Through standardized measurements for clinics: a novel autofocusing framework for digital holographic microscopy

Digital holographic microscopy (DHM) is a phase contrast imaging technology attractive for high-throughput, label-free, and quantitative cell analysis for clinical applications. Accurate and repeatable measurements require a standardized framework with minimal human interference. Current high-speed imaging setups use autofocus algorithms that often require costly calibration beads spiked into samples. Our study introduces a novel, end-to-end autofocusing framework compatible with clinical workflows. It automates evaluation of focus metrics directly on blood cells of processed sample, eliminating the need for calibration materials. This framework employs a compact MobileNet-based vision model trained through knowledge distillation, offering an accurate and repeatable solution for clinical applications.