The tomography of fusion plasmas provides local information on plasma emissivity from line-integrated measurements (projections). However, the corresponding inversion task presents an ill-posed and often underdetermined problem. Compared to industrial and medical tomography systems, data in fusion research are spatially sparse due to the limited number of lines of sight, and they may vary rapidly in time. Therefore, dedicated inversion techniques have been developed that allow for lower spatial resolution and implementation of a priori information and constraints. In this contribution, the main inversion techniques used today are reviewed, with working results and challenges outlined. Special attention is given to techniques that allow for rapid tomography inversions, because of their future potential for real-time applications, and a new combined technique is proposed.