 |
Dr. Andrés Ferrer |
 |
Dr. Natalia Jarzebska |
Cell counting and viability assessments are critical methods for labs working with cell cultures. A novel combination of digital holographic microscopy and machine learning is boosting the speed and reliability of these essential measurements.
Cell count and viability assessment are fundamental methods in biotechnology. They are used for many applications, including drug discovery, toxicity screening, and bioprocessing. However, scientists face certain challenges when measuring cell count and viability, including long operational time, sample contamination, and intra-operator variability, all of which affect the reliability of results. Moreover, biological samples need to be visualized with (fluorescent) staining that binds to specific structural components and can negatively affect the sample’s integrity. Additionally, staining adds operational time to an already laborious process. While automation can reduce manual errors, it is necessary to identify critical quality attributes (CQA) and critical process parameters (CPP) to compensate for sample complexity and overcome these challenges.
Digital holographic microscopy
Digital holographic microscopy is proving to be a powerful technique that enables fast, reproducible, and stain-free imaging of biological samples. It eliminates the need for image-forming lenses and consequently, the need to find the right focal plane, reducing error potential significantly. Thus, digital holographic microscopy can improve accuracy which in turn enhances analytical quality and enables better process understanding. Additionally, the speed and robustness of digital holographic microscopy make it an exceptional candidate for combination with machine learning (ML), which enables highly detailed image analysis without human intervention. The processing speed and reliability of this combination mean it is fast becoming the preferred approach for cell counting and viability assessment.
 |
Figure 1. Schematical representation of the setup of digital holographic microscopy. |
How digital holographic microscopy works
Digital holographic microscopy uses a coherent light source (laser) that illuminates the sample, cells, and media, both of which present a slight difference in refractive index. The light scattered from the cells interferes with the media it passes through, and the interference pattern created enables the reconstruction of a holographic image that is recorded by the instrument (Figure 1). Due to the low light intensity of the digital holographic microscopy laser and the absence of cytotoxic staining, the cellular analysis does not impact cell integrity, allowing researchers to gain a more complete picture of cell health in real-time. Whether the object being studied is an individual cell or a cell set, the image obtained with digital holographic microscopy can be further analyzed using numerical algorithms optimized by ML to extract even more complete information about morphology, cell structure, and viability.
In short, digital holographic microscopy provides significantly more information on cellular status than brightfield microscopy and eliminates the need for staining. This eliminates the possibility of cytotoxicity and intra-operator variability to provide significantly more reliable information about cell viability in less time.
A brief history: holography and microscopy
Holography, which uses both amplitude and phase to construct an image, was developed in the 1940s by Dennis Gabor CBE FRS. This advancement, which would not be fully realized until the development of the laser in the 1960s, led the Hungarian-British electrical engineer and physicist to be awarded the Nobel Prize in Physics in 1971 [1].
Holographic microscopy, in which Gabor [2] used visible light and photographic film to record microscopic images, was originally employed to improve the resolving power of electron microscopes.
Digital holographic microscopy, which digitally reconstructs sample images using interference patterns created by the interplay of an object beam and a reference beam (visible in Figure 1), has become a powerful tool when combined with machine learning to quickly assess large data sets such as cell populations and systems.
The frontrunner in cell-based studies
While there are other automated monitoring techniques, such as impedance-based methods and fluorescent staining, digital holographic microscopy coupled with ML is emerging as the frontrunner thanks to its ability to deliver fast and reliable results. This non-invasive and highly repeatable method provides a microscopic sample image and quantitative parameters with single-cell resolution.
This resolution enables the detailed study of cell morphology and cycle phase, including the potential for distinguishing between apoptosis and necrosis.
Better characterization and control
By training ML models on large digital holographic microscopy datasets, algorithms can classify cells based on their morphology and detect subtle changes in refractive index. This strong indicator of cell status enables analysis of large populations in real-time (Figure 2). This allows better characterization and control of cell cultures while adding a layer of consistency to the way data is analyzed. In conclusion, digital holographic microscopy’s ability to capture microscopic images of viable and dead cells in a stain-free and non-invasive manner makes it a powerful tool for studying complex cellular processes in R&D and manufacturing.
 |
Figure 2. A DHM picture obtained from a HeLa cell suspension measured with the new CytoDirect cell counter. Viable and dead cells are clearly differentiated and respectively labeled (red: dead, green: viable) by the machine learning algorithm. |
References
[1] Wiki: Dennis Gabor. https://en.wikipedia.org/wiki/Dennis_Gabor
[2] Harvard University: Holographic microscopy. https://manoharan.seas.harvard.edu/holographic-microscopy
