Compressed delay-and-sum beamforming for ultrafast ultrasound imaging

Carrillo

2016 IEEE International Ultrasonics Symposium (IUS)

et al. 2016

Self Cite

Abstract-Classical beamforming methods, based on DelayAnd-Sum (DAS) require an extensive number of samples and delay calculations to obtain high-quality images. Compressed Beamforming (CB) proposes an alternative to DAS, based on compressed sensing, which aims at reducing the data rate. However, proposed CB approaches induce a computationally heavy measurement model that hampers their attractiveness for iterative image reconstruction. In this paper, a CB framework, applicable to either radio-frequency or in-phase quadrature data and for both plane wave and diverging wave compounding, is described. The proposed framework exploits a computationally light measurement model which leads to tractable reconstruction. It solves a convex problem and assumes sparsity in a waveletbased model to achieve high-quality image reconstruction from measurements acquired with only few transducer elements.

Section: B Image Reconstructionmentioning

confidence: 99%

A compressed beamforming framework for ultrafast ultrasound imaging

Carrillo

2016 IEEE International Ultrasonics Symposium (IUS)

et al. 2016

Self Cite

“…The measurement operator associated to plane wave (PW) imaging has been derived in previous works [6], [9]. Formally, let us denote by r (x i , t) the element raw data received at time t by a transducer positioned at x i .…”

Section: Sparse Regularization For Ultrasound Imagingmentioning

confidence: 99%

“…In our previous works [5], [6], [9], [15], we suggested to use a concatenation of wavelet bases since it exhibits better reconstruction results than traditional wavelet-based models. The model, called sparsity averaging (SA) model, is composed of the concatenation of Daubechies wavelet transforms with different wavelet mother functions ranging from Daubechies 1 (Db1) to Daubechies 8 (Db8).…”

Section: Sparse Regularization For Ultrasound Imagingmentioning

confidence: 99%

“…Liebgott et al [2] use the sparsity of backscattered echoes in the wave atom frame to reduce the number of required echoes in a pre-beamforming step. In a post-beamforming step, a sparsity prior of radio-frequency (RF) images has been exploited in various models such as: Fourier basis [3], wavelet basis [4], sparsity averaging model [5], [6], α−stable distribution [7] or more recently in learned dictionaries [8].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Morphological component analysis for sparse regularization in plane wave imaging

Carrillo

2016 IEEE International Ultrasonics Symposium (IUS)

et al. 2016

Self Cite

Abstract-Classical ultrasound image reconstruction mainly relies on the well-known delay-and-sum (DAS) beamforming for its simplicity and real-time capability. Sparse regularization methods propose an alternative to DAS which lead to a better inversion of the ill-posed problem resulting from the acoustic wave propagation. In the following work, a new sparse regularization method is proposed which includes a component-based modelling of the radio-frequency images as well as a pointspread-function-adaptive sparsity prior. The proposed method, evaluated on the PICMUS dataset,outperforms the classical DAS in terms of contrast and resolution.

“…These methods are built upon forward models of the problem. David et al [2] and Besson et al [3] have proposed time-domain formulations of the problem. Besson et al have presented a forward model in the Fourier domain in which US propagation is seen as a projection on a non-uniform Fourier space [4].…”

Section: Introductionmentioning

confidence: 99%

USSR: An ultrasound sparse regularization framework

2017 IEEE International Ultrasonics Symposium (IUS)

Martinez

et al. 2017

Self Cite

Abstract-Ultrafast ultrasound (US) imaging uses unfocused waves to insonify the whole medium of interest at once, allowing pulse-echo US imaging to achieve very high frame rates, at the cost of a lower image quality. In this paper, we present USSR, an UltraSound Sparse Regularization framework which permits high-quality imaging at fast rates and with a very low memory footprint. The framework, based on highly parallelizable, parametric, matrix-free formulations of the measurement model and its adjoint as well as on well-chosen sparsity priors, is implemented on multi-threaded architectures and evaluated on the publicly available PICMUS dataset.