Cheetah

Cheetah implements a U-Net convolutional neural network for image segmentation and time-resolved cell analysis to enable real-time cybergenetic control by integrating microscopy imaging with microfluidics and optogenetics.

Key Features:

• U-Net segmentation: Uses a U-Net convolutional neural network architecture for high-fidelity image segmentation of microscopy images.
• Improved accuracy: Delivers segmentation accuracy that surpasses traditional thresholding-based methods for precise cell delineation.
• Cell counting: Provides automated counting of cells from segmented images over time.
• Phenotypic characterization: Extracts cellular characteristics and phenotype-related measurements from image data.
• Cell tracking: Tracks cells over extended periods to enable analysis of long-term cellular dynamics.
• Real-time control signal generation: Converts image-derived measurements into control signals for real-time cybergenetic control and environmental regulation.
• Integration with experimental modalities: Integrates microscopy imaging workflows with microfluidics and optogenetics for experimentally coupled control.
• Dynamic protein expression control: Enables dynamic control of protein expression in mammalian cells via image-guided feedback.

Scientific Applications:

• Bacterial growth analysis: Quantifies and analyzes bacterial cell growth and dynamics from time-lapse microscopy.
• Mammalian cell growth analysis: Quantifies and monitors mammalian cell growth and morphology over time.
• Cybergenetic control platforms: Implements real-time cybergenetic control integrating microscopy-derived signals with environmental regulation.
• Optogenetic and microfluidic perturbations: Enables probing and manipulation of living cells using optogenetics and microfluidic environmental control.
• Long-term behavioral studies: Facilitates study of long-term cellular behaviors and dynamics through continuous segmentation, tracking, and quantification.

Methodology:

Computational steps explicitly include U-Net convolutional neural network–based image segmentation, comparison to thresholding-based methods for accuracy, automated cell counting, phenotypic characterization, temporal cell tracking, and conversion of image-derived measurements into real-time control signals.