Characterizing Cell Health Using Machine Learning
A machine learning-based approach using microscopic images and pixel-level classification enhances cell health characterization in living cells, addressing the limitations of existing methods by providing accurate scoring and high-throughput analysis.
Patent Information
- Application Number
- JP2025545175
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-02-09
- Filing Date
- 2024-02-08
- Publication Date
- 2026-02-27
AI Technical Summary
Existing methods for characterizing protein movement in living cells are limited by interactions with their dense surroundings, making it difficult to analyze cellular phenomena under physiologically relevant conditions with minimal confounding factors.
A machine learning-based approach using microscopic images of fluorescently labeled cells, trained with varying compound concentrations and environmental conditions, to assign cell health scores through pixel-level classification and ensemble models, enabling high-throughput single-molecule tracking (SMT) for enhanced characterization.
The method provides accurate cell health scoring and characterization under physiologically relevant conditions, minimizing confounding factors and enabling high-throughput analysis of cellular dynamics.