Multiplex Fourier ptychographic reconstruction with model-based neural network for Internet of Things

Fourier ptychographic microscopy (FPM) is a newly developed technique to capture wide field-of-view (FOV) and high-resolution images that meet the demands of Internet of Things (IoT). FPM enlarges the equivalent numerical aperture of the system and achieves phase imaging by simply employing an angle-varied illumination module. Recently, researches propose to perform the FPM reconstruction with deep learning servers which is costly and requires large datasets. In this paper, we present a new FPM image reconstruction framework termed multi-NNP for Internet of Medical Things (IoMT). Multi-NNP performs multiplex ptychographic reconstruction with a model-based neural network locally rather than on deep learning servers. Our framework simplifies the process and improves the reconstruction performance which promotes the application of wide-field, high-resolution microscopic images in IoMT. Experimental results demonstrate the performance and effectiveness of the proposed framework.