Enhancing Photoacoustic Tomographic Imaging using Total Generalized Variation Regularization: A Model-based Approach

Photoacoustic tomographic (PAT) imaging has emerged as a promising non-invasive medical diagnostic technology for visualizing biological tissue. However, the inverse problem in PAT is known for its ill-posed nature, and the presence of noise in photoacoustic signals often results in blurred reconstructed images. Although hybrid total variation (TV) regularization methods have been commonly employed for PAT reconstruction, they face challenges in effectively eliminating staircasing artifacts and preserving sharp edges. To overcome these limitations, a novel approach has been proposed that incorporates total generalized variation (TGV) as a regularization term in the objective function. The model-based method utilizes the fast-composite splitting algorithm (FCSA) for implementation. To evaluate the performance of the proposed algorithm, numerical phantoms are simulated and results are compared with other hybrid TV methods as well as conventional TV methods based on performance metrics such as structural similarity index (SSIM), peak signal-to-noise ratio (PSNR), and relative L
2
error. The calculated values of SSIM, PSNR, and relative L
2
error are 0.9120, 40.3708, and 1.1027×10
^-5
respectively for the proposed method. Experimental validation has also been carried out. The accuracy of the reconstructed results demonstrates the potential and significance of the proposed method.