Parametric Reconstruction of Photoacoustic Tomographic Imaging using Gaussian Mixture Model and Evolutionary Methods

Photoacoustic tomographic (PAT) imaging is a non-invasive biomedical imaging methodology based on the photoacoustic effect in the tissue. The basic principle involves that a short pulse laser is used to irradiate the biological tissue where the tissue particles absorb the photon and generate acoustic pressure waves in the sample. The pressure signals are measured by the ultrasonic transducer placed at the multiple locations on the boundary of the sample. Subsequently, the absorbed energy or the initial pressure distribution in the medium is reconstructed from the measured acoustic pressure signals. In this work, to reduce the number of unknowns in the reconstruction problem, an eight component Gaussian mixture model (GMM) is used to parameterize the initial pressure distribution of the tissue medium. In this study, two optimization techniques like Particle swarm optimization (PSO) and Genetic algorithm (GA) have been considered for image reconstruction. Performance of developed algorithms was evaluated using structural similarity index (SSIM) between the reconstructed and the actual images. The obtained values of SSIM of the reconstructed images are 0.9381 for PSO and 0.9357 for GA which shows the accuracy of reconstruction as well as reinforces the potential of the proposed reconstruction algorithms. The proposed algorithm has been illustrated for finger joints like phantom as well.