During cancer surgery, surgeons must decide which lymph nodes to biopsy or preserve. Current intraoperative tools localize draining nodes but lack a real-time readout of tumor involvement, leading to overtreatment, undertreatment, and lymphedema. We developed a bioinspired single-chip imager modeled on mantis shrimp vision that coregisters color, near-infrared (NIR) fluorescence from indocyanine green (ICG), and deep-ultraviolet (UV) autofluorescence on a single sensor within the same field of view. In the ex vivo workflow, NIR fluorescence first localizes draining nodes. After the lymph node is exposed, a deep-UV autofluorescence acquisition provides a label-free readout of nodal involvement. Using stacked photodiodes and pixel-level multiband filters, the device enables coregistered multispectral imaging from 300 to 1000 nm. In immediate ex vivo testing on breast cancer specimens (33 patients, 94 nodes), the UV channel distinguished metastatic from benign nodes (area under the receiver operating characteristic curve (AUC) 0.89; sensitivity 97%; specificity 89%) while preserving standard sentinel node localization with ICG. The results demonstrate a compact route to point-of-care nodal assessment.
Jin et al. (Mon,) studied this question.