Diffusion kurtosis imaging ( DKI ) , introducing a higher-order kurtosis tensor to quantify the non-Gaussian property of water diffusion as well as multiple diffusion weighted factor b-values to model fit-ting, is a powerful probe for assessment of the microstructure of biological tissues, enjoying a unique ad-vantage in neuroscience and clinic diagnosis.However, there is still a lack of comparative analysis on different features of microstructure of brain using different b-value combinations in DKI data acquisition and thoughts of optimizing b-value combinations of DKI data acquisition according to different clinical needs.In this paper, DKI image contrast of cerebrospinal fluid, gray matter and white matter and voxel-based analysis of the whole brain were used to compare the influence of DKI parameter under 5 b-values scheme and 2 b-values scheme and the differences of corpus callosum local tissue microstructure charac-teristic expression.Results show that the combination of low b-value (<1 000 s/mm2 ) and high b-value (>2 000 s/mm2 ) DKI scheme has great resolution ability for tissues, while more b-values collected at the same time increase the acquisition time and the model fitting error as well;2 b-values scheme ( 1 000 s/mm2 , 2 000 s/mm2 ) will be more suitable for general DKI clinical diagnosis.%扩散峭度成像（ DKI）引入高阶峭度张量来量化水分子扩散的非高斯程度，并可采用多种扩散敏感因子b值参与模型拟合，能反映组织复杂结构更微小变化，因而在脑神经科学研究与临床诊断中具有独特优势。但目前尚缺乏不同b值组合DKI数据采集对不同脑微结构特征的比较分析和针对不同临床需求制定DKI数据采集的最优b值组合方案的思路。本文从脑脊液、灰质和白质的DKI图像对比度角度，结合基于体素统计分析方法比较了5个不同b值组合与高、低2个b值组合采集方案对DKI成像指标的影响及在胼胝体局部组织微结构特征表达的差异。结果表明，低b值（＜1000 s／mm2）和高b值（＞2000 s／mm2）采集成像皆有较强的组织分辨能力，而过多增加b值组合个数会增大拟合误差并增加采集时间；使用高、低2个b值（2000 s／mm2与1000 s／mm2）的组合采集方案较适合于一般DKI临床诊断。
展开▼
