Cross-Sectional Complex Permittivity Imaging for Three-Dimensional Microwave Breast Cancer Ablation

To satisfy the critical need to ensure treatment efficacy and patient safety, this paper proposes a method to perform ablation zone monitoring in microwave ablation (MWA) breast cancer treatment based on the incorporation of the time difference of arrival (TDOA) and contrast source inversion (CSI) techniques. The hybrid TDOA-CSI uses TDOA to localize the ablation zone and CSI to reconstruct the associated complex-permittivity changes, as previous studies have been limited to two-dimensional reconstructions. Thus, this study introduces the quasi-three-dimensional (3D) imaging scheme, in which 3D total-field observation data obtained with a circular receiver array are converted into equivalent two-dimensional (2D) cross-sectional data for CSI reconstruction. For this conversion, a nonlinear deep learning approach with fully connected neural network (FCNN) is used, which learns the nonlinear mapping from the 3D observation data to the corresponding 2D cross-sectional data in the frequency domain. The synthetic data generated by full-wave 3D finite-difference time-domain (FDTD) simulations, assuming an MRI-derived realistic phantom, demonstrate that the proposed scheme can provide a cross-sectional reconstruction of the ablation zone and its temporal complex-permittivity changes under conditions of sufficient SNR and the specific phantom model used for training.