AI-Augmented Fluorescence Imaging of Non-Small Cell Lung Cancer Protease Activity via a Dual-Enzyme Nanoprobe

Protease activity mapping plays a vital role in understanding tumor microenvironments and developing diagnostic tools. However, simultaneous quantification of multiple protease activities in complex biological systems remains challenging due to probe instability, signal overlap, and limited analytical throughput. Here, we present a dual-enzyme-responsive Au–Se nanoprobe functionalized with two orthogonally labeled peptide substrates for kallikrein 6 (KLK6) and urokinase-type plasminogen activator (uPA), two proteases implicated in nonsmall cell lung cancer (NSCLC). The use of gold–selenium bonds enhances resistance to glutathione-induced degradation and maintains fluorescence quenching stability in physiological conditions. To augment signal analysis, we integrate deep learning-based segmentation using Cellpose and U-Net to extract quantitative data from single-cell and in vivo imaging, respectively. This AI-assisted workflow significantly improved measurement reproducibility and enabled high-content quantification of protease activity distributions and tumor-to-background ratios, establishing a robust platform for dual-protease sensing in complex biological systems.