A Classification of Flowing White Blood Cells Based on Simplified Optical Flow System and Machine Learning Algorithms

초록

In this study, we present a customized optical flow system that integrates acoustophoresis technology and image capturing optical module for the analysis of flowing white blood cells (WBCs). The system was composed of multiple laser modules, portable function generator with desirable voltage, basic optical components to generate multiple fluorescence signals and images. For the classification of WBCs, the system features a microfluidic channel combined with transducer module to generate focal plane for aligning flowing cells. The system is able to obtain three different fluorescent signals as well as one bright field image. Using fluorescent signals and images, we can classify WBCs with machine learning algorithms. Calibration experiments demonstrated reliable classification of microspheres with varying fluorescence intensities and sizes, and the system accurately distinguished WBC subtypes based on their fluorescence signals and morphological characteristics. Clinical sample testing further validated the system's capabilities, with results comparable to those obtained from commercial optical microscopes. This integrated optical-acoustic cytometry platform offers high-speed, multi-parameter cell analysis with reduced complexity and cost. Its combination of acoustic focusing, robust imaging, and AI-driven analysis makes it a promising tool for both research and clinical applications, particularly in resource-constrained settings.