Variable Density Random Sampling In Parallel Magnetic Resonance Imaging Study

The two important points of Magnetic Resonance Imaging, which are imaging quality and imaging speed, have been studied in software by algorithm of data scanning trajectory and algorithm of image reconstruction.In data scanning, a new sampling method was adopted which was named variable-density random sampling. In MRI, subsampling is used to reduce data scanning time which does not satisfy the Nyquist sampling frequency and result in image aliasing. Besides, the power of an image is mainly concentrated in the central area of K-space. Reconstructed images usually have aliasing containing low frequency information because an equal-space sampling is used in traditional parallel sampling. Unequal-space variable-density random sampling can reduce image aliasing, which sampled the low frequency information of K-space mainly and high frequency information rarely. In image reconstruction, an optimization model was adopted, which was solved by using the theory of nonlinear conjugate gradient.In imaging quality, the result showed that, compared to SENSE and GRAPPA, this new sampling method reduced image aliasing greatly under the same reduction factor. In imaging speed, the result showed that, compared to GRAPPA, data scanning time was reduced because no auto-calibration signal lines were needed in this new reconstruction algorithm. This algorithm can improve imaging speed to some extent.