Deep‐Tissue, Large‐FOV 3D NIR‐II Fluorescence Confocal Microscopy With Hundred‐Nanosecond Equivalent Pixel Dwell Time

Abstract Near‐infrared II (NIR‐II, 900–1880 nm) fluorescence confocal microscopy enables in vivo imaging with high spatial resolution at large depth. Nonetheless, three dimensional (3D) imaging requires capturing substantial pixels and prolonged laser scanning, leading to phototoxicity, exogenous probe metabolic decay, and loss of information on dynamic anatomical structures. Strategies to diminish imaging duration can be considered by decreasing the actual pixel dwell time without deterioration of imaging quality. In this study, a novel approach combining NIR‐II fluorescence confocal microscopy is introduced with deep learning interpolation network, which substantially decreases axial sampling frequency requirements, achieving equivalent hundred‐nanosecond pixel dwell time in 3D visualization in vivo. By applying the cerebral vessel interpolation (CVI) network to large field‐of‐view (FOV) 3D NIR‐II fluorescence microscopic imaging, up to a 16‐fold increase has been achieved in laser scanning speed, reducing pixel dwell time from 8 µs to 500 ns. This significantly reduces laser‐induced damage to biological samples, lessens the need for extending the metabolism time of exogenous probes, and facilitates potential rapid biomedical imaging applications. Benchmarking tests show CVI network achieves the best performance compared to conventional interpolation methods on both lateral and axial cross‐sectional images.