Method and system for realizing water-fat separation

Abstract

The invention relates to a method for realizing water-fat separation. The method comprises the following steps of collecting a plurality of magnetic resonance images with different echo time intervals; dividing the plurality of magnetic resonance images into a plurality of sub blocks with different resolutions; respectively selecting the minimum values in the fitting errors of the plurality of sub blocks as field pattern estimation values of the sub blocks corresponding to the resolutions, and respectively selecting seed blocks corresponding to the resolutions according to the field pattern estimation values; obtaining pixel seed points at the corresponding original image resolution of different resolutions by using the plurality of seed blocks corresponding to the resolutions, and merging the pixel seed points to obtain a final pixel seed point; selecting the real field image value of the rest pixel points by using the final pixel seed point; and performing calculation on the plurality of magnetic resonance images by using the real field pattern values of all of the pixel points to respectively obtain a water pattern and a fat pattern. By adopting the method, the efficiency of selecting the real field pattern value can be effectively improved; and the effective water-fat separation can be performed through the real field pattern value. In addition, the invention also provides a system for realizing water-fat separation.

Description

technical field [0001] The invention relates to the technical field of nuclear magnetic resonance imaging, in particular to a method and system for realizing water-fat separation. Background technique [0002] Water-fat separation technology has a wide range of application values ​​in clinical magnetic resonance imaging (MRI), for example, in the high-value main magnetic field B0 offset field (ie, the field map value f B , in Hz) fat suppression. The traditional water-fat separation method is based on the multi-echo MR signal pixel-by-pixel iterative fitting separation algorithm (Iteration Decomposition of water and fat with Echo Asymmetry and Least-squares, referred to as IDEAL), but this algorithm can easily converge to a local minimum when dealing with high-value B0 field images Cause water-fat separation, and for single-component pixels, such as pixels containing only water or fat, the algorithm may also get wrong separation results, because at this time f B has two so...

AI Technical Summary

Problems solved by technology

Method 3 collects 3 equally spaced MR images, and then uses the golden section algorithm to obtain the candidate solutions of the pixel field map. Since the estimated field map values ​​of the equidistant MR images are periodic, there are many candidate solutions for each pixel. , which will reduce the efficiency of real field image value selection

In addition, method 1 and method 3 only get the phasor vector of the field diagram, but not the real field diagram value

[0008] Therefore, how to effectively improve the efficiency of selecting real field map values ​​and effectively separate water and fat through real field map values ​​is still a technical problem

Images