Smart-phone phase contrast microscope with a singlet lens and deep learning

Abstract

Amounts of cost-effective biological observing and detecting instruments are needed in biology and medical applications. In these demands, a low-cost, portable microscope with light-weight and tele-communication, is very attractive in the resource-limit area. In this manuscript, a portable singlet phase contrast microscope based on smart-phones are proposed. In the optics hardware, an external singlet lens is designed and closely attached before the rear cameras of a smart-phone, which constructs the imaging part of a simplified microscope. And a circular oblique illumination is adopted. In the computational imaging part, the in-focus images recorded by the smart-phone are virtually enhanced and style-transferred to phase contrast images by deep learning methods. Through pairs of experimental image data, we train the deep learning network and demonstrate the effectiveness of this virtual phase contrast imaging approach using multiple samples, e.g., unstained/transparent tumor tissue pathological slide, and H&E (Hematoxylin and Eosin) stained tumor tissue pathological slide. Our results provide a powerful example of low-cost portable phase contrast microscope with high-speed tele-communication abilities in the 5G + mobile internet era.