Breast Tumour Identification based on Chimera Overset Grid with Spline Interpolation in Forward-Backward Time-Stepping

Abstract

Active microwave imaging (MI) technique has been recently recommended as a new technology for early detection of malignant breast tumours. Finite-difference Time-domain (FDTD) method is an effective numerical tool for computational electrodynamics modelling. However, the main drawback of the FDTD method is difficult to model the curved boundaries and small features of arbitrary shaped fibroglandular region due to its restriction to inherent orthogonal grids. To address this issue, Chimera Overset Grid (COG) method with spline interpolation in Forward-Backward Time-Stepping (FBTS) inverse scattering technique was proposed to overcome the limitation of FDTD method. In order to analyse the accuracy of this new method to detect a malignancy tumour embedded in different classes of breast parenchymal density, the homogeneous mostly fatty breast and extremely dense breast were chosen as Case A, and Case B, respectively. The Mean Squared Error (MSE) and normalised cost error functional for reconstructed dielectric profiles utilising the proposed method achieved significantly lower values than the FDTD method in FBTS inverse scattering. The results indicated that the proposed method could accurately identify malignancy tumour location and reconstruct the breast composition in dielectric profiles.