基于流场分布的纤维续跟踪

Fiber Re-Tracking Based on Flow Field Distribution

白质纤维跟踪能够重建大脑纤维走向,但由于分辨率的限制使得成像体素呈现出锯齿形边界,导致重建纤维束在白质区域边界提前终止。此外,信号噪声引起的体素内方向信息缺失也同样会导致纤维跟踪过早终止。针对以上问题,本研究提出一种纤维停止点续跟踪算法使纤维跟踪结果更加准确。首先,建立纤维局部结构的流场分布模型,以流场分布表示体素内任意点的纤维方向信息;然后,基于体素流场分布模型计算纤维跟踪停止点的连续性进行纤维续跟踪,得到更为准确的纤维结果。分别采用两组实验数据集(包含25个主要纤维束的ISMRM 2015挑战数据集和一组来自斯坦福大学数据库的临床数据集)对提出的方法进行验证。首先,在模拟数据集上通过Tractometer量化指标对所提出算法进行定量分析,结果表明,该算法的有效连接束数量(VB),有效连接率(VC)和无效连接率(NC)分别为24、51.3%和12.9%,与同为确定型追踪算法的SD_Stream相比,VB提高2束,VC提高19.1%,NC降低22.8%。其次,通过临床数据集对所提出算法进行定性分析,进一步验证所提出算法的有效性。实验结果表明,基于流场分布的纤维续跟踪算法有效避免了纤维跟踪过早终止这一问题,提高了纤维跟踪的有效纤维比例,降低了无连接纤维比例,提升了纤维跟踪结果的准确性。

White matter fiber tracking can reconstruct the direction of brain fibers.However,due to the limitation of resolution,the imaging voxels show zigzag boundaries,leading to the early termination of the reconstructed fiber bundles at the white matter region boundary.In addition,the lack of voxel direction information caused by signal noise also leads to premature termination of fiber tracking.In view of these problems,a fiber stop tracking algorithm was proposed to obtain more accurate result of the fiber tracking.Firstly,the flow field distribution model of the fiber local structure was established,and the fiber direction information of any point in the voxel was expressed by the flow field distribution.Next,based on the voxel flow field distribution model,the continuity of the fiber tracking stop point was calculated for the fiber tracking,and more accurate fiber results were obtained.Two experimental data sets(including ISMRM 2015 data set of 25 major fiber bundles and a set of clinical data sets from Stanford University database)were used to verify the proposed method.Firstly,the proposed algorithm was quantitatively analyzed on the simulation data set by using the tractometer quantitative index,the results showed that the effective number of valid bundles(VB),valid connections(VC)and no connections(NC)of the algorithm were 24,51.3%and 12.9%,respectively.Compared with SD Stream,VB increased by 2 bundles,VC increased by 19.1%,and NC decreased by 22.8%.The effectiveness of this method was verified through the qualitative analysis of clinical data.The experimental results showed that the algorithm could effectively avoid the premature termination of fiber tracking,improving the effective fiber ratio of the fiber tracking,reducing the proportion of unconnected fibers,and improving the accuracy of fiber tracking results.