"In HCS, segmentation—the accurate identification of cell boundaries and organelles—requires high-quality imagery. Good optics and well-prepared samples are necessary to achieve high signal-to-noise ratios, which enable reliable AI-driven analysis."
In an interview, Boyd Butler, a microscopy and high-content screening expert at Molecular Devices, discussed the role of artificial intelligence (AI) in high-content screening (HCS). Butler emphasized the importance of AI-ready image data, challenges in imaging 3D models, and how advances in optics, automation, and machine learning are influencing drug discovery.
AI-Ready Image Data
Butler noted that AI systems depend on high-quality input data. In HCS, segmentation—the accurate identification of cell boundaries and organelles—requires high-quality imagery. Good optics and well-prepared samples are necessary to achieve high signal-to-noise ratios, which enable reliable AI-driven analysis. This is especially important in toxicity testing, where subtle phenotypic changes must be detected.
Imaging 3D Models
Butler stated that 3D models such as organoids are more physiologically relevant than 2D cultures but present imaging challenges. These include light scattering, refractive index mismatches, and large data volumes from multiple optical sections. Segmenting structures within dense 3D tissue also requires high image quality and computational power.
Role of Optics
Better optics improve resolution and image clarity. High numerical aperture objectives and water immersion objectives reduce refractive index mismatches. Confocal imaging eliminates out-of-focus light in thick samples. Poor resolution can lead to incorrect segmentation by AI models.
Automation and Reproducibility
Automation increases throughput and consistency by handling repetitive tasks and standardizing protocols. It reduces human error and variability, which is critical for reproducibility in pharmaceutical development.
AI in Image Analysis
Butler identified segmentation as the area where AI has the greatest impact. Accurate segmentation is foundational for classification and quality control. Deep learning performs well in complex biological samples where traditional methods struggle.
Guided Workflows
Guided workflows simplify instrument configuration and protocol development, reducing setup time and errors. They improve reproducibility across users and are valuable in core facilities with diverse user skill levels.
Machine Learning for Phenotypic Changes
Machine learning can analyze hundreds of morphological features simultaneously, detecting subtle changes that humans might miss. Examples include changes in endoplasmic reticulum morphology or cytoskeletal organization. This capability is useful in toxicology and cell painting assays.
Scaling Core Labs
AI-enabled platforms reduce the learning curve for new users, allowing them to operate independently. This frees core facility staff to focus on experimental design and data interpretation, increasing imaging capacity without adding staff.
About Boyd Butler
Boyd Butler has a background in biophysics and optical physics and has served as a faculty member, principal investigator, and core facility director. At Molecular Devices, he helps scientists adopt advanced imaging solutions including HCS systems, automated workflows, and AI-powered analytical tools.
About Molecular Devices
Molecular Devices provides bioanalytical measurement systems, software, and consumables for life science research and pharmaceutical development. Their product portfolio includes platforms for high-throughput screening, genomic and cellular analysis, colony selection, and microplate detection.