Adaptive Photoacoustic Beamforming Algorithms for Blood Oxygen Saturation Estimation

Adaptive photoacoustic beamforming algorithms have provided significant improvement in image quality when compared to conventional delay-and-sum (DAS) algorithm. One of the key application areas of photoacoustic imaging (PAI) is the non-invasive estimation of blood oxygen saturation $(\mathbf{sO}_{\boldsymbol{2}})$ through multi-wavelength imaging using adaptive beamforming. Therefore, the objective of this work is to evaluate and compare the performance of adaptive PAI beamforming algorithms in the context of blood $\mathbf{sO}_{\boldsymbol{2}}$ estimation. Analysis involved a hybrid simulation method to simulate diffuse lesion and microvasculature phantoms and comparison among eight beamforming algorithms [Delay-and-sum (DAS), minimum variance (MV), DAS and MV with coherence weighting (DASCF and $\mathbf{MV}_{\mathbf{C}\mathbf{F}}$, filtered delay-multiply-and-sum (F-DMAS), signed DMAS (S-DMAS), photoacoustic sub-aperture processing with DAS and MV $(\mathbf{PSAP}_{\mathbf{D}\mathbf{A}\mathbf{S}}$ and $\mathbf{PSAP}_{\mathbf{M}\mathbf{V}})]$. Quantitative results demonstrated that DAS and MV had lower mean absolute error (MAE) compared to other methods (For microvasculature: median MAE (%) for $\mathbf{DAS}\boldsymbol{=0.16,} \mathbf{MV}\boldsymbol{=0.16,} \mathbf{DAS}_{\mathbf{C}\mathbf{F}}\boldsymbol{=0.25,} \mathbf{MV}_{\mathbf{C}\mathbf{F}}\boldsymbol{=0.25,} \mathbf{F-DMAS}\boldsymbol{=0.22,}$ S $\mathbf{DMAS}\boldsymbol{=0.22,}\ \mathbf{PSAP}_{\mathbf{D}\mathbf{A}}\mathbf{s}\boldsymbol{=0.25,}\ \mathbf{PSAP}_{\mathbf{M}\mathbf{V}}\boldsymbol{=0.25).}$ Therefore, careful attention must be taken while evaluating adaptive beamforming algorithms for $\mathbf{sO}_{\boldsymbol{2}}$ estimation using multi-wavelength PAI.