Sparse reconstruction of log-conductivity in current density impedance tomography

Area of Science:

• Electrical Engineering
• Applied Mathematics
• Medical Imaging

Background:

• Electrical Impedance Tomography (EIT) reconstructs internal conductivity distributions from boundary measurements.
• Sparse reconstruction is crucial for identifying localized conductivity changes.
• Existing methods may struggle with edge enhancement and high-contrast imaging.

Purpose of the Study:

• To develop a new non-linear optimization approach for sparse reconstruction of log-conductivities in current density impedance imaging.
• To improve image quality, contrast, and resolution in EIT.
• To leverage L1 regularization and anisotropic diffusion for enhanced reconstruction.

Main Methods:

• A non-linear optimization framework minimizing an objective functional.
• Incorporation of a least squares fit for interior electric field data.
• Inclusion of L1 regularization for sparsity and Perona-Malik diffusion for edge enhancement.
• Solving an elliptic partial differential equation (PDE) relating conductivity and electric potential.

Main Results:

• The proposed method demonstrates superior image reconstructions of various log-conductivity patterns.
• Numerical experiments show effectiveness compared to existing methods.
• The L1 regularization and Perona-Malik diffusion effectively promote sparsity and enhance edges.

Conclusions:

• The new non-linear optimization approach provides effective sparse reconstruction of log-conductivities.
• The method offers improved contrast and resolution in EIT image reconstruction.
• This framework shows promise for advanced applications in electrical impedance imaging.