The fastest, clinically validated ultraviolet photoacoustic microscopy for label-free intraoperative tissue assessment enhanced by unsupervised deep learning

We present a label-free, the fastest platform for intraoperative tissue assessment using ultraviolet photoacoustic computed microscopy enhanced by unsupervised deep learning. An orthogonal excitation-detection design with a 21 × 21 UV array and 256-channel transducer enables 1.4 µm resolution at 8.82 million pixels per second. The system captures thick, fixed or fresh tissues—including lung, brain, and breast—with fine microanatomical detail. An unsupervised CycleGAN-based pipeline provides H&E-equivalent images with minimal training data. This real-time, slide-free method addresses critical challenges in intraoperative histopathology.