organoid

organoid performs automated pixel-level segmentation, labeling, temporal tracking, and morphological quantification of single organoids in brightfield and phase-contrast microscopy images to measure count, size, and morphology for analysis of drug responses.

Key Features:

• Automated pixel-by-pixel recognition and tracking: Performs pixel-level segmentation, labeling, and temporal tracking of single organoids in brightfield and phase-contrast microscopy images.
• Quantitative accuracy: Reports 95% agreement for organoid count and 97% agreement for organoid size compared to manual measurements across pancreatic, lung, colon, and adenoid cystic carcinoma organoids.
• Temporal tracking stability: Maintains single-organoid tracking accuracy above 89% over extended time-lapse microscopy studies up to four days.
• Morphological feature extraction: Computes metrics including circularity, solidity, and eccentricity to quantify organoid morphology and assess treatments such as chemotherapy.
• High-throughput data handling: Supports processing of large, data-intensive microscopy datasets for high-throughput experiments.

Scientific Applications:

• Drug Discovery: Enables measurement of organoid count, size, and morphology over time to assess compound effects and drug responses.
• Personalized Medicine: Supports analysis of patient-derived organoids for personalized drug screening and response profiling.
• High-Throughput Screening: Facilitates large-scale organoid screens through automated segmentation, tracking, and quantitative output generation.

Methodology:

Uses deep learning algorithms trained on a diverse set of brightfield and phase-contrast microscopy images with pixel-by-pixel segmentation; training included pancreatic cancer organoids and validation was performed across pancreatic, lung, colon, and adenoid cystic carcinoma organoids, including time-lapse studies up to four days.