Ultra-low photodamage three-photon microscopy assisted by MSAD-net: enabling high-fidelity in vivo monitoring of muscle regeneration via deep learning
Peer-Reviewed Publication
Updates every hour. Last Updated: 25-Dec-2025 02:11 ET (25-Dec-2025 07:11 GMT/UTC)
Skeletal muscle regeneration research is hindered by the "photodamage-imaging quality" trade-off in three-photon microscopy (3PM). A team from Zhejiang University developed the Multi-Scale Attention Denoising Network (MSAD-Net) to address this: combining MSAD-Net with 3PM reduces excitation power to 1.0–1.5 mW (1/4–1/2 of conventional levels) and scanning time to 2–3 μs/pixel (1/6–1/4 of standard), while maintaining 0.9932 SSIM and real-time denoising (80ms/frame). The system enables five-channel deep in vivo imaging of mouse muscle, uncovering key roles of macrophages and blood vessels in muscle stem cell-mediated repair.
De-Wei Gao's research group at ShanghaiTech University has developed a new method for efficient and highly selective boron-heteroatom functional group exchange reactions. Their method overcomes the inherent difficulty of primary radical instability in traditional free radical chemistry and can achieve highly selective conversion of primary carbon-boron bonds to a variety of heteroatom functional groups. This strategy has been successfully applied to a sugar-derived 1,n-diboron compound system, achieving modular modification and efficient synthesis of sugar molecules and has shown to have potential application in the rapid construction of bioactive molecules. These results were published as an open access article in CCS Chemistry, the flagship journal of the Chinese Chemical Society.