pith. sign in

arxiv: 2312.05059 · v3 · pith:SVERQ63Pnew · submitted 2023-12-08 · 🧮 math.NA · cs.NA· eess.SP

The Kernel Method for Electrical Resistance Tomography

Antonello Tamburrino , Vincenzo Mottola This is my paper

Pith reviewed 2026-05-24 05:20 UTC · model grok-4.3

classification 🧮 math.NA cs.NAeess.SP
keywords electrical resistance tomographykernel methodinverse obstacle problemnon-iterative reconstructionconductivity anomaliesboundary measurementspower densityNeumann data
0
0 comments X

The pith

The Kernel Method recovers arbitrary conductivity anomalies by finding boundary currents that produce identical measurements with or without them.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents the Kernel Method, a non-iterative technique for the inverse obstacle problem in Electrical Resistance Tomography. It assumes a known background conductivity and aims to locate one or more anomalies of arbitrary shape, topology, and size. The approach rests on locating special boundary current densities that generate the same measurements whether the anomaly is present or absent. When these currents are applied solely to the background material, the power density vanishes inside the anomaly region, which reveals its location. The method requires only forward solves on the known background and therefore supports low-cost, real-time imaging.

Core claim

If a boundary current density exists such that it produces the same measurements in the presence and absence of the anomaly, then that current, when applied to the background conductivity alone, produces a power density that vanishes exactly inside the region occupied by the anomaly. The Kernel Method uses this property to retrieve the anomalies.

What carries the argument

Identification of Neumann boundary data that yield identical boundary measurements with and without the anomaly, causing the power density computed on the background conductivity to vanish inside the anomaly region.

If this is right

  • Multiple anomalies of arbitrary shape and topology can be recovered without iteration.
  • Reconstruction reduces to forward solves on the known background conductivity.
  • Numerical implementation is simple and requires very low computational cost.
  • The approach is positioned for real-time applications in electrical resistance tomography.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same vanishing-power-density principle might be tested in related inverse problems such as acoustic or thermal tomography where an inclusion can be rendered invisible to selected excitations.
  • Accurate prior knowledge of the background conductivity is required for the method to locate the anomaly correctly.
  • Numerical experiments with added measurement noise would indicate how robust the search for the special currents remains in practice.
  • The method could be extended by combining multiple such currents to improve resolution of closely spaced anomalies.

Load-bearing premise

There exists a boundary current density that produces identical measurements with and without the anomaly and that produces power density vanishing exactly inside the anomaly when applied to the background conductivity.

What would settle it

For a known test anomaly, search numerically for a boundary current that exactly matches the measurements obtained with and without the anomaly, then verify whether the power density vanishes throughout the anomaly region when the same current is applied to the background model alone.

Figures

Figures reproduced from arXiv: 2312.05059 by Antonello Tamburrino, Vincenzo Mottola.

Figure 1
Figure 1. Figure 1: Perturbation of the eigenvalues of the key operator ∆Λ. The domain Ω is a circle centered in the origin with radius 10 cm, while the anomaly D is a circle centered in the origin and radius 4 cm. The electrical conductivities are σbg “ 200 S m and σa “ 1 S m. The norm of the operator N is δ. of the perturbed operator are plateauing when their amplitude is of the order of δ. It is remarkable how they differ … view at source ↗
Figure 2
Figure 2. Figure 2: The Kernel Method is applied to retrieve the unknown anomaly D. For a given eigenfunction gn applied as Neumann boundary data to the reference configuration pσpxq “ σbg in Ωq), the corresponding solution u n bg of (2.1) is u n bgpr, θq “ R n ? πσbg ´ r R ¯n cospnθq, with a related power density given by pnpr, θq “ σbg |∇ubg| 2 “ 1 σbgπ ´ r R ¯2n´2 . (6.2) As already pointed out in Section 5, the unknown an… view at source ↗
Figure 3
Figure 3. Figure 3: Behavior of the reconstructed internal radius ˜ri , with respect to the order of the selected eigenfunction. is depicted. As it can be seen, the reconstructed radius tends to the actual one for n Ñ `8. 6.2. Holed circle. This example aims to clearly show one of the limits of the proposed method, which is related to the underlying physical model of the problem. Consider the configuration depicted in [PITH_… view at source ↗
Figure 4
Figure 4. Figure 4: The Kernel Method is applied to retrieve the circular crown r1 ď r ď r2. of variables leads to gnpθq “ 1 ? π cospnθq λn “ r3 nσbg 2 ` r2 r3 ˘n “` r1 r2 ˘n ´ ` r2 r1 ˘n ‰ pσ 2 bg ´ σ 2 d q dn dn “ ˆ r1 r2 ˙n „ˆr3 r2 ˙n pσd ´ σbgq 2 ` ˆ r2 r3 ˙n pσ 2 d ´ σ 2 bgq ȷ ´ ˆ r2 r1 ˙n „ˆr3 r2 ˙n pσd ` σbgq 2 ` ˆ r2 r3 ˙n pσ 2 d ´ σ 2 bgq ȷ . As for the previous example, the Kernel Method requires one to drive the re… view at source ↗
Figure 5
Figure 5. Figure 5: Behavior of the reconstructed outer radius ˜r2 of the cir￾cular crown, with respect to the order of the selected eigenfunction. which tends to r2 as n Ñ 8. 7. Numerical Examples This last section is dedicated to showing the effectiveness of the proposed method in some examples of application. Specifically, the aim is to determine the shape, position and dimension of one or more anomalies less conductive th… view at source ↗
Figure 6
Figure 6. Figure 6: Reconstructions for η “ 0 (within the machine pre￾cision). In black the reconstructed anomaly D˜, while in red the boundary of the actual anomaly D [PITH_FULL_IMAGE:figures/full_fig_p017_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Reconstructions for η “ 10´3 . In black the recon￾structed anomaly D˜, while in red the boundary of the actual anom￾aly D [PITH_FULL_IMAGE:figures/full_fig_p018_7.png] view at source ↗
read the original abstract

This paper treats the inverse problem of retrieving the electrical conductivity of a material starting from boundary measurements in the framework of Electrical Resistance Tomography (ERT). In particular, the focus is on non-iterative reconstruction methods suitable for real-time applications. In this work, the Kernel Method, a new non-iterative reconstruction method for Electrical Resistance Tomography, is presented. The imaging algorithm addresses the problem of retrieving one or more anomalies of arbitrary shape, topology, and size embedded in a known background (the inverse obstacle problem). The foundation of the Kernel Method is based on the idea that if a proper current density applied at the boundary (Neumann data) of the domain exists such that it is able to produce the same measurements with and without the anomaly, then this boundary source produces a power density that vanishes in the region occupied by the anomaly, when applied to the problem involving the background material only. This new tomographic method has a simple numerical implementation that requires a very low computational cost. In this paper, the theoretical foundation of the Kernel Method is provided, and an extensive numerical campaign proves the effectiveness of this new imaging method.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper introduces the Kernel Method, a non-iterative reconstruction technique for the inverse obstacle problem in Electrical Resistance Tomography. It claims that for anomalies of arbitrary shape, topology and size embedded in a known background conductivity, there exist boundary current densities (Neumann data) g such that the boundary measurements coincide with and without the anomaly; when this g is applied to the background problem alone, the resulting power density vanishes inside the anomaly region. The method is presented with a theoretical foundation, a simple numerical implementation of low computational cost, and validation via an extensive numerical campaign.

Significance. If the existence of such non-trivial g and the exact vanishing property held, the approach would supply an efficient, direct (non-iterative) solver for real-time ERT imaging of arbitrary anomalies. The numerical experiments are described as demonstrating practical effectiveness, which would be a useful contribution if the underlying construction were valid.

major comments (2)
  1. [Abstract / theoretical foundation] Abstract and theoretical foundation: The central construction assumes the existence of a non-trivial boundary current density g such that the Dirichlet-to-Neumann maps satisfy DN(g) = DN0(g) and the power density ∫ σ0 |∇u0|^2 vanishes exactly on a positive-measure anomaly region D when the background conductivity σ0 is used. For the insulating case (σ=0 in D) this requires ∇u0 ≡ 0 on an open set inside D; unique continuation for the elliptic equation with regular σ0 then forces u0 constant throughout Ω, hence g = 0. The same obstruction applies, with only minor modifications, to finite-contrast anomalies. This directly falsifies the existence assumption on which the kernel construction rests.
  2. [Abstract] Abstract: The claim that the method 'retrieves one or more anomalies of arbitrary shape, topology, and size' is load-bearing on the above existence result. No counter-example or additional regularity assumption on σ0 or on the anomaly contrast is supplied that would evade unique continuation, so the theoretical foundation cannot support the stated generality.
minor comments (1)
  1. [Abstract] The abstract states that an 'extensive numerical campaign proves the effectiveness' but supplies no error tables, mesh sizes, or quantitative metrics; these details would be needed even if the theoretical issue were resolved.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thorough review and for identifying the key mathematical issue with the proposed Kernel Method. The comments correctly highlight that the central existence assumption is incompatible with unique continuation for elliptic equations. We address each point below and indicate that we will revise the manuscript to remove the unsupported claims.

read point-by-point responses

  1. Referee: [Abstract / theoretical foundation] Abstract and theoretical foundation: The central construction assumes the existence of a non-trivial boundary current density g such that the Dirichlet-to-Neumann maps satisfy DN(g) = DN0(g) and the power density ∫ σ0 |∇u0|^2 vanishes exactly on a positive-measure anomaly region D when the background conductivity σ0 is used. For the insulating case (σ=0 in D) this requires ∇u0 ≡ 0 on an open set inside D; unique continuation for the elliptic equation with regular σ0 then forces u0 constant throughout Ω, hence g = 0. The same obstruction applies, with only minor modifications, to finite-contrast anomalies. This directly falsifies the existence assumption on which the kernel construction rests.

    Authors: We agree with the referee's analysis. The unique continuation principle for solutions to elliptic equations with sufficiently regular coefficients indeed implies that if the power density vanishes on a set of positive measure inside an open anomaly region D (hence ∇u0 = 0 on an open subset), then u0 must be constant throughout the domain, forcing g = 0. Our construction did not include any counter-example or modified regularity assumptions that would evade this obstruction, and none exist under the standard hypotheses of the inverse obstacle problem in ERT. Consequently, non-trivial g satisfying both DN(g) = DN0(g) and exact vanishing of the power density inside D cannot exist. This falsifies the theoretical foundation as presented. We will revise the manuscript to explicitly acknowledge this limitation, withdraw the general claims, and restrict or remove the method's description accordingly. revision: yes

  2. Referee: [Abstract] Abstract: The claim that the method 'retrieves one or more anomalies of arbitrary shape, topology, and size' is load-bearing on the above existence result. No counter-example or additional regularity assumption on σ0 or on the anomaly contrast is supplied that would evade unique continuation, so the theoretical foundation cannot support the stated generality.

    Authors: We concur that the abstract claim of retrieving anomalies of arbitrary shape, topology, and size depends entirely on the existence of such non-trivial g, which is ruled out by unique continuation. No regularity assumptions or counter-examples were provided (or possible) to circumvent the obstruction. We will revise the abstract and all related statements in the manuscript to eliminate this unsupported generality. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation self-contained from stated physical principle

full rationale

The paper presents the Kernel Method as derived directly from the physical principle that a suitable boundary current density producing identical measurements with and without the anomaly will yield vanishing power density inside the anomaly region when applied to the background conductivity alone. This foundation is stated explicitly in the abstract and introduction without any parameter fitting to data subsets, self-referential definitions, or load-bearing self-citations. The numerical implementation follows from this principle as an independent construction, with no reduction of the central claim to its inputs by construction. The derivation chain remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no explicit free parameters, axioms, or invented entities can be extracted. The method implicitly assumes a known background conductivity and the existence of suitable boundary currents, but these are not quantified.

pith-pipeline@v0.9.0 · 5723 in / 1108 out tokens · 19459 ms · 2026-05-24T05:20:23.634547+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

32 extracted references · 32 canonical work pages

  1. [1]

    Cheng J, Liu S and Li Y 2019 Design and optimization of liquid level sensor based on elec- trical tomography 2019 IEEE International Conference on Imaging Systems and Techniques (IST) pp 1–6

    work page 2019

  2. [2]

    Durdevic P, Hansen L, Mai C, Pedersen S and Yang Z 2015 IF AC-PapersOnLine48 147– 153 ISSN 2405-8963 2nd IFAC Workshop on Automatic Control in Offshore Oil and Gas Production OOGP 2015

    work page 2015

  3. [3]

    Mann R 2009 Ert imaging and linkage to cfd for stirred vessels in the chemical process industry 2009 IEEE International Workshop on Imaging Systems and Techniques pp 218– 222

    work page 2009

  4. [4]

    Holder D 1994 Thorax 49 626–626

    work page 1994

  5. [5]

    Tarassenko L and Rolfe P 1984 Electron. Lett. 20 574

    work page 1984

  6. [6]

    34 543–550

    Meier T, Luepschen H, Karsten J, Leibecke T, Grossherr M, Gehring H and Leonhardt S 2008 Intensive Care Med. 34 543–550

    work page 2008

  7. [7]

    Adler A, Berthiaume Y, Guardo R and Amyot R 1995 Proceedings of 17th International Conference of the Engineering in Medicine and Biology Society 1 557–558 vol.1 URL https: //api.semanticscholar.org/CorpusID:61366527

    work page 1995

  8. [8]

    Conway J 1987 Clin. Phys. Physiol. Meas. 8 141–146

    work page 1987

  9. [9]

    Pham T M T, Yuill M, Dakin C and Schibler A 2011 Eur. Respir. J. 37 919–924

    work page 2011

  10. [10]

    Zain N M and Chelliah K K 2014 Asian Pac. J. Cancer Prev. 15 1327–1331

    work page 2014

  11. [11]

    Holder D S, Rao A and Hanquan Y 1996 Physiol. Meas. 17 A179–A186

    work page 1996

  12. [12]

    Kruger M V P 2003 Tomography as a metrology technique for semiconductor manufacturing Tech. Rep. UCB/ERL M03/11 EECS Department, University of California, Berkeley

    work page 2003

  13. [13]

    Jordana J, Gasulla M and Pall` as-Areny R 2001 Measurement Science and Technology 12 1061–1068

    work page 2001

  14. [14]

    Hou T C, Loh K J and Lynch J P 2007 Nanotechnology 18 315501

    work page 2007

  15. [15]

    Linderholm P, Marescot L, Loke M H and Renaud P 2008 IEEE Transactions on Biomedical Engineering 55 138–146

    work page 2008

  16. [16]

    1088/0967-3334/25/1/021

    Lionheart W R B 2004 Physiological Measurement 25 125 URL https://dx.doi.org/10. 1088/0967-3334/25/1/021

    work page 2004

  17. [17]

    Colton D and Kirsch A 1996 Inverse Problems 12 383 URL https://dx.doi.org/10.1088/ 0266-5611/12/4/003

    work page 1996

  18. [18]

    Kirsch A 1998 Inverse Problems 14 1489

    work page 1998

  19. [19]

    Br¨ uhl M 2001SIAM Journal on Mathematical Analysis 32 1327–1341

    work page

  20. [20]

    org/10.1515/jiip.1999.7.3.255

    Ikehata M 1999 Journal of Inverse and Ill-posed Problems 7 255–272 URL https://doi. org/10.1515/jiip.1999.7.3.255

    work page doi:10.1515/jiip.1999.7.3.255 1999

  21. [21]

    Devaney A J Super-resolution processing of multi-static data using time reversal and music northeastern University preprint

    work page

  22. [22]

    Tamburrino A and Rubinacci G 2002 Inverse Problems 18 1809–1829

    work page 2002

  23. [23]

    Mottola V, Esposito A C, Faella L, Piscitelli G, Prakash R and Tamburrino A 2025 The inverse obstacle problem for nonlinear inclusions ( Preprint 2408.08040) URL https:// arxiv.org/abs/2408.08040

    work page arXiv 2025

  24. [24]

    Gisser D G, Isaacson D and Newell J C 1990 SIAM Journal on Applied Mathematics 50 1623–1634

    work page 1990

  25. [25]

    Gebauer B 2008 Inverse Problems & Imaging 2 251–269

    work page 2008

  26. [26]

    Harrach B and Ullrich M 2013 SIAM Journal on Mathematical Analysis 45 3382–3403

    work page 2013

  27. [27]

    Ikehata M 1998 Journal of Inverse and Ill-Posed Problems 6 127–140 THE KERNEL METHOD 21

    work page 1998

  28. [28]

    Soc.) pp 65–73

    Calderon A P 1980 On an inverse boundary Seminar on Numerical Analysis and its Appli- cations to Continuum Physics, Rio de Janeiro, 1980 (Brazilian Math. Soc.) pp 65–73

    work page 1980

  29. [29]

    Alessandrini G 2012 Journal of Mathematical Analysis and Applications 386 669–676 ISSN 0022-247X

    work page 2012

  30. [30]

    Garofalo N and Lin F 1986 Indiana University Mathematics Journal 35 245–268

    work page 1986

  31. [31]

    Rudin W 1991 Functional Analysis 2nd ed International series in pure and applied mathe- matics (McGraw-Hill) ISBN 0070542368; 9780070542365

    work page 1991

  32. [32]

    Kato T 1995 Perturbation theory for linear operators Classics in Mathematics (Springer) ISBN 9783540586616,354058661X

    work page 1995