An Accurate Diffraction Ultrasound Tomography Images Reconstruction Algorithm Based on LS_NUFFT

According to the Fourier diffraction projection theory, the Fourier transform of projection data from multi-band frequency ultrasonic is distributed over complex plane in form of different radius arcs. The classical filtered back projection (FBP)algorithm isn’t fit for distributing of the k space data, and the conventional gridding data interpolation on non-Cartesian data would produce much artifact . The least squares nonuniform fast Fourier transform(LS_NUFFT) is brought forth by Liu and Nguyen and applied in MRI image reconstruction. The LS_NUFFT is apply successfully in diffraction ultrasound tomography image reconstruction for the first time and the best kernel matrix size is proved by experiment. The reconstruction experiments based on LS_NUFFT, Min-max NUFFT and GFFT algorithms are done respectively, the reconstruction quality of LS_NUFFT is much better than Min-max method and GFFT method .