Lens-Free Holographic Imaging-Based Immunosensor Using Unpaired Data Set Signal-to-Noise Ratio-Enhanced Modal Transformation for Biosensing

Conventional microscopes have limited capacities to reconcile the trade-off between the lens and field of view (FOV). Thus, the imaging field and accuracy of immunosensors remain restricted. In this study, a holographic deep learning unpaired modal transformation-assisted immunosensor is presented, combining a portable lens-free holographic imaging device with a CuO₂@SiO₂ nanoparticle-based click reaction signal amplification strategy for accurate antibiotic detection. The immunosensor achieves both large FOV imaging (10.3-fold improvement over the microscope) and signal-to-noise ratio-enhanced holographic reconstruction (signal-to-noise ratio of 32.65 dB, structural similarity index of 0.83) by constructing a modal transformation model with unpaired data sets, thus resolving the complexity of one-to-one matching of data sets required by conventional methods. The immunosensor detects chloramphenicol with high sensitivity and a wide linear range (limit of detection = 3.54 pg/mL, dynamic range of 10 pg/mL to 50 ng/mL) within 40 min. As a portable detection device, it demonstrates potential as a sensitive and on-site detection platform for food safety inspection and clinical diagnosis.