不均匀场下基于压缩感知和并行计算的无参考扫描epi扭曲矫正和螺旋mri稀疏重建电子与通信工程专业论文.docxVIP

Free-Reference Free-Reference EPI Distortion Correction and Spiral MRI Reconstruction under Strong Inhomogeneous Field based on Compressed SensingSenslng andan Parallelarall omputl ABSTRACT As typical ultra-fast sequences，the echo planar imaging(EPI)and the spiral M对 play important roles in functional MRI and dynamic MIU．However,the ultra—fast sequences are sensitive to inhomogeneous field，leading to image distortions in EPI and blurring in spiral-MIU． Most of the published methods either need additional acquisitions to obtain the field map，or are not capable of the strong inhomogeneous problems．In this thesis，the proposed EPI correction algorithm is able to correct strong and local inhomogeneous problems without any extra acquisitions for the field map．The field map is represent as a linear combination of the two·dimensional Fourier bases．Applying the technologies of compressed sensing，convex optimization，phase unwrapping，and cuckoo search，both EPI images and field maps could be reconstructed．Simulations and experiments demonstrate that the proposed method is effective to correct the shape distortion and intensity variation of EPI image under strong inhomogeneous field． Most of the traditional algorithms for spiral MIU are based on gridding or non—unformed fast Fourier transformation(NUFFT)．However，kernel functions are often imperfect and the error of applying gridding kernel functions is unavoidable．In this thesis，A spatial—temporal transform(STT)matrix is introduced in the Ii norm optimize problem based on compress sensing．Applying the STT matrix，the sampled data is able to be a fidelity term directly rather than the traditional afier-gridding data, which avoids the gridding errors．111e simulation demonstrate that the STT results have less error than the NUFFT results．Further more，the algorithm is also suitable for blurting correction in spiral MIU under inhomogeneous field．Simulation shows that the blurring is effectively removed． Parallel computing on Graphics