Consistent Region-Growing Based Dixon Water and Fat Separation for Images with Disconnected Objects

Hua Ai¹, Jingfei Ma¹

Consistent water and fat separation in images with disconnected objects is difficult for a region-growing based Dixon method. Here, we propose to monitor and record the quality index of a recently-proposed algorithm for region-growing at each step. The quality index is then used to automatically segment the disconnected objects into separate sub-images. Finally, the sub-images are consistently recombined on the basis of water and fat spectral asymmetry and slice-to-slice phase correlation. The proposed method was tested on a total of 1106 axial in vivo leg images and was shown to reduce the number of inconsistent slices from 203 to 6.