北京生物医学工程

用于幼儿白质纤维束分析的快速密度聚类算法

Fast density-peaks clustering for pediatric white matter tract analysis

作者：段煜茁樊鑫程世超张宇夕程华

单位：大连理工大学软件学院(辽宁大连 116620) 首都医科大学附属北京儿童医院医学影像中心(北京100045)

关键词：快速聚类；白质纤维束；弥散张量成像；二叉树；幼儿纤维发育

分类号：R318.04

出版年·卷·期（页码）：2019·38·2（134-144）

摘要：

目的从弥散张量成像(diffusion tensor imaging,DTI)中聚类白质(white matter,WM)纤维束对于定量分析幼儿大脑发育是非常重要的中间步骤。经典的基于图谱的分析方法通常是针对经过配准算法转化到通用模板空间之后的空间平均弥散张量做分析,其中配准算法不可避免地平滑了纤维束的局部特性而带来误差。更为直观可行的方法是首先将交织的白质纤维分离成单一类型的纤维束,然后在本地空间对来自不同主体的同一类型的纤维束进行统计分析,以避免来自配准算法的平滑误差。最近发表的密度峰值聚类(density peaks clustering,DP)算法,在没有任何现有模板的情况下聚类结构复杂的白质纤维依然具有很强的鲁棒性。尽管如此,密度的计算是DP的核心步骤,特别是当纤维数量巨大时,计算非常耗时。方法首先本文提出一个快速密度峰值聚类算法(fast density peaks clustering,FDP),将过程中的密度和关键参数的全局计算转换成局部计算,并利用一种二叉树结构来有序地存储这些用于局部计算的近邻,从而约50倍地加速了密度计算步骤。然后通过在基准数据集上的实验和与经典聚类算法的纤维聚类效果对比验证了FDP的加速效果和精确度。最后,证明了本算法在幼儿纤维发育分析中应用的合理性与可行性。结果密度计算基于上述结构的直接访问,因此显著降低了计算量;基于该算法的纤维数据统计结果也与基于图谱分析方法的一般结论一致。结论本文提出的快速密度聚类算法具备快速、准确和鲁棒的优势,其在幼儿白质纤维发育分析中的应用合理地降低了传统方法中繁杂的人工和计算消耗。

Objective Clustering of white matter (WM) tracts from diffusion tensor imaging (DTI) is an important intermediate step for quantitative analysis on pediatric brain development. The classical atlas-based method typically analyzes the spatially averaged DTI quantities transformed into a common template space by registration algorithms. A more intuitive and feasible method is to separate the entangled WM fiber into onesingle type of fiber bundle,and then perform statistical analysis on a common type of bundles from different subjects in the native space so as to circumvent the smoothing from registration algorithms. The recently published density peaks (DP) clustering algorithm still demonstrates strong robustness in clustering WM fibers without any priori. However,density calculation is the core step of DP,especially when the number of fibers is huge,which is extremely time-consuming. Methods Firstly,we propose a fast density peaks (FDP) clustering algorithm which accelerates the density calculation about 50 times. We transform the global computation of density and key parameters into local calculation,and develop a binary tree structure to store these neighbors for local computation in an orderly way. Additionally,with experiments on the benchmark data set and comparing results of our FDP algorithm and existing clustering methods,we can validate the efficiency and effectiveness of FDP. Finally,we prove the feasibility and rationality of the proposed algorithm in the analysis on pediatric WM tract development. Results Therefore,the density calculation results are based on the direct access of the above structure,leading significantly computational reducing; the statistical results based on FDP are also consistent with the general conclusions of traditional analysis methods. Conclusions The FDP algorithm has the advantages of rapidness, accuracy and robustness. Its application in the analysis of pediatric WM fiber development achieves common result of atlas based methods without such expensive manual labors and computational loads.

参考文献：

[1]Basser PJ, Mattiello J, Lebihan D. MR diffusion tensor spectroscopy and imaging[J].Biophysical Journal,1994,66(1):259-267.

[2]Ciulli S, Citi L, Salvadori E, et al. Prediction of impaired performance in Trail Making Test in MCI patients with small vessel disease using DTI data[J].IEEE Journal of Biomedical & Health Informatics,2016,20(4):1026-1033.

[3]Smith SM, Jenkinson M, Johansen - Berg H, et al. Tract - based spatial statistics:voxelwise analysis of multi-subject diffusion data [J].NeuroImage,2006,31(4):1487-1505.

[4]Aghajani M, Veer IM, van Lang ND, et al. Altered white - matter architecture in treatment-naive adolescents with clinical depression.[J].Psychological Medicine,2013,44(11):2287-2298.

[5]Bashat DB,Granot Halevy R,Weinstein M,et al. Abnormal white matter integrity in young children with autism and correlation with behavior[J].International Journal of Developmental Neuroscience,2012,32(4):667-667.

[6]Liu X,Lai Y,Wang X,et al. Reduced white matter integrity and cognitive deficit in never medicated chronic schizophrenia: a diffusion tensor study using TBSS[J].Behavioural Brain Research,2013,252:157-163.

[7]Lebel C,Gee M,Camicioli R, et al. Diffusion tensor imaging of white matter tract evolution over the lifespan [J].NeuroImage, 2012,60(1):340-352.

[8]Ouyang M, Cheng H, Mishra V, et al. Atypical age dependent effects of autism on white matter microstructure in children of 2-7 years[J].Human Brain Mapping,2016,37(2):819-832.

[9]Smith SM,Johansen Berg H,Jenkinson M,et al. Acquisition and voxelwise analysis of multi-subject diffusion data with tract based spatial statistics[J].Nature Protocols,2007,2(3):499-503.

[10]Jin Y, Huang C, Daianu M, et al. 3D tract specific local and global analysis of white matter integrity in Alzheimer ＇ s disease[J].Human Brain Mapping,2017,38(3):1191-1207.

[11]O’ Donnell LJ,Golby AJ,Westin CF. Fiber clustering versus the parcellation based connectome[J].Neuroimage, 2013,80(8):283-289.

[12]O’ Donnell LJ, Westin CF. White matter tract clustering and correspondence in populations[M]//Medical Image Computing and Computer Assisted Intervention—MICCAI 2005. Berlin: Springer International Publishing,2005:140-147.

[13]Reichenbach A,Goldau M,Heine C,et al. V-bundles:clustering fiber trajectories from diffusion mri in linear time[C]//Medical Image Computing and Computer-Assisted Intervention—MICCAI 2005. Berlin:Springer International Publishing,2015:191-198.

[14]O’ Donnell LJ, Kubicki M, Shenton ME, et al. A method for clustering white matter fiber tracts[J].American Journal of Neuroradiology,2006,27(5):1032-1036.

[15]Wang Q, Yap PT, Wu G, et al. Application of neuroanatomical features to tractography clustering[J].Human Brain Mapping, 2012,34(9):2089-2102.

[16]Zhang T, Chen H, Guo L, et al. Characterization of U-shape streamline fibers: methods and applications[J].Medical Image Analysis,2014,18(5):795-807.

[17]Jin Y, etingül HE. Tractography-embedded white matter stream clustering[C]//2015 IEEE 12th International Symposium on Biomedical Imaging. New York,NY,USA:IEEE,2016:432-435.

[18]Shao J, Hahn K, Yang Q, et al. Hierarchical density-based clustering of white matter tracts in the human brain[J].International Journal of nowledge Discovery in Bioinformatics, 2010,1(4):1-25.

[19]Mai ST,Goebl S, Plant C. A similarity model and segmentation algorithm for white matter fiber tracts[C]//2012 IEEE 12th International Conference on Data Mining. Brussels, Belgium: IEEE Press,2013:1014-1019.

[20]Li H, Xue Z, Guo L, et al. A hybrid approach to automatic clustering of white matter fibers[J].NeuroImage,2010,49(2):1249-1258.

[21]Jin Y,Shi Y,Zhan L,et al. Automatic clustering of white matter fibers in brain diffusion MRI with an application to genetics[J].NeuroImage,2014,100:75-90.

[22]Rodriguez A,Laio A. Clustering by fast search and find of density peaks[J].Science,2014,344(6191):1492-1496.

[23]Chen P,Fan X,Liu R,et al. Fiber segmentation using a density-peaks clustering algorithm [C]//2015 IEEE 12th International Symposium on Biomedical Imaging. New York,USA:IEEE Press,2015:633-637.

[24]Kamali T, Stashuk D. Automated segmentation of white matter fiber bundles using diffusion tensor imaging data and a new density based clustering algorithm [J].Artificial Intelligence in Medicine,2016,73:14-22.

[25]Zvitia O,Mayer A,Shadmi R,et al. Co-registration of white matter tractographies by adaptive-mean-shift and Gaussian mixture modeling[J].IEEE Transactions on Medical Imaging,2010,29(1):132-145.

[26]Jiang H,van Zijl PCM,Kim J,et al. DtiStudio:resource program for diffusion tensor computation and fiber bundle tracking [J].Computer Methods and Programs in Biomedicine,2006,81( 2):106-116.

[27]Jenkinson M, Beckmann CF, Behrens TEJ, et al. FSL[J].NeuroImage,2012,62(2):782-790.

[28]Woods RP, Grafton ST, Watson JDG, et al. Automated image registration: II. Intersubject validation of linear and nonlinear models[J].Journal of Computer Assisted Tomography,1998,22(1):153-165.

[29]Smith SM. Fast robust automated brain extraction[J].Human Brain Mapping,2002,17(3):143-155.

[30]Basser PJ,Pajevic S,Pierpaoli C,et al. In vivo fiber tractography using DT-MRI data[J].Magnetic Resonance in Medicine,2000, 44(4):625-632.

[31]Mayer A,Zimmerman Moreno G, Shadmi R, et al. A supervised framework for the registration and segmentation of white matter fiber tracts[J].IEEE Transactions on Medical Imaging,2011,30(1):131-145.

[32]Wakana S,Caprihan A,Panzenboeck MM,et al. Reproducibility of quantitative tractography methods applied to cerebral white matter[J].NeuroImage,2007,36(3):630-644.

[33]Johnson RT,Yeatman JD,Wandell BA,et al. Diffusion properties of major white matter tracts in young, typically developing children[J].NeuroImage,2014,88:143-154.

[34]Hermoye L,Saint Martin C,Cosnard G,et al. Pediatric diffusion tensor imaging: normal database and observation of the white matter maturation in early childhood[J].NeuroImage,2006,29(2):493-504.

服务与反馈：

【文章下载】【加入收藏】

提示：您还未登录，请登录！点此登录