Continuing advances in computer architecture and data analysis are making imaging an increasingly attractive option for tracking bioprocesses. “The development of various deep learning architectures and training methods suitable for image processing, and the advent of high-performance GPUs (graphics processing units) that can process them, have enabled the application of more advanced image processing techniques,” said Sang-Kyu Jung, PhD, associate professor of biochemical engineering at Hongik University in Sejong, South Korea.
Today, image analysis can give scientists much more information about the state of bioprocesses. As Jung explained, image analysis with machine learning “can now be used to classify objects according to their properties and predict important phenomena.”
CHO cells
As an example, a team of scientists from the Beijing Key Laboratory of Enze Biomass Fine Chemicals in China applied image analysis to Chinese hamster ovary (CHO) cells in culture. Specifically, the scientists used a deep learning approach to analyze images of these cells. In addition to providing online, real-time information, the scientists noted that “precise and automatic statistics of geometric properties facilitate the optimization and control of the culture process.”
Various imaging techniques are available for bioprocess intensification. For example, Dr. Jean-Sébastien Guez, senior scientist at the Institut Pascal in France, and his colleagues used in situ microscopy to analyze antibody production by mammalian cells in bioreactors. The scientists concluded that “analysis of these images with AI-based methods provided estimates of cell density and viability with very high accuracy, even under suboptimal culture conditions.”
Imaging-based methods in bioprocessing will continue to attract attention in the future.16Number For example, at the Bioprocessing Summit in Boston, Dr. Theodore Randolph, professor of chemical and biological engineering at the University of Colorado, will speak about using machine learning to analyze images for the formation of aggregates during bioprocessing.
In the future, imaging will be used to analyze even more features of bioprocesses in real time, leading to more efficient and productive biomanufacturing.
