Lab Manager | Run Your Lab Like a Business

Article

Close up of scientist's hands working with cell culture products in a biosafety cabinet.

The Benefits of AI and Automation in the Cell Culture Process

Bringing in AI and automation can improve every step of cell culturing, leading to improved efficiency in research and drug development

by
Shantanu Dhamija

Shantanu Dhamija is the vice president of strategy and innovation at Molecular Devices.

ViewFull Profile.
Learn about ourEditorial Policies.
Register for free to listen to this article
Listen with Speechify
0:00
5:00

When scientists talk about the potential for artificial intelligence (AI) to enhance life sciences research and development, they often talk in futuristic terms. Indeed, the technology hasn’t yet reached its full potential but there’s one area where AI, coupled with automation, can greatly improve laboratory research today: cell culture.

Traditional cell culture workflows require 24/7, hands-on processes that are time consuming, prone to error, and hard to reproduce. Bringing in AI and automation can improve every step of cell culturing, leading to faster and better decision-making, ultimately improving efficiency in research and drug development.

Cell culture protocols are generally divided into multiple stages. While most of the work within a stage can be automated, manual inspection and review (generally via microscopy) are necessary at the end of nearly every stage. Advances in automated imaging and cloud storage have allowed remote inspection, but researchers still have to look at the images and data and decide what happens next. This need for routine monitoring and decision-making by a human has made end-to-end automation unachievable.

With AI, however, researchers can create customized workflows that follow the natural flow of the cell culture process without the need for modularization and human intervention. Machine learning (ML)-powered algorithms can trigger decision-making remotely, based on the algorithm training or parameters established by the scientists running the experiment. This can make tasks, like feeding and passaging cell cultures, autonomous—like how self-driving cars have the intelligence to determine when to slow down, speed up, or switch lanes based on information coming into the system. An alternative option would be a co-piloted approach. In this case, AI provides recommendations that are accepted or rejected by the scientist remotely, eliminating the need for 24/7 hands-on time and freeing researchers for other tasks.

Researchers can also program automated systems to send real-time alerts when milestones have been reached, or if certain events require immediate attention. While the technology is in its infancy, the opportunities are endless. Whichever way you go, automation with AI-driven decision-making helps standardize cell culture processes through the incorporation of objective, quantified metrics, while improving the consistency and reproducibility of the experiments being conducted across multiple sites.

Supporting 3D cell models

The ability to automate tasks improves 2D cell culture processes significantly but is a potential game-changer when it comes to the development of 3D cell models such as organoids. These models are particularly valuable in life sciences research because they are designed to reflect the biology of human organs accurately. However, these models, which often encompass multiple cell types, are also extraordinarily complex, requiring robust culture protocols that are highly reliant on the knowledge and judgment of the scientific staff performing this work.

Automation and AI make it easier to maintain and monitor 3D cell cultures round-the-clock, ensuring that they are developing correctly and preventing them from being damaged or aspirated during feeding. AI can also enable the development of methods for scaling up 3D models in ways that ensure consistency and reproducibility. Furthermore, it allows scientists to identify outliers at the well, plate, or experiment level, and exclude these from the experiment completely. Finally, researchers can grow spheroids or organoids from different cell sources simultaneously.

Imaging inside

AI-driven decision-making is predicated on the ability to obtain accurate information that guides these decisions. Visual inspection via microscopy has been the norm for decision-making in manual cell culture. Advances in computer vision and ML have enabled the accurate quantification of both basic characteristics (e.g., confluence) and complex morphology (e.g., organoid crypts), which is used for monitoring culture health and decision-making.

One area where automated imaging can be particularly valuable is in automated quality control. When organoids are grown for a long period of time, there is a higher likelihood of contamination with mold or fungus. AI can detect that contamination with image analysis, running quietly in the background, having been trained to look for the common phenotypes of contamination. If it sees fungus or mold in the culture, it can notify the scientist of a potential contamination event, and identify all plates that were in close proximity to the infected plate, enabling effective quarantining, and thus preventing cross-contamination.

AI in action

Cell cultures for cancer research are a prime example of the benefits of AI and automation. Oncologists are increasingly building 3D cancer spheroids—clusters of cells that provide insights into physiological changes and the tumor microenvironment. With automated cell-culture workflows, researchers can start with 2D cultures of tumor cells and move seamlessly to cell differentiation, spheroid formation, and maintenance. From there, they can perform widefield and confocal imaging and analysis—all automated and assisted by AI.

The rapid rise of AI may spark concerns among some scientists that technology will take their place in the lab, but the truth is, it’s more likely to gift them more time to dive deeper into their research, make more discoveries, and develop better therapies. Scientists will be essential for training AI algorithms to make smart decisions in the cell culture process. Then they’ll step out of the process and let automation take over, eliminating the risk of mistakes or inconsistencies in hands-on processes that can compromise experiments. The result will be enhanced productivity, optimized efficiency, and a fast path to preclinical trials for promising therapeutics.