Measuring the velocity of fluorescently labelled red blood cells with a keyhole tracking algorithm
Measuring the velocity of fluorescently labelled red blood cells with a keyhole tracking algorithm
复制标题
DOI：
10.1111/j.1365-2818.2007.01877.x
复制
发表时间：
2008-01-01
影响因子：
2
通讯作者：
Tozer, G. M.
中科院分区：
文献类型：
--
作者：
Reyes-Aldasoro, C. C.;Akerman, S.;Tozer, G. M.
在本文中，我们提出了一种跟踪算法来测量荧光标记的红细胞穿过肿瘤微血管的速度，生长在背部皮瓣窗口室，植入小鼠。预处理从图像中去除噪声和伪影，然后从背景中分割细胞。跟踪算法是基于一个“锁眼”模型，描述了一个视频序列的连续帧之间的分割细胞的可能的运动。当存在细胞移动的历史时，细胞的过去、现在和预测的着陆位置将定义类似于钥匙孔形状的两个概率区域。此锁眼模型用于确定相邻帧中的单元是否应链接以形成轨迹，也可作为后处理工具来连接分割轨迹并丢弃可能由于噪声或不确定性而形成的链接。当没有历史记录时，围绕父细胞的质心的圆形区域被用作概率区域。基于轨迹的平均速度的分布去除异常值。由于每个细胞的位置和时间都被记录下来，因此可以从轨迹中获得大量的统计测量。该算法进行了测试，两组实验。首先，分析了8个不同几何形状的肿瘤的血管，平均速度范围为86至372 μ m s（-1），最小和最大轨迹速度分别为7和1212 μ m s（-1）。其次，在对两个肿瘤施用血管阻断剂后进行速度的纵向研究，并分析24小时内的时间行为。在其中一个肿瘤中，脉管系统完全关闭，而在另一个肿瘤中，在30分钟时速度明显降低，随后在6小时时恢复。该跟踪算法能够在活体视频序列内同时测量多条血管中的红细胞速度，从而能够分析癌症小鼠模型中的血流异质性和对治疗的反应。
In this paper we propose a tracking algorithm to measure the velocity of fluorescently labelled red blood cells travelling through microvessels of tumours, growing in dorsal skin flap window chambers, implanted on mice. Preprocessing removed noise and artefacts from the images and then segmented cells from background. The tracking algorithm is based on a 'keyhole' model that describes the probable movement of a segmented cell between contiguous frames of a video sequence. When a history of cell movement exists, past, present and a predicted landing position of the cells will define two regions of probability that resemble the shape of a keyhole. This keyhole model was used to determine if cells in contiguous frames should be linked to form tracks and also as a postprocessing tool to join split tracks and discard links that could have been formed due to noise or uncertainty. When there was no history, a circular region around the centroid of the parent cell was used as a region of probability. Outliers were removed based on the distribution of the average velocities of the tracks. Since the position and time of each cell is recorded, a wealth of statistical measures can be obtained from the tracks. The algorithm was tested on two sets of experiments. First, the vasculatures of eight tumours with different geometries were analyzed; average velocities ranged from 86 to 372 mu m s(-1), with minimum and maximum track velocities 7 and 1212 mu m s(-1), respectively. Second, a longitudinal study of velocities was performed after administering a vascular disrupting agent to two tumours and the time behaviour was analyzed over 24 h. In one of the tumours there is a complete shutdown of the vasculature whereas in the other there is a clear decrease of velocity at 30 min, with subsequent recovery by 6 h. The tracking algorithm enabled the simultaneous measurement of red blood cell velocity in multiple vessels within an intravital video sequence, enabling analysis of heterogeneity of flow and response to treatment in mouse models of cancer.