A massively parallel SIMD framework for fast 3D microwave tomography

Microwave Tomography (MT) is one of the most promising emerging imaging modality in recent years, which has shown potential to detect small tumors in human breast. The computational complexity and memory requirement of the MT algorithms are major challenges in further development. This paper presents a Single Instruction Multiple Data (SIMD) framework for fast 3D microwave tomography. A time domain inverse scattering algorithm was used to reconstruct the dielectric properties of human breast. The proposed SIMD framework was evaluated on anatomically accurate numerical breast phantoms, and reconstruction results are presented.