Cell Segmentation: 50 Years Down the Road
Cell Segmentation: 50 Years Down the Road
复制标题
DOI：
10.1109/msp.2012.2204190
复制
发表时间：
2012-09-01
影响因子：
14.9
通讯作者：
Meijering, Erik
中科院分区：
文献类型：
--
作者：
Meijering, Erik
自从19世纪初承认细胞是生命的基本构建单位的细胞理论建立以来，生物学家一直试图解释其潜在的原理。在几十年的研究过程中取得了重大发现[1]，但对充分了解细胞机制以及如何操纵它们以改善健康的追求一直延续到今天，比以往任何时候都更大的预算、更多的思维和更复杂的工具。细胞生物学的许多进步可以归功于光学显微镜[2]，这是一种工具。自从安东尼·范·列文霍克斯在17世纪70年代第一次改进和利用显微成像在细胞水平上研究生命以来，这个领域已经走过了很长的路。今天的生物学家不仅拥有大量不同的、互补的显微成像技术，使他们能够可视化现象，甚至远低于经典的光的衍射极限，先进的显微镜系统还允许他们在短短几个小时内轻松获取大量图像。现代成像实验中产生的数据的丰富性、异质性、维度和复杂性排除了手动图像管理、处理和分析的可能性。因此，用于执行这些任务的计算机化技术对于细胞生物学的进一步发展变得至关重要。图像分割是许多研究中的一个中心问题，也是图像分析的基石。具体地说，由于细胞形态是指示细胞的生理状态的重要表型特征，而且由于细胞轮廓通常需要用于随后的细胞内过程分析(放大到纳米尺度)或细胞社会学分析(缩小到毫米尺度)，因此细胞分割问题在过去几年中受到越来越多的关注[7]。在这里，我们反思这个领域多年来是如何演变的，以及过去的事态发展如何推论到未来。
Ever since the establishment of cell theory in the early 19th century, which recognized the cell as the fundamental building unit of life, biologists have sought to explain the underlying principles. Momentous discoveries were made over the course of many decades of research [1], but the quest to attain full understanding of cellular mechanisms and how to manipulate them to improve health continues to the present day, with bigger budgets, more minds, and more sophisticated tools than ever before. One of the tools to which a great deal of the progress in cell biology can be attributed is light microscopy [2]. The field has come a long way since Antoni van Leeuwenhoeks first steps in the 1670s toward improving and exploiting microscopic imaging for studying life at the cellular level. Not only do biologists today have a plethora of different, complementary microscopic imaging techniques at their disposal that enable them to visualize phenomena even way below the classical diffraction limit of light, advanced microscope systems also allow them to easily acquire very large numbers of images within just a matter of hours. The abundance, heterogeneity, dimensionality, and complexity of the data generated in modern imaging experiments rule out manual image management, processing, and analysis. Consequently, computerized techniques for performing these tasks have become of key importance for further progress in cell biology [3][6]. A central problem in many studies, and often regarded as the cornerstone of image analysis, is image segmentation. Specifically, since cellular morphology is an important phenotypic feature that is indicative of the physiological state of a cell, and since the cell contour is often required for subsequent analysis of intracellular processes (zooming in to nanoscale), or of cell sociology (zooming out to millimeter scale), the problem of cell segmentation has received increasing attention in past years [7]. Here we reflect on how the field has evolved over the years and how past developments can be expected to extrapolate into the future.