Layer stripping: a direct numerical method for impedance imaging

Inverse problems for PDEs (35R30) Numerical methods for ill-posed problems for initial value and initial-boundary value problems involving PDEs (65M30) Motion of charged particles (78A35) Biomedical imaging and signal processing (92C55)

Review text

The authors consider the Neumann problem \(\nabla\cdot\sigma\nabla u=0\) in \(\Omega\), \(\sigma(\partial u/\partial N)=h\) on \(\partial\Omega\) assuming the integrals of \(u\), \(h\) over \(\partial\Omega\) are zero. Here \(o\in C^ 3(\overline{\Omega})\) and \(\Omega\) is a bounded smooth domain in \(\mathbb{R}^ 2\) or \(\mathbb{R}^ 3\). Then the map \(\Lambda: h\to u\) on \(\partial\Omega\) is well-defined. The subject of this paper is a natural and promising algorithm for numerical reconstruction of \(\sigma\) from \(\Lambda\). The authors derive a formula for reconstruction of \(\sigma\) at \(\partial\Omega\) and consider the example when \(\Omega\) is a disk. They obtain a Riccati type (operator) equation for \(\Lambda(r)\) at \(\partial\Omega_ r=\{| x|<r\}\) and consider a rotationally invariant \(\sigma\) in more detail. In section 2 they describe the algorithm which consists of 1) subdivision of \([0,1]\) into subintervals \([R_{j-1},R_ j]\), 2) calculation of \(\sigma=\sigma_ j\) on \(\partial\Omega_{R_ j}\), 3) recalculation of \(\Lambda\) from \(\partial\Omega_{R_ j}\) onto \(\partial\Omega_{R_{j-1}}\) assuming \(\sigma=\sigma_ j\) when \(R_{j-1}<| x|<R_ j\) (Cauchy problem, for an elliptic equation). No convergence proofs and only few numerical examples are given. For close ideas we refer to \textit{E. Curtis} and \textit{J. Morrow} [SIAM J. Appl. Math. 50, No. 3, 918-930 (1990; Zbl 0717.35092)].

Recommendations

Cited in

(26)

This page was built for publication: Layer stripping: a direct numerical method for impedance imaging

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3987867)