Tumor photoacoustic image reconstruction method based on deep learning

Photoacoustic imaging combines the advantages of ultrasound imaging and optical imaging, has the characteristics of cross-scale, deep penetration, and non-radiation, providing a new plan for cancer prevention and treatment. The combination of deep learning and tumor photoacoustic image reconstruction has important technological innovation and clinical application value for tumor prevention and treatment. In this paper, a photoacoustic image reconstruction network SEU-Net is designed. Through the algorithm evaluation system and network training strategy, the network can achieve fast and accurate image reconstruction on the basis of sparse sampling, effectively reducing equipment costs. Based on the analysis of tumor tissue structure and physiological characteristics, two models of geometrical figure and blood vessel grayscale image are designed to simulate tumor tissue, and a simulation data set is established through k-Wave. The model is trained to remove artifacts and the initial sound pressure signal map reconstruction, and compared with other classic reconstruction networks, the algorithm proposed in this paper has a good reconstruction effect.