Kecheng Yuan / University of Science and Technology of China
Changliang Wang / University of Science and Technology of China
Penghui Luo / University of Science and Technology of China
Jiantai Zhou / University of Science and Technology of China
Huabin Zhang / University of Science and Technology of China
QingYun Liu / University of Science and Technology of China
In MRI, image contrast is reduced by the influence of fat tissue, which is not conducive to the detection of potential lesions. How to suppress fat is challenging because of the interference of complex background noise, rapidly changing magnetic fields, and eddy currents. To accurately separate water and fat, we propose a water/fat separation method based on Regional Iterative Phasor Extraction (RIPE) and a new weighted smoothing map. To resolve the ambiguity of the phase error of the images acquired at two different TEs, a new weighted map which combines magnitude map and distance map in equal proportions, is used for smoothing of field vector. In particular, the distance map is composed of a Gaussian function centered on each processed pixel, with a mean value of 0 and a variance of 0.5. Compared with the traditional unweighted RIPE method, the proposed method introduces information based on the magnitude and the distance of the smooth pixels, which can effectively avoid the generation of errors in low-quality areas and then propagated to high-SNR regions, failing overall image smoothing. The proposed method was evaluated on simulation data and in vivo liver data to test the performance. Experimental results demonstrate that the proposed method is more robust and can be applied to applications of water/fat separation that require fast imaging, such as breath-hold imaging or dynamic imaging.