Similarity analysis of cell movements in video microscopy
Similarity analysis of cell movements in video microscopy
复制标题
DOI：
10.1109/biovis.2012.6378595
复制
发表时间：
2012-10
期刊：
影响因子：
--
通讯作者：
Jens Fangerau;Burkhard Hockendorf;J. Wittbrodt;H. Leitte
中科院分区：
文献类型：
--
作者：
Jens Fangerau;Burkhard Hockendorf;J. Wittbrodt;H. Leitte
现代3D+T视频显微镜技术使生物学家能够在时间和空间上以前所未有的分辨率获取生物体的数据。这些数据集包含大量生物相关和可量化的信息，例如复杂生物体中所有单个细胞的运动。然而，这些信息的提取、验证和分析既具有挑战性又耗时。在本文中，我们提出了一种计算技术，分类和验证生物体中的细胞运动和细胞分裂的相似模式，包括多达数千个细胞。我们的算法确定跟踪路径的跟踪细胞表现出相似的功能和形状结构。这些相似性值被分配给我们的聚类算法，聚类路径成组的连贯行为。这些数据可以在2D投影和3D细胞运动表示中进行交互式探索。这是第一次，这种可视化允许生物学家在整个生物体的规模上详尽地评估分裂模式和细胞迁移的相似性和差异。为了验证，我们将我们的方法应用于一个合成数据集和两个真实的数据集，包括斑马鱼时期从囊胚期到早期外包和生长斑马鱼尾巴。我们表明，我们的方法成功地检测基于形状和细胞运动的特征的相似性。
Modern 3D+T video microscopy techniques enable biologists to acquire data of living organisms with unprecedented resolution in time and space. These datasets contain a wealth of biologically relevant and quantifiable information, e.g. the movements of all individual cells in a complex organism. However, extraction, validation, and analysis of this information are both challenging and time-consuming. In this paper, we present a computational technique that classifies and validates similar patterns of cell movements and cell divisions in organisms that consist of up to thousands of cells. Our algorithm determines tracking paths of traced cells that exhibit similar features and shape structures. These similarity values are assigned to our cluster algorithm that clusters paths into groups of coherent behavior. The data can be interactively explored in 2D projections and a 3D cell movement representation. For the first time, this visualization allows biologists to exhaustively assess similarities and differences in division patterns and cell migration on the scale of an entire organism. For validation, we applied our method on a synthetic dataset and two real datasets including zebrafish periods from blastula stage to early epiboly and growing zebrafish tail. We show that our method succeeds in detecting similarities based on shape and cell-movement based features.