Regularized Capon Beamformer Using $\ell_{1}$ -Norm Applied to Photoacoustic Imaging

TLDR: This paper proposes to add a $\ell_{1}$-norm regularization term to the conventional MV minimization problem and create a new sparse beamforming method, named Modified-Sparse-Mv (ms-MV)algorithm, which results in more noise reduction and sidelobe suppression compared to MV.

Abstract: Delay-and-Sum (DAS), as a non-adaptive beamforming method, is one of the most common algorithms used in Photoacoustic imaging due to its simple implementation. The results obtained from this algorithm suffer from low resolution and high sidelobes. The adaptive Minimum variance (MV) method improves the image quality compared to DAS in terms of resolution and contrast. In this paper, it is proposed to add a $\ell_{1}$ -norm regularization term to the conventional MV minimization problem and create a new sparse beamforming method, named Modified-Sparse-Mv (ms-Mv)algorithm. In fact, the sparsity of the output is forced to the beampattern by adding this new sparse added term, which results in more noise reduction and sidelobe suppression compared to MV. The minimization problem is convex, and therefore, it can be solved using an iterative algorithm. The results show that the proposed MS-MV method improves the signal-to-noise-ratio for about 5.36 $dB$ and 6.44 $dB$ compared to DAS and MV, respectively, for the designed wire phantom.

Figures

TABLE I THE CALCULATED FWHM (mm) FOR THE DESIGNED WIRE PHANTOM.

Fig. 2. The lateral variations of the reconstructed PA images shown in Fig. 1. Arrows and circles demonstrate the improvement caused by MS-MV algorithm.

Fig. 1. The reconstructed experimental PA images using (a) DAS, (b) MV and (c) MS-MV. All the images are shown with a dynamic range of 50 dB. Two wires are used as the imaging target.