3-D ultrasound reconstruction associated with augmented reality allows physicians to explore a region of interest in 3-D in an intuitive and user-friendly way, while leveraging the advantages of 2D ultrasound imaging: simple, low cost and non-ionizing. It may assist many clinical tasks, such as practician training, procedure assistance or visualization of tissues difficult to interpret through 2D visualization. Recently, new unsupervised deep learning techniques based on a continuous description of the 3-D field, showed promising results in terms of 3-D model estimation, robustness to noise and uncertainty, and efficiency. Inspired by these approaches, the objective of this work is to propose a 3-D ultrasound reconstruction method based on neural implicit representations, adapted to the challenges of an augmented reality pipeline. Results on simulated and experimental data show the precision and efficiency of the reconstruction compared to state-of-the-art neural and traditional reconstructions.