An End-to-End Sensor Signal Processing Framework for Automated Enumeration of Circulating Tumor Cells in Multichannel Immunofluorescence Microscopy

Manual enumeration of cells in immunofluorescence microscopy is time-consuming and subjective, hindering reproducibility in biomedical analysis. Circulating tumor cells (CTCs), rare yet vital indicators of cancer metastasis, are difficult to detect due to their low prevalence and imaging noise. We propose an end-to-end automated pipeline integrating Gaussian Mixture Model (GMM)-based noise reduction, gradient-based radial intensity enhancement, YOLOv7 detection, and multichannel marker aggregation. Applied to CTC detection, our system improves the F1 score from 61.6% to 89.8% and reduces analysis time from 4–8 hours to 1–2 hours. This modular framework offers an accurate, efficient, and reproducible solution for high-throughput cell analysis.