Automatic Classification of Circulating Blood Cell Clusters based on Multi-channel Flow Cytometry Imaging.

Area of Science:

• Biomedical Engineering
• Computational Biology
• Hematology

Background:

• Circulating blood cell clusters (CCCs) are key biomarkers for thrombosis, infection, and inflammation.
• Flow cytometry with fluorescence staining is standard for analyzing CCC morphology and protein profiles.
• Existing machine learning tools primarily focus on single cells, not complex CCCs.

Purpose of the Study:

• To develop a computational framework for automated analysis of circulating blood cell cluster images.
• To accurately identify cell types within heterogeneous CCCs.
• To improve the analysis of flow cytometry data for disease biomarker discovery.

Main Methods:

• A two-step computational framework was developed for CCC image analysis.
• The YOLOv11 model was fine-tuned for classifying images into cell cluster and non-cluster groups.
• Cluster contours were overlaid with multi-channel fluorescence stains to identify cell phenotypes within clusters.

Main Results:

• The framework achieved over 95% accuracy in both cluster classification and cell phenotype identification.
• The YOLOv11 model outperformed traditional CNNs and ViT in image categorization.
• The method demonstrated robustness against cell debris and staining artifacts.

Conclusions:

• An automated computational framework effectively analyzes circulating blood cell cluster images from flow cytometry.
• The framework leverages both bright-field and fluorescence data for enhanced accuracy.
• This approach has potential applications beyond blood cells, including immune and tumor cell cluster analysis.