Binocular vision fusion enhanced 3D NIR-II in vivo imaging of bone and vessel networks

Taking advantages of weak light scattering and minimal amount of auto-fluorescence background, optical imaging through the second near-infrared window (NIR-II, 1000-1700 nm) allows resolving the microscopic structures in deep tissues. However, current 2D wide-field imaging systems cannot provide the axial resolution to reveal depth information and the lateral resolution thereof has been limited by the pixel numbers of NIR-II cameras. We aim to improve 3D in vivo imaging at a large field of view (FOV) to achieve high-resolution 3D imaging of bone and vessel simultaneously. We developed a 3D NIR-II imaging technique utilizing iterative convolutional sub-pixel image reconstruction (ICSP-IR) and convolutional neural networks (CNN) in a binocular system. The instrumentation and CNN powered data analytics enable 3D in vivo volumetric imaging with a lateral resolution of 153 µm and an axial resolution of 480 µm at a FOV of 42.5 mm × 53 mm × 14 mm. High-resolution 3D imaging of both tibia structures and vessel networks can be simultaneously achieved by the 808-nm and 980-nm dual-beam excitations of NIR-Ⅱa type nanoparticles that emit at 1060 nm and NIR-Ⅱb type nanoparticles that emit at 1530 nm, respectively.