An Efficient Image Formation Algorithm for Real-time Linear-Array Photoacoustic Tomography

In photoacoustic (PA) imaging using commercial ultrasound transducers, two dimensional PA images are produced by commonly used delay-and-sum (DAS) as the reconstruction method. However, the reconstructed image is affected by noise and artifacts. Here, we investigate the performance of a nonlinear (NL) beamforming method for linear-array PA image formation. The proposed algorithm uses the pth root of the recorded signals and imposes a computational complexity in the order of DAS (O(M)). We have evaluated the performance of the proposed method numerically, having a signal-to-noise ratio of 30 dB and -10 dB. It is shown that at the depth of 15 mm, the NL 3 (NL beamformer with a p=3) outperforms DAS, DMAS and NL 2 for about 26 dB, 11 dB and 11 dB, in terms of level of sidelobe, respectively. In addition, DAS, DMAS, NL 2 and NL 3 lead to a full-width-half-maximum of about 1.4 mm, 1.04 mm, 1.04 mm and 0.86 mm, respectively. The proposed method can be a great choice for sentinel lymph node imaging.