An Inverse Problems Approach to MR-EPT Image Reconstruction

Area of Science:

• Medical Imaging
• Biophysics
• Electromagnetism

Background:

• Magnetic Resonance-Electrical Properties Tomography (MR-EPT) maps electrical conductivity and permittivity using MRI.
• Existing MR-EPT methods often rely on differentiating RF field data, increasing noise sensitivity.
• The Helmholtz equation describes the time-harmonic RF field within the scanner.

Purpose of the Study:

• To present an inverse approach for MR-EPT reconstruction that avoids data differentiation.
• To improve the accuracy and reduce noise sensitivity in electrical properties mapping.
• To evaluate the performance of Quadratic (QR) and Total Variation (TV) regularization schemes in MR-EPT.

Main Methods:

• Developed an inverse MR-EPT approach using a forward model for RF field magnitude and phase.
• Employed a fitting strategy to estimate the electrical properties distribution.
• Implemented and validated QR and TV regularization techniques through numerical simulations and experimental data.

Main Results:

• The proposed inverse MR-EPT approach accurately reconstructs geometry and identifies contrasts in both simulated and experimental data.
• The method demonstrates reduced noise sensitivity compared to differentiation-based approaches.
• Total Variation (TV) regularization effectively reconstructs sharp spatial transitions in electrical properties.

Conclusions:

• The novel inverse MR-EPT approach offers a robust and accurate method for mapping electrical properties.
• Avoiding data differentiation enhances noise resilience and image quality.
• TV regularization provides superior reconstruction of fine details in electrical property distributions.