Experimental Evaluation of Sensor Directivity Using Isotropic Total Variation Minimization in Photoacoustic Tomography

Single-element transducers with circular scanning geometry are widely used in photoacoustic (PA) tomography. The finite aperture of transducers introduce banding artefact in the reconstructed images. The present work proposes a model-based method to address this problem. The finite aperture and the directivity effects of the transducer were included in the model matrix. For each pixel-detector pair, the spatial impulse response function of the transducer was first convolved with the PA signal from a spherical absorber. The model matrix was filled with these convolved signals. Image reconstruction was accomplished using the isotropic total variation (TV) minimization and backprojection (BP) algorithms. The experiments were also conducted using two test phantoms. The signal-to-noise ratio (SNR) was calculated as a figure of merit The TV method produced better PA images compared to BP method as the structures of the PA sources were reconstructed with higher accuracy. This is due to the inherent directivity correction in the TV protocol. It is also evident from the SNR values. The suggested approach is simple to understand and straightforward to implement.