Standardization of <scp>MR</scp> Electrical Properties Tomography: A Guideline From the <scp>ISMRM</scp> Electro‐Magnetic Tissue Properties Study Group

MR Electrical Properties Tomography (MR-EPT) is an MRI-based technique that allows non-invasive measurement of the electrical properties of human tissue (i.e., electrical conductivity σ and relative permittivity εr). Over the past two decades, numerous methods have been published on how to reconstruct such properties from MRI data, as reported in a comprehensive review [1]. However, published in vivo results show a large variability in reconstructed conductivity and relative permittivity values, which may arise from the different data processing approaches, as also observed in the first MR-EPT reconstruction challenge [2, 3]. This guideline complements two other MR-EPT guidelines, that is, on phantoms [4] and acquisitions [5], and aims to standardize the reporting of methods and results in scientific publications to support reproducibility and comparability of new acquisition and reconstruction methods for MR-EPT, while indirectly improving MR-EPT reconstruction quality, study validity, and efficacy. Reporting phantom recipes, geometry, and reference values of electrical properties is of paramount importance. For details on what to report in publications, we refer the reader to the guideline on phantoms for MR-EPT [4]. Pre/post-processing—the following should be reported: Type of filters applied before and/or after EPs reconstruction, including filters kernel size (mm), weights, constraints to tissue-specific regions. Reconstruction method—the following should be reported: Type of reconstruction, for example, direct, inverse, differential/integral-based, image-based, deep-learning, 2D, 3D; Type of output, that is, conductivity (S/m) only or both conductivity (S/m) and relative permittivity (-); The reconstruction algorithm or reference(s) to a publication(s) explaining it; Imaging data used for the reconstruction, for example, transceive phase for phase-only MR-EPT; Assumptions on B1-fields, field-components, or EPs, such as the local homogeneity assumption, and on the use of the signal magnitude, for example, for constraining the reconstruction to specific regions with the same signal intensity; Input parameters for example, kernel or patch size (mm) and weights; Reconstruction time (min:s), including training time (min:s) for deep-learning methods and computing hardware specifications; If applicable, size of training and validation datasets. Tissue structure—If a priori knowledge on tissue structure was used either for the pre/post-processing steps or EPs reconstruction, the following information should be provided: Tissue structure and/or tissue boundary information used; How this was derived, for example, segmentation, image contrast, boundary detection methods; Software packages and parameters used to obtain this information. To define a standardized way of reporting both MR-EPT images and quantitative values, a Delphi process, similar to that used to standardize color-maps for MR relaxometry [8], was performed within the Electro-Magnetic Tissue Properties (EMTP) Study Group of the International Society of Magnetic Resonance in Medicine (ISMRM), including both clinical and technical expertise as well as industry representatives. The process relied on a panel of up to 25 experts from the EMTP Study Group who were asked to answer four questionnaires between January and April 2025. Details on the process and questionnaires are provided in Data S1. The following statements about color-maps, ranges, data availability, and metrics reached consensus. Alongside these consensus statements, recommendations are provided to facilitate benchmarking of new reconstruction methods. Recommendation: Since a consensus on the color-maps to be used was not reached, we recommend using color-maps reported in Crameri et al. [9], that are in agreement with the statements that reached consensus. To facilitate comparison between scientific publications, we recommend using Lipari for conductivity and Navia for relative permittivity. If necessary, these maps may be adapted to include a color clearly distinguishable from the others, dedicated to regions of unknown EPs. A consensus about how to display areas with erroneous EPs values (either negative or too high) was not reached. However, based on the trend of answers, negative EPs may be displayed with the color of the background. Questions were not formulated on how to display EPs with unrealistically high values. However, we recommend displaying these with the color of the upper bound of the color-map. Recommendation: The suggested ranges for relative permittivity did not reach consensus, while the range of conductivity that reached consensus applies to the brain at 3 T (128 MHz). To facilitate comparison between scientific publications, we recommend using 0–2.5 S/m for conductivity and 30–100 for relative permittivity, both at 1.5 T (64 MHz) and 3 T (128 MHz). Testing new reconstruction methods on the following datasets provided here: Reporting the results generated by the provided analysis scripts using the available templates. Datasets, analysis scripts to be run on reconstructed electrical properties maps using the provided input datasets, and reporting templates (including an example of a standardized report) are available here: https://doi.org/10.5281/zenodo.17879937 We encourage researchers to publish datasets and code alongside their publications. A link to these repositories can be added to the community website: EMTPhub.org. The authors thank the ISMRM-EMTP study group, which reviewed both the original and the revised version of the guideline. S.M. received funding from NWO-VENI 18078 for the coordination of this work. S.M., A.A., T.M., and L.Z. acknowledge funding from the 24DIT01 APULEIO project. The project 24DIT01 APULEIO has received funding from the European Partnership on Metrology, co-financed from the European Union's Horizon Europe Research and Innovation Programme and by the Participating States. I.I.G. acknowledges funding from NIH K99 EB035163. Datasets, analysis scripts to be run on reconstructed electrical properties maps using the provided input datasets, and reporting templates are available here: https://doi.org/10.5281/zenodo.17879937. Data S1: Delphi process information and rounds. Table S1: Results Round 1. Table S2: Results Round 2. Table S3: Results Round 3. Table S4: Results Round 4. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.