Research On Adaptive Beamforming Algorithms For Synthetic Aperture Ultrasound Imaging

Due to the advantages of real time,non-invasive,nonionizing radiation and low cost,medical ultrasound imaging has been taken as an important medical inspection method and widely used in clinical diagonosis.In ultrasound imaging,beamforming plays a key role in the final generated ultrasound image quality.Conventional ultrasound imaging uses the delay-and-sum(DAS)to reconstruct the ultrasound image,however DAS only simply sums the echo signals and cannot distinguish desired signals and clutter,and thus lead to low quality image with wide mainlobe and high sidelobe.Over recent years,adaptive beamforming techniques beased on characteristics of echo signals have been demonstrated that they can provide high quality beams with narrow mainlobe and low sidelobe and have become a research hotspot in the field of ultrasound imaging.The exsisting beamforming methods are modified and new beamforming methods are proposed to be implemented in the synthetic transmit aperture(STA)ultrasound imaging to enhance the image quality in this thesis.The main works and results are presented as follows:First,the eigenspace-basaed minimum variance(ESBMV)beamformer can effectively suppress sidelobes,however it will introduce dark-region artifacts(DRA)beside the hyperechoic point target.To address the problem of DRA,the spatial coherene is utilized to adaptively select the eigenvalue threshold in ESBMV.Besides,the adjusted ESBMV is combined with short-lag spatial coherence(SLSC)to further suppress sidelobes.Second,to address the tradeoff of the generalized coherence factor(GCF)between sidelobe suppression and speckle preservation,a dynamic GCF(DGCF)is proposed in this thesis.On one hand,the amplitude standard deviation of aperture data is utilized to adaptively select the cut-off frequency in conventional GCF to retain high resolution and contrast performance.On the other hand,the spatial frequency spectrum of the output of convolution is taken as the spectrum of aperture data in the estimation of DGCF to enhance the preservation of speckle.Experimental results demonstrate that the proposed method improves resolution and contrast significantly,and simultaneously preserving speckle signals.Third,the signal mean-to-standard-devation factor(SMSF)based on the mean-tostandard-devation ratio(MSR)is proposed in this thesis.To improve the robustness of SMSF,the estimation of SMSF is extended to the spatial frequency domain by using the spectrum and a square neighborhood average technique to improve the preservation of speckle statistics.Experimental results demonstrate that SMSF has a superior resolution performance and can significantly improve contrast.Moreover,the frequency-domain SMSF partially overcomes the problem of speckle corruption introduced by the timedomain SMSF.Finally,the covariance mean-to-standaed-deciation factor(CMSF)is proposed in this thesis to overcome the trade-off between resolution and speckle statistics of SMSF.The CMSF is defined as the average of MSR values estimated from the row vectors of the covariance matrix.In CMSF,adaptive subarray length and diagonal reducing are combined to suppress incoherence off-axis scattering in the conventional DAS image.Simulation and experimental results indicate that the proposed CMSF retains the high contrast and resolution performances of SMSF,and simultaneously preserving the speckle.In summary,ESBMV and GCF beamforming methods are modified to overcome their limitations in ultrasound imaging.Furthermore,new beamforming methods based on the SMSF proposed in this thesis effectively improve the resolution and contrast.Specially,CMSF significantly enhances the lesion detactability,and thus have the potential to provide an instructive attempt for clinical diagnosis.