Integrated microfluidics-fluorescence machine-vision sensing system for rapid, high-accuracy somatic cell counting in raw cow milk: Addressing fouling, bubble interference, and cell clustering

Compact MFMV-SCC integrates staining, antifouling channels, and vision counting. • G6.5-COOH channels reduce fouling and visible bubbles during raw-milk flow. • MV-FCC-PCM mitigates high-SCC undercounting from connected cell contours. • SCC is quantified in 2 min over 39–1106 × 10 3 cells/mL, with an LOD of 51 × 10 3 cells/mL. • Validation supports portable, low-cost, field-oriented SCC screening. Somatic cell count (SCC) is a key indicator of milk quality and udder health, with elevated SCC levels linked to mastitis and economic losses in the dairy industry. However, rapid SCC analysis in raw cow milk remains challenging because staining efficiency, microchannel fouling, bubble-induced imaging interference, and undercounting of clustered cells can affect analytical performance. Here, we report a miniaturized microfluidics-fluorescence machine-vision platform for rapid, high-accuracy SCC analysis in raw cow milk. The platform integrates an optimized SYBR GOLD-based rapid staining protocol, G6.5-COOH-modified microchannels that reduced protein adsorption, somatic-cell adhesion, and visible bubble interference, and a rule-based cluster-aware image-analysis algorithm, termed the Machine Vision-based Fluorescent Cell-Cluster Precise Counting Method (MV-FCC-PCM), for quantifying connected fluorescent cell contours. Under the optimized image-processing conditions, MV-FCC-PCM achieved a counting accuracy of 98.11% ± 0.80% against manual counting for high-density fluorescent cell images. The system enables frame-based fluorescence detection and automated SCC quantification within 2 min per sample. It provides a quantification range from 39 × 10 3 to 1106 × 10 3 cells/mL and achieves a limit of detection (LOD) of 51 × 10 3 cells/mL. Validation using certified reference materials and 58 field-collected individual Holstein cow milk samples showed close agreement with reference methods, good reproducibility, and robust performance against selected milk-matrix interferents. Overall, this integrated platform provides a compact workflow for rapid raw cow milk SCC screening and supports further validation for on-site milk-quality monitoring.