Efficient nonlinear beamformer based on P’th root of detected signals for linear-array photoacoustic tomography : application to sentinel lymph node imaging

In linear-array transducer-based photoacoustic (PA) imaging, B-scan PA images are formed using the raw channel PA signals. Delay-and-sum (DAS) is the most prevalent algorithm due to its simple implementation, but it leads to low-quality images. Delay-multiply-and-sum (DMAS) provides a higher image q...

Full description

Saved in:
Bibliographic Details
Main Authors: Mozaffarzadeh, Moein, Makkiabadi, Bahador, Periyasamy, Vijitha, Pramanik, Manojit
Other Authors: School of Chemical and Biomedical Engineering
Format: Article
Language:English
Published: 2018
Subjects:
Online Access:https://hdl.handle.net/10356/88990
http://hdl.handle.net/10220/46046
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:In linear-array transducer-based photoacoustic (PA) imaging, B-scan PA images are formed using the raw channel PA signals. Delay-and-sum (DAS) is the most prevalent algorithm due to its simple implementation, but it leads to low-quality images. Delay-multiply-and-sum (DMAS) provides a higher image quality in comparison with DAS while it imposes a computational burden of O  (  M2  )  . We introduce a nonlinear (NL) beamformer for linear-array PA imaging, which uses the p’th root of the detected signals and imposes the complexity of DAS [O  (  M  )  ]. The proposed algorithm is evaluated numerically and experimentally [wire-target and in-vivo sentinel lymph node (SLN) imaging], and the effects of the parameter p are investigated. The results show that the NL algorithm, using a root of p (NL_p), leads to lower sidelobes and higher signal-to-noise ratio compared with DAS and DMAS, for (p  >  2). The sidelobes level (for the wire-target phantom), at the depth of 11.4 mm, are about −31, −52, −52, −67, −88, and −109  dB, for DAS, DMAS, NL_2, NL_3, NL_4, and NL_5, respectively, indicating the superiority of the NL_p algorithm. In addition, the best value of p for SLN imaging is reported to be 12.