Metasurface with deep learning for biomedical imaging applications (Invited Paper)

This talk explores the transformative power of deep learning (DL) in metasurface-based imaging systems. We will showcase how DL algorithms are pushing the boundaries of performance in different applications, including endo-microscopy, meta-miniscopy, and confocal microscopy. In endo-microscopy, we achieve high-resolution brain images with detailed vasculature within 0.1 seconds (50x faster) by combining DL networks with existing endoscopes. This significantly reduces image acquisition time and system complexity. For DL assisted meta-miniscope, the integration of DL models with a meta-miniscope facilitates the transformation of conventional bright-field images into edge-enhanced images with high contrast accuracy. This novel approach not only enhances image quality but also minimizes system complexity and processing times, thus opening new avenues for improved microscopy techniques. This talk will be of interest to researchers and developers in optics, bioimaging, and deep learning, to highlight significant advancements through metasurfaces with deep learning.