A device for performing tomosynthesis in real time is described. Multiple imaging sources (such as x-ray sources) may be energized in parallel and collimated towards a field of view. Objects within the field of view cast shadows onto one or more detectors. An imaging system may read the one or more detectors and acquire multiple views corresponding to the multiple imaging sources to produce a reconstructed image of an object of interest. From this reconstructed image, a target of the radiation therapy can be located, and the delivery of the radiation can be adjusted, as needed. The approach provides a real-time tomosynthesis design that can produce enhanced contrast for guidance of, for example, lung tumor treatment. Higher frame rates can be achieved to better compensate for changes in the position of the target during radiation therapy due to, for example, respiratory or cardiac motion.
