Image3C, a multimodal image-based and label-independent integrative method for single-cell analysis.

Area of Science:

• Cell Biology
• Bioinformatics
• Computational Biology

Background:

• Image-based cell classification is vital for identifying cellular phenotypic changes.
• Current methods rely on species-specific reagents (e.g., antibodies) and prior knowledge, limiting their application.
• Many research organisms lack these specific reagents, hindering image-based analysis.

Purpose of the Study:

• To develop a novel image-based cell classification methodology.
• To overcome limitations of species-specific reagents in cell analysis.
• To enable de novo cell composition analysis for understudied organisms.

Main Methods:

• Developed Image3C (Image-Cytometry Cell Classification) methodology.
• Combined image-based flow cytometry with unbiased, high-throughput cell clustering.
• Integrated convolutional neural networks (CNNs) for classification.
• Utilized intrinsic cellular features and non-species-specific dyes.

Main Results:

• Image3C successfully performs cell classification without species-specific reagents.
• The method enables de novo cell composition analysis.
• Detected changes between different experimental conditions using intrinsic cellular features.

Conclusions:

• Image3C expands the utility of image-based cell analysis to a broader range of research organisms.
• The methodology is applicable even when detailed cellular phenotypes are unknown or reagents are unavailable.
• Facilitates robust cell population analysis in diverse biological systems.