Automated recognition of intracellular organelles in confocal microscope images
Automated recognition of intracellular organelles in confocal microscope images
复制标题
DOI：
10.1034/j.1600-0854.2002.30109.x
复制
发表时间：
2002-01-01
期刊：
影响因子：
4.5
通讯作者：
Hayes, B
中科院分区：
文献类型：
--
作者：
Danckaert, A;Gonzalez-Couto, E;Hayes, B
识别细胞内蛋白质的定位对于了解它们的功能是必不可少的。它通常是通过了解细胞生物学专家对特征良好的细胞内细胞器的分布情况并与之进行比较而得出的。我们自动化了这一过程，以实现对细胞内蛋白质分布的更客观和定量的评估，这可以由经验较少的细胞生物学家使用，并可能被用作经验不足的用户的培训程序，或作为功能分析中新基因的高通量定位程序。在这里，我们描述了一种基于模块化神经网络的分类系统的开发和测试，该系统通过荧光染色的细胞系中的几组共聚焦切片来训练，以寻找关键细胞内结构的标记。尽管单个细胞之间的模式不同，但系统运行良好，准确率为97%，这给了我们对该方法及其未来发展的信心。预计该计划将有助于设计进一步的实验，利用共定位与已知的细胞器标记蛋白，以确定假定的运输途径和蛋白质相互作用的目的蛋白。
Recognition of the localisation of intracellular proteins is essential to the understanding of their function. It is usually made through knowledge of and comparison to the distribution of well-characterised intracellular organelles by experts in cell biology. We have automated this process in order to achieve a more objective and quantitative assessment of the protein distribution within the cell, which ran be employed by the less experienced cell biologist and may be utilised as a training program for inexperienced users, or as a high throughput localisation program for novel genes in functional analysis. Here we describe the development and testing of a classification system based on a modular neural network trained with sets of confocal sections through cell lines fluorescently stained for markers of key intracellular structures. The system functioned well in spite of the variability in pattern that occurs between individual cells and performed with 97% accuracy, which gives us confidence in the method and in its future development. It is envisaged that this program will aid the design of further experiments utilising colocalisation with known organelle marker proteins, in order to confirm putative trafficking pathways and protein-protein interactions of the protein of interest.