Depth- and Size-Dependent Correction of Tumor Conductivity In Phase-Based MR Electrical Properties Tomography
Abstract
Purpose
The quantitative accuracy of MR Electrical Properties Tomography (MR-EPT) is strongly influenced by tissue heterogeneity and lesion geometry. This study systematically investigates the effects of tumor depth and tumor size on conductivity reconstruction errors in phase-based (PB) MR-EPT and proposes depth- and size-dependent correction strategies based on phantom experiments and numerical simulations.
Methods
Tumor conductivity was modeled using saline-filled spherical inserts (true conductivity: 1.2 S/m) embedded in a cylindrical ECT phantom with a background conductivity of 0.3 S/m. Tumor depth was examined by placing a 37-mm-diameter sphere at multiple depths, while tumor size effects were evaluated using spheres ranging from 5 to 37 mm in diameter. MR-EPT data were acquired with an ultrashort echo time (UTE) sequence, and conductivity was reconstructed using a phase-based algorithm. Corresponding numerical simulations were conducted in COMSOL using both homogeneous phantom and heterogeneous human brain models. Reconstruction error was quantified as the relative difference between reconstructed and true conductivity values. Simulation-derived correction coefficients were obtained and linearly fitted as functions of tumor depth and size.
Results
Both phantom measurements and simulations demonstrated systematic underestimation of tumor conductivity with increasing depth and decreasing tumor size. In phantom experiments, reconstructed conductivity decreased from 1.17 S/m at a depth of 35 mm to 0.67 S/m at 105 mm. For decreasing tumor size, conductivity dropped from 1.11 S/m for 37-mm spheres to approximately 0.85 S/m for 10-mm spheres. Similar trends were observed in heterogeneous brain simulations. Applying the proposed depth- and size-dependent correction models significantly improved reconstruction accuracy, reducing relative conductivity errors from up to 0.44 to below 0.12 in phantom experiments.
Conclusion
Tumor depth and size introduce predictable and systematic errors in PB-based MR-EPT conductivity reconstruction. Correction models derived from combined phantom and heterogeneous brain simulations substantially improve quantitative accuracy, supporting the clinical potential of MR-EPT for tumor characterization.