Integrated microfluidic platform for the capture and identification of circulating tumor cells using dielectrophoresis and image recognition.
Abstract
Background
Circulating tumor cells (CTCs), which detach from primary tumors and enter the peripheral blood, serve as biomarkers for monitoring a patient's status following cancer therapy.
Methods
In this study, we developed a microfluidic platform integrating dielectrophoresis and image recognition for the capture and detection of CTCs from blood sample. Experimental parameters included applied electric field strengths ranging from 12.5 to 32.5 × 104 V/m (at 1 MHz) and sample flow rates of 6.25-31.25 mm/min. Fluorescent antibodies were employed to label captured CTCs for real-time detection. Furthermore, a detection software utilizing OpenCV was developed to label CTCs through image processing. The identification performance of the self-developed software was evaluated and compared with machine learning algorithms.
Results
AND
Conclusions
Results demonstrated that the optimal CTC separation efficiency of 78 ± 2% was achieved at an electric field of 25 × 104 V/m and a flow rate of 18.75 mm/min. While the self-developed software achieved a lower recall than the YOLOv8 model (85 ± 4% vs. 92 ± 2%), its precision matched that of YOLOv8 (90 ± 3% vs. 89 ± 4%).