A single-chip, dual-gain color–NIR camera for fluorescence-guided lung and breast cancer surgery

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Abstract Near infrared (NIR) fluorescence guided surgery can highlight tumors and sentinel lymph nodes, but most clinical fluorescence imaging systems use multi-camera architectures in which residual color–NIR misregistration can mislocalize fluorescence relative to anatomy, and weak signals often require dimming operating room lights. We developed a single-chip dual-gain scientific CMOS (sCMOS) sensor with red, green, blue (RGB), and NIR channels that provides intrinsically coregistered color reflectance and NIR fluorescence under typical operating room illumination (50–80 kLux). A backside illuminated array with division of focal plane (DoFP) pixelated interference filters achieves an input-referred read noise of 2.25 electrons and a fused dynamic range >90 dB, enabling indocyanine green (ICG) detection at ~100 pM under dark-background conditions and ~500 pM under standard 50–80 kLux operating-room illumination, with quantification over four orders of magnitude. We integrated the sensor into a surgical camera and evaluated it in 20 patients with lung cancer receiving the cathepsin-activated ICG probe VGT-309 and 35 patients with breast cancer undergoing ICG-guided sentinel lymph node biopsy. In lung cancer, lesion level sensitivity was 0.90 (18/20) and positive predictive value was 0.78 (18/23), with at least one clinically significant event in 45% (9/20) of patients. In breast cancer (90 specimens), high-gain NIR imaging improved classification of gamma probe-designated sentinel lymph nodes by node-to-background ratio (ROC AUC 0.94 vs 0.60) while preserving a white light view. These results establish a single-chip route to lights-on color and NIR guidance in the operating room.