In recent years, ultrasound imaging equipment has developed in diverse ways to meet the needs of various clinical applications. However, the in-herent characteristics of ultrasound imaging, including phenomena like diffraction, attenuation, interference, and refraction, as well as issues like speckle, artifacts, and noise, adversely affect spatial resolution. This signifi-cantly impacts the accuracy of clinical diagnosis and poses an obstacle to its widespread application. To improve the quality of ultrasound imaging, traditional methods have focused mainly on hardware enhancements and reconstruction method optimization. However, hardware improvements increase manufacturing difficulty and cost, while reconstruction algo-rithm optimization often comes at the expense of temporal resolution. Therefore, finding more exquisite methods to break through the spatio-temporal resolution limits of ultrasound imaging, promoting the preci-sion, intelligence, miniaturization, and cost-effectiveness of medical ultra-sound equipment, and ensuring accurate diagnosis are foundational keys to advancing precision intelligent ultrasound healthcare. In this chapter, we introduced advanced deep learning techniques applied to ultrasound image reconstruction and explored the challenges and potential future trends in this evolving field.