Rapid Single-Cell Phenotypic Antifungal Susceptibility Testing on a SlipChip Enabled by Deep Learning.

Fungal infections cause over 1.5 million deaths annually, yet current antifungal susceptibility testing (AFST) methods require 24-48 h, delaying critical treatment decisions. Here, we present a rapid, label-free, single-cell AFST (sc-AFST) platform that integrates a SlipChip microfluidic device with deep learning-based image analysis to determine minimal inhibitory concentrations (MICs) within 4 h. The SlipChip partitions fungal cells into picoliter droplets for exposure to antifungal gradients, while a ResNet-34 classifier and U-Net segmentation model automatically identify and quantify cellular growth directly from bright-field images. We validated the platform using Candida albicans and clinical isolates, achieving >98.8% sensitivity in cell detection and >93% segmentation accuracy. The sc-AFST results showed 100% categorical agreement with gold standard broth microdilution test across all 10 clinical isolates, despite one minor MIC discrepancy. This approach enables high-resolution single-cell growth analysis and population-level drug response profiling, offering a scalable, rapid, and clinically actionable tool for AFST.