction.