Fluoro-forest: A random forest workflow for cell type annotation in high-dimensional immunofluorescence imaging.

Area of Science:

• Biomedical Imaging
• Computational Biology
• Immunology

Background:

• Cyclic immunofluorescence (IF) enables high-resolution cell phenotyping and tissue organization analysis.
• Current workflows often use unsupervised clustering and cluster-level annotation, lacking statistical rigor and risking misclassification.
• Marker expression averages in existing methods do not provide robust cell type assignment.

Purpose of the Study:

• To develop an end-to-end pipeline for accurate cell type annotation in cyclic IF data.
• To improve upon existing clustering-based annotation methods by incorporating statistical evaluation.
• To enhance the interpretability and accuracy of cell type assignment in complex biological samples.

Main Methods:

• A semi-supervised, random forests approach for predicting cell type annotations.
• Cluster-based sampling for efficient training data generation.
• Downstream visualization techniques for interpretability of cell annotations.

Main Results:

• The proposed workflow significantly improves cell type annotation accuracy.
• Accurate annotation was achieved with a training set comprising less than 5% of the total cells.
• The pipeline provides cell type annotation probabilities and model performance metrics.

Conclusions:

• The developed pipeline offers a statistically robust and accurate method for cell type annotation in cyclic IF data.
• This approach enhances the reliability of cell phenotyping and tissue organization quantification.
• The tool can augment existing clustering-based workflows for complex IF data analysis.