摘要
目的采用CUBE-FSE-FLEX序列测量水脂混合溶液模型的脂肪浓度、猪瘦肉及肥肉容积,研究基于此技术的脂肪浓度定量及容积测量准确性。方法(1)水脂混合模型:分别制作浓度范围2%~30%、浓度梯度为2%的脂肪乳溶液和浓度范围0%~100%、浓度梯度为10%的水脂溶液,分别采用CUBE-FSE-FLEX和IDEAL-IQ序列扫描水脂溶液和脂肪乳溶液模型,获得两组序列所测脂肪分数,采用配对t检验分别比较两组测量值与实际脂肪分数的差异,采用线性回归方法及Bland-Altman散点图分析两组测量值与实际脂肪浓度的相关性及一致性;(2)肥瘦肉模型:采用CUBE-FSE-FLEX序列扫描肥瘦肉模型,分别获得手工分割(MS)和半自动分割算法[多维阈值法(MDT)及区域增长法(RG)]测量的肥肉和瘦肉样本的体积,采用配对t检验、Pearson相关法及Bland-Altman散点图行3种方法容积测量值与实际容积的差异性、相关性及一致性检验;对肥瘦肉模型采用的MDT测得值进行Pearson相关法及Bland-Altman散点图分析。结果(1)水脂溶液:CUBE-FSE-FLEX-FaF与实际脂肪浓度无显著性差异(t=0.113,P=0.913),IDEAL-IQ-FaF与实际浓度差异显著(t=-41.084,P<0.001),两种测量值与实际脂肪浓度均有高度直线相关,分别为0.941(CUBE-FSE-FLEX)与1(IDEAL-IQ)。测量值与实际值差值的95%置信区间为:CUBE-FSE-FLEX(-19.1708,18.5199),IDEAL IQ(-4.7593,-3.4589)。(2)脂肪乳溶液:CUBE-FSE-FLEX及IDEAL-IQ测量值与实际脂肪浓度均有显著性差异,(t=34.668,P<0.001)、(t=21.363,P<0.001),与实际脂肪浓度均呈高度直线相关,R^2分别为0.999(CUBE-FSE-FLEX)与1(IDEAL-IQ),测量值与实际值差值的95%置信区间为:CUBE-FSE-FLEX(-4.1762,-2.68),IDEAL-IQ(-2.0076,0.990)。(3)肥瘦肉模型:配对t检验示肥瘦肉模型MS测量值与实际值均无显著差异(t=0.777,P=0.443;t=1.727,P=0.94);MDT及RG测量值与实际值均有显著差异(P<0.001)。肥瘦肉模型中三种测量值均与实际容积值高度相关(R^2均>0.
Objective To use the CUBE-FSE-FLEX sequence to measure the fat concentration,lean pork and fat volume in a water-fat mixed solution model,and to study the fat concentration quantification and volume measurement accuracy based on this technology.Methods(1)Water-fat mixing model:a fat emulsion solution with a concentration range of 2%to 30%and a concentration gradient of 2%,and a water-fat solution with a concentration range of 0%to 100%and a concentration gradient of 10%,respectively,using CUBE-FSE-FLEX and IDEAL-IQ sequence scans of the hydrolipid solution and fat emulsion solution model to obtain the fat fractions measured in the two groups of sequences.The paired t test was used to compare the differences between the measured values and the actual fat fractions respectively.The linear regression method and The Bland-Altman scatter plot analyzes the correlation and consistency between the two groups of measured values and the actual fat concentration;(2)Fat and lean meat model:The CUBE-FSE-FLEX sequence was used to scan the fat and lean meat model,and manual segmentation(MS)and semi-automatic segmentation algorithm,multi-dimensional threshold method(MDT)and regional growing method(RG)were used to measure the volume of fat and lean samples,using paired t test.Pearson correlation method and Bland-Altman Scatter plots were used to analyse the differences,correlations,and consistency of volume measurement values with actual volume in three methods.Pearson correlation method and Bland are used for measured values of MDT used in fat and lean meat phantoms-Altman scatter plot analysis.Results(1)Water-fat solution:There was no significant difference between CUBE-FSE-FLEX-FaF and actual fat concentration(t=0.113,P=0.913),and IDEAL-IQ-FaF was significantly different from the actual concentration(t=-41.084,P<0.001).The two measurements are highly linearly related to the actual fat concentration.The values of R^2 were 0.941(CUBE-FSE-FLEX)and 1(IDEAL-IQ),respectively.The 95%confidence interval for the difference between the
作者
刘欢
张菁
盛晓兰
熊蝶
戴丽卉
汤翔宇
王秋霞
LIU Huan;ZHANG Jing;SHENG Xiaolan(Department of Radiology,the Affiliated Tongji Hospital to Tongji Medical College,Huazhong University of Science and Technology,Wuhan,Hubei Province 430030,P.R.China)
出处
《临床放射学杂志》
CSCD
北大核心
2020年第1期169-176,共8页
Journal of Clinical Radiology
基金
国家自然科学基金项目(编号:81771793、81301192)
湖北省自然科学基金项目(编号:2017CKB900)。
关键词
体外模型
容积测量
脂肪定量
对比研究
In vitro model
Measurement of volume
Fat quantification
Comparative study
