This paper presents a miniaturized integrated microscope achieving superior optical performance compared to commercial microscopes, while being significantly smaller and lighter. A progressive optimization pipeline combines aspherical lenses, diffractive optical elements, and deep learning for spatially varying deconvolution, resulting in a tenfold improvement in depth-of-field. The microscope's compact design allows for integration into a cell phone, demonstrating its potential for portable diagnostics.
