阈值提取法用于压缩感知3D mDIXON肝脏脂肪定量的可行性

目的 探讨肝脏阈值提取法用于压缩感知3D mDIXON肝脏脂肪定量的可行性。方法 对28名健康志愿者以加速因子2行肝脏压缩感知3D mDIXON序列扫描。由2名影像科医师分别采用阈值提取法和传统轴位ROI法测量全肝脂肪含量，并记录2种方法测量脂肪分数的时间；分析2种方法定量肝脏脂肪分数的可重复性及2种方法测量结果的相关性。结果 2名医师采用阈值法获得的脂肪分数分别为（3.53±1.34）%及（3.40±1.24）%，一致性较好（ICC=0.996，P=0.001）；采用传统轴位ROI法测量的脂肪分数分别为（2.35±1.35）%及（2.31±1.13）%，一致性亦较好（ICC=0.994，P=0.001）。阈值提取法（3.47±1.29）%]和轴位ROI法（2.45±1.30）%]获得的肝脏脂肪分数呈高度正相关（r=0.997，P=0.001）。阈值提取法平均测量时间（3.2±0.8）min，轴位ROI法平均测量时间（5.0±0.3）min。结论 阈值提取法可方便快捷地用于压缩感知3D mDIXON肝脏脂肪定量。

Feasibility of threshold extraction method for compressed sensing 3D mDIXON liver fat quantification

Objective To explore the feasibility of threshold extraction method for compressed sensing 3D mDIXON liver fat quantification. Methods Totally 28 healthy volunteers underwent liver compressed sensing 3D mDIXON sequence liver scanning with the acceleration factor 2. The whole liver fat content was measured by 2 imaging physicians using the threshold extraction method and traditional axial ROI method, respectively, and the time of fat fraction measurement with 2 methods were recorded. The repeatability of the two methods in quantifying liver fat fraction, and the correlation of the results measured with two methods was analyzed. Results The fat fraction obtained by 2 physicians with threshold extraction method were (3.53±1.34)% and (3.40±1.24)%, respectively, with good consistency (ICC=0.996, P=0.001), while those measured with the traditional axial ROI method were (2.35±1.35)% and (2.31±1.13)%, respectively, also with good consistency (ICC=0.994, P=0.001). The liver fat fraction obtained with threshold extraction method (3.47±1.29]%) was highly positively correlated with that with axial ROI method (2.45±1.30]%) (r=0.997, P=0.001). The average measuring time of threshold extraction method was (3.2±0.8)min, of axial ROI method was (5.0±0.3)min. Conclusion Threshold extraction method could be conveniently and easily used for compressed sensing 3D mDIXON liver fat quantification.