Xiao J, Christen R, Minimo C, Bartels P H, Bibbo M
Department of Pathology and Cell Biology, Jefferson Medical College, Thomas Jefferson University, Philadelphia, Pennsylvania 19107-5244.
Anal Quant Cytol Histol. 1994 Aug;16(4):240-6.
An image segmentation algorithm, based on boundary tracking, was introduced to achieve automatic segmentation of nuclei. This will improve reliability and reproducibility for the computer-assisted grading of routinely stained material, especially from biopsies, which often offer only scanty clinical material. Nuclear grading systems using karyometric features were developed earlier. However, hematoxylin and eosin-stained tissues have proven difficult for automatic segmentation, which is a crucial part of an objective grading system. In this paper we describe an automatic tracking method that traces nuclear boundaries on the basis of edge information and local boundary features. There were two phases to the procedure. First, approximate boundaries were extracted by automatic thresholding; then, boundaries were refined through interactive tracking. The results are encouraging.
引入了一种基于边界跟踪的图像分割算法，以实现细胞核的自动分割。这将提高对常规染色材料（尤其是活检材料，其临床材料通常很少）进行计算机辅助分级的可靠性和可重复性。早期开发了使用核测量特征的核分级系统。然而，苏木精和伊红染色的组织已被证明难以进行自动分割，而自动分割是客观分级系统的关键部分。在本文中，我们描述了一种基于边缘信息和局部边界特征来追踪核边界的自动跟踪方法。该过程有两个阶段。首先，通过自动阈值提取近似边界；然后，通过交互式跟踪对边界进行细化。结果令人鼓舞。
