Li Simon, Buehnemann Claudia, Hassan Bass, Noble J
Department of Engineering, University of Oxford, UK.
Annu Int Conf IEEE Eng Med Biol Soc. 2010;2010:4813-6. doi: 10.1109/IEMBS.2010.5628032.
The increasing amounts of microscopy data generated in cell biology requires the development of automated tools for the quantitative analysis of images. Clumps of cells are difficult to segment due to the frequent lack of clear boundaries between cells and are often ignored, but communication between cells is an intrinsic part of the response of cells to their environment. In addition cells often show a large variation in their responses, even within a clump, and an accurate segmentation is therefore vital to prevent the unwanted averaging of measurements over multiple cells. Here we present a method for segmenting clumps of cells by using a multi-scale ridge filter to enhance unclear boundaries. A multi-phase level set method incorporating a region competition term is used to identify a boundary for each cell based on the ridge filter response.
细胞生物学中生成的显微镜数据量不断增加，这就需要开发用于图像定量分析的自动化工具。由于细胞之间常常缺乏清晰的边界，细胞团块很难进行分割，并且常常被忽略，但是细胞间的通讯是细胞对其环境做出反应的固有组成部分。此外，即使在一个细胞团块内，细胞的反应也常常表现出很大的差异，因此准确的分割对于防止在多个细胞上进行不必要的测量平均化至关重要。在此，我们提出一种通过使用多尺度脊线滤波器来增强不清晰边界的方法来分割细胞团块。一种结合区域竞争项的多相水平集方法用于根据脊线滤波器响应为每个细胞识别边界。
