[Paper Review] Admittivity imaging from multi-frequency micro-electrical impedance tomography

Habib Ammari, Laure Giovangigli+2 more|arXiv (Cornell University)|Mar 22, 2014

Electrical and Bioimpedance Tomography31 references3 citations

TL;DR

This paper proposes a stable and convergent optimal control-based algorithm for reconstructing both conductivity and permittivity distributions from multi-frequency micro-electrical impedance tomography (M-EIT) data. By formulating the admittivity imaging problem as a minimization of a functional derived from internal potential measurements across a frequency band, the authors prove convergence and stability of the Landweber-type iterative scheme under appropriate conditions on boundary measurements and operator regularity.

ABSTRACT

The aim of this paper is to propose an optimal control optimization algorithm for reconstructing admittivity distributions (i.e., both conductivity and permittivity) from multi-frequency micro-electrical impedance tomography. A convergent and stable optimization scheme is shown to be obtainable from multi-frequency data. The results of this paper have potential applicability in cancer imaging, cell culturing and differentiation, food sciences, and biotechnology.

Motivation & Objective

• To develop a robust and stable reconstruction method for admittivity distributions (conductivity and permittivity) from multi-frequency internal electrical measurements in micro-electrical impedance tomography.
• To address the ill-posedness of the inverse admittivity imaging problem by leveraging multi-frequency data and optimal control theory.
• To establish rigorous convergence and stability guarantees for the proposed iterative reconstruction scheme.
• To demonstrate the feasibility of high-resolution imaging of biological tissues using internal potential measurements at multiple frequencies.
• To provide a theoretical foundation for applying spectroscopic admittivity imaging in biomedical applications such as cancer detection and cell analysis.

Proposed method

• Formulates the admittivity imaging problem as a minimization of a functional over the space of conductivity and permittivity distributions, using multi-frequency internal potential measurements as data.
• Applies optimal control theory to derive the Fréchet derivative of the data misfit functional, enabling gradient-based minimization.
• Constructs an initial guess by solving a boundary value problem for the complex-valued potential governed by the frequency-dependent conductivity and permittivity.
• Implements a Landweber-type iterative scheme with a projection operator T to ensure iterates remain within the feasible set K, improving convergence and stability.
• Incorporates a cutoff function in the Landweber iteration to control the convergence rate and ensure the sequence remains bounded and convergent.
• Proves convergence of the iterative sequence to the true admittivity distribution under assumptions of Lipschitz continuity of the Fréchet derivative and a coercivity condition on the operator.

Experimental results

Research questions

• RQ1 Can a stable and convergent optimization algorithm be constructed for reconstructing both conductivity and permittivity from multi-frequency internal electrical measurements in micro-EIT?
• RQ2 What conditions on the boundary measurements ensure the well-posedness and stability of the admittivity reconstruction problem?
• RQ3 How can the ill-posedness of the inverse admittivity problem be mitigated using multi-frequency data and optimal control techniques?
• RQ4 What is the role of the initial guess and the projection operator T in ensuring convergence of the iterative scheme?
• RQ5 Under what regularity and coercivity conditions does the Landweber-type iteration converge to the true admittivity distribution?

Key findings

• The proposed optimal control-based iterative algorithm converges to the true admittivity distribution under suitable assumptions, including Lipschitz continuity of the Fréchet derivative and a coercivity condition on the forward operator.
• The convergence is guaranteed when the initial guess is sufficiently close to the true solution and the step size μ is small enough, ensuring stability of the iteration.
• The Landweber iteration with a cutoff function and projection operator T ensures that the iterates remain in the feasible set and converge to the solution at a controlled rate.
• The method achieves stability and convergence even though the logarithmic admittivity transport equation cannot be solved via the method of characteristics due to the complex-valued nature of the potential.
• The theoretical analysis confirms that the minimization functional is Fréchet differentiable and its derivative is explicitly computed, enabling gradient-based reconstruction.
• The framework is applicable to high-resolution imaging of thin biological layers using micro-electrode arrays, with potential applications in cancer imaging, cell culturing, and biotechnology.