Amgen, Inc., Thousand Oaks, 91320, CA, USA.
Illinois Institute of Technology, Chicago, 60616, IL, USA.
Pharm Res. 2024 Aug;41(8):1659-1669. doi: 10.1007/s11095-024-03749-4. Epub 2024 Jul 24.
In biotechnology, microscopic cell imaging is often used to identify and analyze cell morphology and cell state for a variety of applications. For example, microscopy can be used to detect the presence of cytopathic effects (CPE) in cell culture samples to determine virus contamination. Another application of microscopy is to verify clonality during cell line development. Conventionally, inspection of these microscopy images is performed manually by human analysts. This is both tedious and time consuming. In this paper, we propose using supervised deep learning algorithms to automate the cell detection processes mentioned above.
The proposed algorithms utilize image processing techniques and convolutional neural networks (CNN) to detect the presence of CPE and to verify the clonality in cell line development.
We train and test the algorithms on image data which have been collected and labeled by domain experts. Our experiments have shown promising results in terms of both accuracy and speed.
Deep learning algorithms achieve high accuracy (more than 95%) on both CPE detection and clonal selection applications, resulting in a highly efficient and cost-effective automation process.
在生物技术中，显微镜细胞成像常用于识别和分析细胞形态和细胞状态，适用于多种应用。例如，显微镜可用于检测细胞培养样本中是否存在细胞病变效应（CPE），以确定病毒污染情况。显微镜的另一个应用是在细胞系开发过程中验证克隆性。传统上，这些显微镜图像的检查是由人工分析师手动完成的。这既繁琐又耗时。在本文中，我们提出使用监督深度学习算法来自动化上述细胞检测过程。
所提出的算法利用图像处理技术和卷积神经网络（CNN）来检测 CPE 的存在，并验证细胞系开发中的克隆性。
我们使用经过领域专家收集和标记的图像数据来训练和测试算法。我们的实验在准确性和速度方面都取得了有希望的结果。
深度学习算法在 CPE 检测和克隆选择应用中均实现了高精度（超过 95%），从而实现了高效且具有成本效益的自动化过程。
