Boundary-artifact-free quantitative phase imaging with Fourier ptychographic microscopy

Fourier ptychographic microscopy (FPM) enables high-resolution, wide-field imaging of both amplitude and phase, presenting significant potential for applications in digital pathology and cell biology. However, artifacts commonly observed at the boundaries of reconstructed images can significantly degrade imaging quality and phase retrieval accuracy. These boundary artifacts are typically attributed to the use of the fast Fourier transform (FFT) on non-periodic images. Another significant physical factor that should not be overlooked is the transverse diffraction of light across boundaries. Here, we introduce a boundary extension reconstruction framework for FPM, termed BE-FPM, which provides boundary-artifact-free quantitative phase imaging (QPI) with minimal computational overhead. In this method, the reconstructed image is initialized with zero-padding and then self-extrapolated during the subsequent iterative reconstruction process. This approach allows for partial restoration of the sample beyond the boundary, ensuring sample consistency around the boundary and addressing the boundary artifact problem fundamentally. We demonstrate the effectiveness of the proposed BE-FPM on both microlens array and live cells, establishing it as an effective FPM solver for boundary-artifact-free QPI and accurate phase characterization for various types of samples.