Three-dimensional projection imaging with half the number of projections

This study demonstrates a sampling scheme for three-dimensional projection imaging with half the number of projections. The angular distribution of projections is designed so that the oversampling of the low spatial frequencies, in tandem with a partial Fourier algorithm, can be used to recreate purposely missing projections. The performance of the sampling scheme and its associated reconstruction algorithm are illustrated with computer-simulated as well as experimental data sets. We find that this technique produces images of comparable quality to the conventional projection imaging scheme. Although a 30% loss of signal-to-noise ratio (SNR) results from its use, the algorithm should prove useful for applications where robustness against motion artifacts and reduced T2* signal loss are desired.