Pimentel J A, Corkidi G
Engineering and Computer Science, Universidad Nacional Autónoma de México, 04510 CU Distrito Federal, México.
Annu Int Conf IEEE Eng Med Biol Soc. 2010;2010:4777-80. doi: 10.1109/IEMBS.2010.5626633.
In this work, we describe a segmentation cell method oriented to deal with experimental data obtained from 3D+t microscopical volumes. The proposed segmentation technique takes advantage of the pattern of appearances exhibited by the objects (cells) from different focal planes, as a result of the object translucent properties and its interaction with light. This information allows us to discriminate between cells and artifacts (dust an other) with equivalent size and shape that are present in the biological preparation. Using a simple correlation criteria, the method matches a 3D video template (extracted from a sample of cells) with the motile cells contained into the biological sample, obtaining a high rate of true positives while discarding artifacts. In this work, our analysis is focused on sea urchin spermatozoa cells but is applicable to many other microscopical structures having the same optical properties.
在这项工作中，我们描述了一种面向处理从3D+t显微镜体积获得的实验数据的分割细胞方法。所提出的分割技术利用了由于物体（细胞）的半透明特性及其与光的相互作用而在不同焦平面上呈现的物体（细胞）外观模式。这些信息使我们能够区分生物制剂中存在的大小和形状相同的细胞和伪像（灰尘等）。该方法使用简单的相关性标准，将3D视频模板（从细胞样本中提取）与生物样本中包含的活动细胞进行匹配，在丢弃伪像的同时获得了很高的真阳性率。在这项工作中，我们的分析集中在海胆精子细胞上，但也适用于许多具有相同光学特性的其他微观结构。
