Recovering missing diffusion signal data for reconstructing fiber orientations

Q-ball imaging (QBI) is a recent high angular resolution diffusion imaging technique for studying the orientation of fibers in the brain or heart, but it requires a high number of acquisitions, thus considerably lengthening acquisition time. Reducing diffusion data to be acquired while achieving the same fiber orientation reconstruction quality is then of fundamental importance. In this paper, we propose a new method of fiber orientation reconstruction from much less data acquisitions. The method is based on recovering missing diffusion data. It consists in imposing constraints both in q-space and in spatial space, applying data consistent principle in q-space, and iterating the process until no change is observed in the reconstructed orientation distribution function (ODF). The results show that using diffusion data in 12 directions, the reconstructed ODFs are rather close to those obtained with 42 diffusion directions, which allows shortening considerably acquisition time.