Real-time visualization of dextran extravasation in intermittent hypoxia mice using noninvasive SWIR imaging.

ABSTRACT

Imaging tools are crucial for studying the vascular network and its barrier function in various physiopathological conditions. Shortwave infrared (SWIR) window optical imaging allows noninvasive, in-depth exploration. We applied SWIR imaging, combined with vessel segmentation and deep learning analyses, to study real-time dextran probe extravasation in mice experiencing intermittent hypoxia (IH)-a characteristic of obstructive sleep apnea associated with potential cardiovascular alterations due to early vascular permeability. Evidence for permeability in this context is limited, making our investigation significant. C57Bl/6 mice were exposed to normoxia or intermittent hypoxia for 14 days. Then SWIR imaging between 1,250 and 1,700 nm was performed on the saphenous artery and vein and on the surrounding tissue after intravenous injection of labeled dextrans of two different sizes (10 or 70 kDa). Postprocessing and segmentation of the SWIR images were conducted using deep learning treatment. We monitored high-resolution signals, distinguishing arteries, veins, and surrounding tissues. In the saphenous artery and vein, after 70-kD dextran injection, tissue/vessel ratio was higher after intermittent hypoxia (IH) than normoxia (N) over 500 seconds (P < 0.05). However, the ratio was similar in N and IH after 10-kD dextran injection. The SWIR imaging technique allows noninvasive, real-time monitoring of dextran extravasation in vivo. Dextran 70 extravasation is increased after exposure to IH, suggesting an increased vessel permeability in this mice model of obstructive sleep apnea.NEW & NOTEWORTHY We demonstrate that SWIR imaging technique is a useful tool to monitor real-time dextran extravasation from vessels in vivo, with a high resolution. We report for the first time an increased real-time dextran (70 kD) extravasation in mice exposed to intermittent hypoxia for 14 days compared with normoxic controls.