Calibration-free quantitative phase imaging in multi-core fiber endoscopes using end-to-end deep learning

Fiber endoscopes have emerged as a vital tool for highresolution Recent advancements have adopted deep learning techniques microscopic imaging in hard-to-reach areas. In contrast to expedite the QPI image reconstruction process [12, 13]. Moreover, to conventional endoscopes with a typical diameter of extant literature indicates the potential of decrypting an several millimeters, fiber endoscopes, which could be submillimeter encoded phase directly from speckle images utilizing deep learning, thin and flexible [1-5], can pass through the organ's although only in simulated environments [14]. This demonstrates intricate pathways without causing harm inside the body [6], the theoretical possibility of reconstructing the original making them particularly suitable for procedures requiring utmost phase directly from speckle images using deep learning for MCF precision and minimal invasiveness. The reduced size and phase imaging, however, networks trained on simulated data adaptability of fiber endoscopes ensure less discomfort for the can hardly achieve accurate phase reconstructions in real-world patient, leading to quicker recovery times and a lower risk of optical systems.