Deep Learning-Assisted Lens-Free Holography Integrated Pd Nanozyme-Armed Phages for the Rapid and Extraction-Free Detection of Viable Bacteria

Bacterial infections represent a serious global threat to public health, often leading to severe infectious diseases and even fatalities. The most effective approach to prevent outbreaks is through the implementation of rapid, accurate, and simple point-of-care testing. Here, we developed a deep learning-assisted lens-free holography biosensing platform integrated with Pd nanozyme-armed phages for the rapid and extraction-free detection of viable bacteria, enabling on-site testing. Phages specifically recognize and capture target viable bacteria. Surface ligand-engineered Pd nanozyme-armed phages then trigger a tyramine signal amplification reaction, resulting in the formation of polystyrene microsphere-bacteria-magnetic nanoparticle complexes. Consequently, the number of unbound polystyrene microspheres decreases with bacterial capture. Additionally, the combination of Pd nanozyme and phage can eliminate bacteria <i>in situ</i> while preventing secondary cross-contamination. Portable lens-free holographic microscope offers several advantages, including an ultrawide field of view, low cost, and a lightweight design, enabling the accurate capture of microsphere holograms in the supernatant through a high-throughput manner. The transformer-based deep learning algorithm integrated with digital holographic reconstruction has been trained to precisely process the probe holograms at high speed. As proof of concept, our approach has successfully enabled the quantitative detection of viable <i>Salmonella typhimurium</i> with high sensitivity (∼20 CFU/mL) in 16 min, without additional nucleic acid extraction. It has been evaluated using various real samples, demonstrating great promise as an intelligent bioassay for next-generation biosensing.