Real-time characterisation of the harvesting process for adherent mesenchymal stem cell cultures based on on-line imaging and model-based monitoring

• We model the morphological cell response to harvesting solutions.
• We assess in real-time the harvesting efficiency.
• We predict the optimal time-point to inhibit the enzymatic reaction.
• The method proposed is generic and scalable.

Determining the optimal time to inhibit the enzymatic reaction during cell harvesting is crucial to successfully detaching adherent cells from the substrate before irreversible cell damage is caused by prolonged cell exposure to the enzymes. This study aimed at developing a non-invasive methodology to determine objectively and automatically the optimal time-point at which to inhibit the enzymatic reaction during cell harvesting. The harvesting process was monitored under a microscope and image analysis was used to measure the cells' morphology (circularity) from the images. System identification techniques were used to model the cells' circularity in response to the harvesting solution and to quantify the time when the cells become circular and reach a plateau phase (inhibition time). ARX models were used to accurately model the process both in culture flasks and in a bioreactor (R2 ≥ 0.98 and 0.95 respectively) and to quantitatively determine the inhibition time. First-order Recursive-ARX models were applied to predict the inhibition time in real-time. When the model parameters converged the median error was 21 s (Min 0 s, Max 75 s). The error decreased monotonically as new data were collected. The developed approach was generic and could be applied both in flasks and in a clinical-scale bioreactor. By automating decision-making, it will be possible to create standards and reproducibility that are necessary for large scale, robust and cost-effective cell cultures with reduced cell culture variability and consistent cell batches.