基于灰质体积共变模式的正常男性丘脑分割  

Parcellation of the thalamus based on gray matter volume covariance in healthy males

作　　者：李颖[1] 激扬[1] 王雪莉[1] LI Ying;JI Yang;WANG Xueli(Department of Radiology,the First Affiliated Hospital of Xi'an Medical Uni-uersity ,Xi'an 710077,China)

机构地区：[1]西安医学院第一附属医院影像科,陕西西安710077

出　　处：《实用放射学杂志》2019年第3期345-348,共4页Journal of Practical Radiology

基　　金：陕西省自然科学基础研究计划项目(2014JM2-8195)。

摘　　要：目的探讨使用灰质体积共变(GMVC)模式方法划分健康成年男性丘脑亚区的可行性和准确性。方法103名男性健康被试[年龄19~50岁,平均(35.5±7.7)岁,受教育程度6~16年,平均(10.5±2.3)年]采集T13D结构像,数据经CAT12标准流程预处理(配准、分割、标准化、平滑等)得到各体素的灰质体积,计算双侧丘脑各体素与全脑其余体素间基于GMVC矩阵,以Cramer’s V指标结合抽样交叉验证法确定丘脑最佳分区数并用增强型K均值聚类法做划分。结果左、右侧丘脑最优划分为11个和13个亚核团,双侧均得到前核、腹前核、外背侧核、外后侧核、腹外侧核、腹后内侧核、腹后外侧核、中线核、内背侧核以及枕核;其中右侧丘脑前核分为背侧和腹侧,中线核分为内侧和外侧,外背侧和外后侧核团合并为外侧核团群。左侧丘脑外后侧核分为前、后侧2部分。亚核团分割结果吻合解剖上14个亚核团的构造。结论GMVC模式方法活体划分的丘脑亚核团与解剖分割一致性强,为丘脑解剖、功能研究提供方法,为以丘脑为靶点的治疗提供定位依据。Objective To explore the feasibility and accuracy of segmentation of the thalamus in healthy males with gray matter volume covariance (GMVC) covariance pattern. Methods T1 3D structural MRI data were collected in 103 male participants [age from 19 to 50 y, mean=(35.5 + 7.7) y,education from 6 to 16 y,mean=(10.5± 2.3) y]. After preprocessed by CAT12 software with standard procedure ( registration, segmentation, modulation and smooth ), the voxel-wise bilateral thalamus gray matter volume (GMV) and the whole brain's voxel-wise GMV data were extracted to calculate the covariance matrix of all participants.Extended K-means clustering algorithm was applied to segment thalamus into sub-areas. Random sub-sampling and crossing-validation technique were used to generate 100 sub-datasets and verify the similarity index.Cramers'V (CV) was calculated and the K relevant to the highest mean of CV was determined as the optimal number of sub-areas for thalamus. Results The left and right thalamus were optimally segmented into 11 and 13 sub-areas by GMV covariance pattern.Both the left and right thalamus comprises Anterior Nuclear,ventral anterior nucleus,lateral dorsal nucleus,lateral posterior nucleus,ventral lateral nucleus, ventral posterior medial nucleus, ventral posterior lateral nucleus, midline nucleus, medial dorsal nucleus and pulvinar. Additionally, for the right thalamus, anterior nuclear group was divided into 2 parts, dorsal part and ventral part,and the midline nucleus into medial part and lateral part. The lateral dorsal and lateral posterior nucleus were merged into lateral nuclear group.For the left thalamus, the lateral posterior nucleus was further divided into 2 parts, anterior part and posterior part.The segmentation results were almost a line with the an atomical 14 partition of thalamus.Conclusion The parcellation based on GMVC covariance pattern is an effective and reliable tool for segmentation of brain gray matter nucleus. Our findings provide experiment basis for thalamus functional and structural st

关 键 词：灰质体积共变 K均值聚类 交叉验证 丘脑 分割 

分 类 号：R322.81[医药卫生—人体解剖和组织胚胎学] R445.2[医药卫生—基础医学]