Pith. sign in

REVIEW 4 major objections 3 minor 17 references

Combining Deep Learning and 3D Contrast Source Inversion in MR-based Electrical Properties Tomography

T0 review · 4 major / 3 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Seeding 3D contrast source inversion with a deep-learning EPT map lowers whole-volume error at 3T and 7T and improves structure around the ventricles.

desk verdict A sensible hybrid MR-EPT idea with honest simulation results, but the single-subject evaluation and undisclosed training/test overlap for the Duke model leave the generalization claim under-supported. read the letter →

arxiv 1908.04542 v1 pith:WCWAXHRH submitted 2019-08-13 physics.med-ph eess.IV

Reijer L. Leijsen , Cornelis A.T. van den Berg , Andrew G. Webb , Rob F. Remis , Stefano Mandija This is my paper · ORCID
classification physics.med-pheess.IV
keywords electricalpropertiestomographyMR-EPTdeeplearningcontrastsourceinversionconductivityimagingpermittivityhybridreconstructionB1+mapping
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Magnetic resonance electrical properties tomography (MR-EPT) aims to map tissue conductivity and permittivity from the MRI transmit field, but each existing reconstruction route has a weak point: Helmholtz-based maps amplify noise, contrast source inversion is slow and depends on where it starts, and deep-learning methods need very large training sets. The paper tests whether starting 3D contrast source inversion from a deep-learning EPT reconstruction, rather than from a homogeneous mask or a Helmholtz map, combines the strengths of both. Using simulated head data at 3T and 7T, it finds that this hybrid gives lower whole-volume relative residual error than standard 3D contrast source inversion at both field strengths, and cleaner conductivity structure around the ventricles than deep-learning EPT alone. The reason matters clinically: a good initialization can shorten iterations and reduce the data burden for deep learning, moving EPT closer to routine use.

What carries the argument

The machinery is contrast source inversion, an iterative solver that reconstructs the electrical properties by minimizing a cost functional built from a contrast function and a contrast source, with a conjugate-gradient update; the paper's contribution is to seed it with a deep-learning EPT map instead of a homogeneous mask. The deep-learning map comes from a conditional generative adversarial network trained on simulated transmit-field data, providing a noisy but anatomically informed starting point; the inversion then re-fits the model to the measured field through Maxwell's equations, which the paper calls data consistency. The update runs for up to 500 iterations or until a tolerance of $10^{-5}$ is reached, and reconstructions are bounded to physiologically plausible ranges.

What would settle it

Retrain the deep-learning network on the same head-model population but with the specific test head removed, then rerun the hybrid and standard contrast source inversion reconstructions on that held-out head; if the whole-volume relative residual error of the hybrid stops being lower than standard contrast source inversion, the reported advantage depends on the test subject being seen during training.

Watch

Extended reading notes

Core claim

The central claim is that a two-step reconstruction, in which a deep-learning EPT map supplies the initial guess for a standard three-dimensional contrast source inversion, improves on either method used alone. On simulated head data at 3T and 7T, the hybrid's whole-volume relative residual error is lower than that of standard 3D contrast source inversion with the usual homogeneous initialization; the deep-learning initialization removes the low-electric-field artifacts that otherwise appear as artificial bands, and the inversion step enforces consistency with the measured field, improving the periventricular conductivity structure that deep-learning EPT alone blurs. Permittivity maps improve less than conductivity maps, and the gain at 7T is obtained even though the deep-learning network was only trained at 3T, because the inversion step supplies data consistency for the subject at hand.

Load-bearing premise

The evaluation assumes that the specific head model used for testing was not part of the training data for the deep-learning network; if it was, the deep-learning initialization is an in-distribution guess and the reported improvement over standard contrast source inversion would be optimistic.

Editorial extensions

If this is right

  • At both 3T and 7T, the hybrid cuts whole-volume relative residual error in conductivity and permittivity compared to standard 3D contrast source inversion, and removes the artificial-band artifacts caused by the homogeneous start.
  • The inversion step improves tissue structure around the ventricles relative to deep-learning EPT alone, so the hybrid addresses a main weakness of the deep-learning output.
  • Because the inversion enforces data consistency, the hybrid can work with noisy data at signal-to-noise ratio 100 with only minor degradation, unlike the Helmholtz-initialized hybrid.
  • A deep-learning network trained at 3T still benefits 7T reconstructions when followed by the inversion step, suggesting the data-consistency step can partly compensate for a mismatched training distribution.
  • The approach may reduce the need for exhaustive deep-learning training sets, since the inversion step adapts the reconstruction to the subject at hand.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • This inference goes beyond the paper: the reported comparison would be a fairer test of generalization if the specific head model used for evaluation were held out of the deep-learning training set; the paper does not state that it was, so part of the hybrid's advantage could come from an in-distribution initialization.
  • This inference goes beyond the paper: because the inversion step enforces data consistency, the same two-step recipe could be applied to other transmit coils or body regions whenever a deep-learning surrogate can supply an initial map, making the improvement a general strategy rather than a head-only fix.
  • This inference goes beyond the paper: a region-specific error metric for the ventricles would quantify the visible improvement in tissue structure, which the paper currently reports mainly qualitatively.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 3 minor

Summary. The manuscript proposes a hybrid MR-EPT approach in which standard Helmholtz-based MR-EPT or deep-learning EPT (DL-EPT) reconstructions are used as initialization guesses for standard 3D contrast source inversion EPT (CSI-EPT). The authors evaluate the resulting hybrid methods, denoted MR-CSI and DL-CSI, on realistic FDTD simulations of the Duke head model at 3 T and 7 T, with and without added noise at SNR=100, and compare them against standard CSI-EPT initialized with a homogeneous mask (H-CSI). The central quantitative claim is that DL-CSI yields lower whole-volume relative residual error than H-CSI at both field strengths (e.g., conductivity RRE 0.55 vs 0.52 at 3 T and 0.46 vs 0.43 at 7 T in Fig. 3) and improves the structure of conductivity reconstructions around the ventricles compared with DL-EPT alone. The paper concludes that the hybrid approach combines the power of data-driven DL-EPT with the data consistency provided by CSI-EPT, potentially improving generalization and reducing the need for exhaustive DL training sets.

Significance. If the result holds, the hybrid DL-CSI approach would be a practically useful way to improve CSI-EPT initialization while mitigating the generalization limitations of DL-based EPT, and the paper is clearly written in terms of the proposed comparison. The use of realistic electromagnetic simulations at two field strengths, quantitative whole-volume RRE metrics, and region-based mean/standard-deviation statistics is a strength, and the paper makes a falsifiable quantitative prediction about the ranking of reconstruction methods. However, the significance is currently limited by the evaluation being restricted to a single head model, a single noise level, and no independent validation of the DL-EPT component; the lack of a clear statement on training/test separation for the DL-EPT network is a specific concern that could bias the central comparison. No code, data, or trained network is shipped, which limits reproducibility. The core idea is promising, but the evidence presented is not yet sufficient to support the generalization claim made in the conclusion.

major comments (4)
  1. [Sec. 2.2.2] The DL-EPT network was trained on 1120 unique B1 fields obtained from 20 head models described as variations of the Duke and Ella models from the Virtual Family, and the only test subject used in this paper is the Duke head model. The manuscript nowhere states that the exact Duke simulation used for evaluation was excluded from the DL-EPT training set. If the Duke model or its simulated B1 fields were part of the training data, the DL-EPT output for Duke is an in-distribution prediction with near-ground-truth tissue structure, and the subsequent small RRE improvements reported for DL-CSI over H-CSI (Fig. 3: conductivity RRE 0.55 to 0.52 at 3 T and 0.46 to 0.43 at 7 T) may reflect starting the CSI iterations near the truth rather than a generalizable property of the hybrid method. The conclusion that DL-CSI 'facilitates a better generalization' is therefore not supported by the present single-subject demonstration. The authors should either explicitly disclose the training/test separation, or add a held-out subject and show that the improvement persists.
  2. [Sec. 3 / Figs. 2-3] The entire evaluation is based on one head model (Duke) and one noise condition (Gaussian noise leading to SNR=100), with no multiple noise realizations or uncertainty measures on the reported RRE values. Because the central claim is about improved generalization and noise robustness, the evidence is too limited: a single favorable subject and a single noise realization do not establish that DL-CSI will outperform H-CSI across subjects, coil configurations, or noise levels. Please either add multiple subjects/noise realizations with error bars, or substantially temper the generalization claim to a proof-of-concept statement.
  3. [Sec. 2.2.4 / Sec. 3] The 7 T DL-CSI reconstructions are initialized with DL-EPT maps obtained at 3 T, because the available DL network was only trained at 3 T. Since tissue electrical properties are frequency-dependent and the network was trained on 3 T data, the 7 T comparison is not a clean test of the hybrid concept at 7 T; the improvement over H-CSI at 7 T may partly reflect the frequency offset of the initialization rather than the data-consistency mechanism. The discussion should either include a 7 T-trained DL network or explicitly frame the 7 T DL-CSI results as a preliminary cross-frequency test with this confound acknowledged.
  4. [Sec. 2.2.2 and reproducibility] The DL-EPT network used as a central component of the proposed method is not described in this manuscript beyond a reference to prior work, and no code, trained weights, or data are provided. As a result, the quantitative comparisons in Figs. 1-3 cannot be reproduced or independently verified by other groups. The authors should state the availability of the network and code, or provide sufficient architectural and training details to allow replication, or clearly mark the DL-EPT outputs as a non-public black box whose use limits reproducibility.
minor comments (3)
  1. [Sec. 2.3, Eq. (3)] Equation (3) defines the relative residual error using the Euclidean norm over the complete domain of interest, but the text does not specify whether the domain excludes the coil region or background voxels; please state the exact mask used for the RRE computation.
  2. [Table S1] In the supplementary table, the DL-EPT rows for 7 T are marked with dashes, yet the DL-CSI 7 T results are obtained using 3 T DL-EPT maps as initialization; a brief note clarifying this inconsistency would help the reader.
  3. [Fig. 3] The subcaptions in Fig. 3 list RRE values but the color scale for permittivity error maps is not explicitly stated; adding the color-bar range to the figure caption would improve readability.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity; score 2 reflects minor same-group self-citation and a disclosure gap about training/test overlap.

full rationale

The paper's central comparison is empirical: DL-EPT and MR-EPT reconstructions are used as initializations for 3D CSI-EPT, and the resulting maps are scored against simulation ground truth via the RRE of Eq. (3). No parameter appearing in Eq. (1) or Eq. (2) is fitted to the ground-truth maps used for evaluation; the CSI iterations minimize a data-consistency functional (Eq. 2) with respect to measured B1+ fields only. The DL-EPT network (ref 13) is a prior published method from the same group, but its use here is as an off-the-shelf initializer, not as a derived conclusion; the improvement of DL-CSI over H-CSI is measured with held-out error metrics, not implied by construction. The one legitimate concern is that Section 2.2 states the DL-EPT network was trained on 1120 B1+ fields from 20 head models that are variations of Duke and Ella, while the test subject is Duke, and the paper never discloses whether the exact Duke model used for evaluation was excluded from training. If it was included, the DL-EPT map is an in-distribution recall rather than a prediction, making the generalization claim weaker. This is a missing-support/data-leakage issue, not a circular derivation: it does not make any equation equal to its own input. The self-citations to refs 12 and 13 are not load-bearing in a circular sense because both methods are independently published and the present contribution is their empirical combination.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The paper introduces no new physical entities. Its core assumptions are that the electromagnetic simulations are accurate, that the DL-EPT surrogate model transfers to the test subject and to 7T, and that CSI-EPT improves on its initialization. The homogeneous mask values for the H-CSI baseline are hand-chosen and affect the comparison.

free parameters (5)
  • Homogeneous mask conductivity (3T) = 0.53 S/m
    Used as initialization for the standard H-CSI baseline at 3T; chosen by hand as the average expected conductivity.
  • Homogeneous mask permittivity (3T) = 51
    Used as initialization for H-CSI at 3T.
  • Homogeneous mask conductivity (7T) = 0.59 S/m
    Used as initialization for H-CSI at 7T.
  • Homogeneous mask permittivity (7T) = 43
    Used as initialization for H-CSI at 7T.
  • Reconstruction constraint bounds = sigma in [0, 2.6] S/m, epsilon_r in [1,100]
    Applied to all final reconstructions; maxima are about 20% higher than the maximum values in the ground truth models.
assumptions (5)
  • domain assumption Maxwell's equations govern the electromagnetic fields, and the FDTD simulations accurately model the coil and head.
    The paper relies on XFDTD simulations as the ground truth and as the data source for reconstructions (Section 2.1).
  • standard math The Helmholtz equation (Eq. 1) correctly relates the Laplacian of B1+ to tissue conductivity and permittivity.
    Used by the MR-EPT initialization method (Section 2.2.1).
  • domain assumption The CSI-EPT cost functional (Eq. 2) and its conjugate-gradient minimization converge to the true electrical properties when given a good initialization.
    The central hybrid claim assumes that CSI-EPT improves the DL-EPT initialization rather than corrupting it (Section 2.2.3).
  • domain assumption The DL-EPT network trained on noisy 3T data generalizes to the test head model and, at 7T, provides a useful initialization despite the field-strength mismatch.
    The DL-CSI 7T results use a 3T-trained network as initialization, which is acknowledged as a limitation (Sections 2.2.2 and 2.2.4).
  • domain assumption The transceive phase can be used in place of the transmit phase for DL-EPT.
    The DL-EPT network was trained with transceive phase because transmit phase is not measurable in MRI (Section 2.2.2, citing ref 16).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Combining Deep Learning and 3D Contrast Source Inversion in MR-based Electrical Properties Tomography." pith.science (2026). https://pith.science/paper/WCWAXHRH

@misc{pith2026190804542,
  author       = {Pith},
  title        = {Pith review of: Combining Deep Learning and 3D Contrast Source Inversion in MR-based Electrical Properties Tomography},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WCWAXHRH}},
  note         = {Machine review of arXiv:1908.04542}
}
read the original abstract

Magnetic resonance-electrical properties tomography (MR-EPT) is a technique used to estimate the conductivity and permittivity of tissues from MR measurements of the transmit magnetic field. Different reconstruction methods are available, however all these methods present several limitations which hamper the clinical applicability. Standard Helmholtz based MR-EPT methods are severely affected by noise. Iterative reconstruction methods such as contrast source inversion-EPT (CSI-EPT) are typically time consuming and are dependent on their initialization. Deep learning (DL) based methods require a large amount of training data before sufficient generalization can be achieved. Here, we investigate the benefits achievable using a hybrid approach, i.e. using MR-EPT or DL-EPT as initialization guesses for standard 3D CSI-EPT. Using realistic electromagnetic simulations at 3 T and 7 T, the accuracy and precision of hybrid CSI reconstructions are compared to standard 3D CSI-EPT reconstructions. Our results indicate that a hybrid method consisting of an initial DL-EPT reconstruction followed by a 3D CSI-EPT reconstruction would be beneficial. DL-EPT combined with standard 3D CSI-EPT exploits the power of data driven DL-based EPT reconstructions while the subsequent CSI-EPT facilitates a better generalization by providing data consistency.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

17 extracted references · 17 canonical work pages

  1. [1]

    ,AGENDIJK*.(YPERTHERMIATREATMENTPLANNING.PhyS-EdBIol2000; 45(5):261

    work page

  2. [2]

    -2)-BASED ANATOMICAL MODEL OF THE HUMAN HEAD FOR SPECIfiCABSORPTIONRATEMAPPING.-EdBIolENgCoMP2008;46(12): 1239–1251

    -AKRIS.,ANGELONE,,4ULLOCH3,3ORG3,KAISER*,KENNEDYD,BON- MASSAR G. -2)-BASED ANATOMICAL MODEL OF THE HUMAN HEAD FOR SPECIfiCABSORPTIONRATEMAPPING.-EdBIolENgCoMP2008;46(12): 1239–1251

    work page

  3. [3]

    J -agNRESoNIMagINg2015;42(2):371–378

    3HIN*,KIM-,,EE*,.AM9,KIM-,CHOI.,KIM3,KIMD.)NITIAL STUDYONINVIVOCONDUCTIVITYMAPPINGOFBREASTCANCERUSING-2). J -agNRESoNIMagINg2015;42(2):371–378

    work page

  4. [4]

    CORRELA- TIONBETWEENCONDUCTIVITYANDPROGNOSTICFACTORSININVASIVEBREAST CANCERUSINGMAGNETICRESONANCEELECTRICPROPERTIESTOMOGRAPHY (-2%04).EuRRadIol2016;26(7):2317–2326

    KIM 3, 3HIN *, KIM D, KIM -, KIM %, -OON (, 9OON *. CORRELA- TIONBETWEENCONDUCTIVITYANDPROGNOSTICFACTORSININVASIVEBREAST CANCERUSINGMAGNETICRESONANCEELECTRICPROPERTIESTOMOGRAPHY (-2%04).EuRRadIol2016;26(7):2317–2326

    work page

  5. [5]

    2ECENT PROGRESS AND FUTURE CHAL- LENGESIN-2ELECTRICPROPERTIESTOMOGRAPHY

    KATSCHER 5, KIM D(, 3EO *K. 2ECENT PROGRESS AND FUTURE CHAL- LENGESIN-2ELECTRICPROPERTIESTOMOGRAPHY. CoMPuT-aTh-EThod -2013;2013:1–11

    work page 2013

  6. [6]

    :HANG8,,IU*,(EB.-AGNETIC-RESONANCE-BASEDELECTRICALPROPER- TIESTOMOGRAPHY:AREVIEW.IEEEREvBIoMEdENg2014;7:87–96

    work page

  7. [7]

    PhyS-EdBIol 1991;36(6):723–734

    (AACKE%,0ETROPOULOS,,.ILGES%,7UD.%XTRACTIONOFCONDUCTIVITY ANDPERMITTIVITYUSINGMAGNETICRESONANCEIMAGING. PhyS-EdBIol 1991;36(6):723–734

    work page 1991

  8. [8]

    IEEETBIo-MEdENg2017;64(11):2515–2530

    ,IU*,7ANG9,KATSCHER5,(EB.%LECTRICALPROPERTIESTOMOGRAPHY BASEDONB1MAPSIN-2):PRINCIPLES,APPLICATIONSANDCHALLENGES. IEEETBIo-MEdENg2017;64(11):2515–2530

    work page

Show all 17 references
  1. [9]

    KATSCHER5,VANDENBERGC.%LECTRICPROPERTIESTOMOGRAPHY:BIO- CHEMICAL,PHYSICALANDTECHNICALBACKGROUND,EVALUATIONANDCLINICAL APPLICATIONS.N-RBIoMEd2017;30(8):E3729

  2. [10]

    IEEET-EdIMagINg2015;34(9):1788–1796

    BALIDEMAJ%,VANDENBERGC,4RINKS*,VAN,IERA,.EDERVEENA, 3TALPERS,,CREZEE(,2EMIS2.C3)-%04:ACONTRASTSOURCEINVERSION APPROACHFORIMPROVED-2)-BASEDELECTRICPROPERTIESTOMOGRAPHY. IEEET-EdIMagINg2015;34(9):1788–1796

  3. [11]

    ARDUINOA,:ILBERTI,,CHIAMPI-,BOTTAUSCIO/.C3)-%04INPRESENCE OF2&-SHIELDFOR-2-COILS.IEEET-EdIMagINg2017;36(7):1396– 1404

  4. [12]

    ,EIJSEN2,BRINK7,VANDENBERGC,7EBBA,2EMIS2.3-DCON- TRAST 3OURCE )NVERSION-%LECTRICAL 0ROPERTIES 4OMOGRAPHY.IEEE T -EdIMagINg2018;37(9):2080–2089

  5. [13]

    -ANDIJA3,-ELIADò%,(UTTINGA.,,UIJTEN0,VANDENBERGC./PEN- INGANEWWINDOWON-2-BASED%LECTRICAL0ROPERTIES4OMOGRAPHY WITHDEEPLEARNING.SCIENTIfiCREPoRTS2019;9(1):8895

  6. [14]

    PhyS-EdBIol2018;55(2): 23–38

    CHRISTA,KAINZ7,(AHN%G,(ONEGGERK,:EFFERER-,.EUFELD%, 2ASCHER7,*ANKA2,BAUTZ7,CHEN*,OTHERS.4HE6IRTUAL&AMILY- DEVELOPMENTOFSURFACE-BASEDANATOMICALMODELSOFTWOADULTSAND TWOCHILDRENFORDOSIMETRICSIMULATIONS. PhyS-EdBIol2018;55(2): 23–38

  7. [15]

    -agNRESoN-Ed2018;80(1):90–100

    -ANDIJA3,3BRIZZIA,KATSCHER5,,UIJTEN0,VANDENBERGC.%RROR ANALYSISOFHELMHOLTZ-BASED-2-ELECTRICALPROPERTIESTOMOGRAPHY. -agNRESoN-Ed2018;80(1):90–100

  8. [16]

    VAN,IERA,2AAIJMAKERSA,6OIGT4,,AGENDIJK*,,UIJTEN0,KATSCHER 5,VANDENBERGC.%LECTRICALPROPERTIESTOMOGRAPHYINTHEHUMAN BRAINAT1.5,3,AND74:ACOMPARISONSTUDY.-agNRESoN-Ed2014; 71(1):354–363

  9. [17]

    ,EIJSEN2,&UCHS0,BRINK7,7EBBA,2EMIS2.DEVELOPMENTSIN %LECTRICAL0ROPERTIES4OMOGRAPHYBASEDONTHECONTRAST3OURCE )NVERSION-ETHOD.JIMagINg2019;5(2):25. COMBININGD,-%04AND3DC3)-%04 3UPP,E-E.4A2Y-A4E2IA, 0 1 2 ;3/M] 1 20 40 60 80 100 4RANSVERSAL 34/bracehtipdownleft /bracehtipupright/br...

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.