Two-step automated method for robust Fourier ptychographic microscopy LED misalignment calibration

Abstract

This article introduces a fully automated and robust hardware calibration method for aligning the LED matrix in Fourier Ptychographic Microscopy (FPM). The proposed method is divided into two independent procedures targeting translational and rotational misalignments. Translational errors are corrected using the Automatic LED Calibration (ALC) metric, which evaluates the symmetry of Fourier spectra obtained from two images illuminated from opposite angles. Rotational misalignments are handled through a subpixel image registration algorithm that estimates direction and magnitude of the angular deviation. Neither step requires specialized calibration targets or complex modeling. Experimental results confirm that method reliably compensates for a wide range of misalignments, delivering stable and reproducible outcomes across diverse conditions and calibration objects. This dual-step approach significantly reduces alignment effort and improves the quality of image reconstruction. Once calibrated, the system enables high-fidelity reconstructions using both Quasi-Newton and Gerchberg-Saxton algorithms. The method is compatible with open-source platforms like FPMapp, streamlining its integration into existing workflows. By simplifying system alignment while maintaining precision, this calibration strategy is poised to accelerate the broader adoption of FPM in digital pathology and biomedical imaging. The proposed approach is presented as a step-by-step procedure suitable even for non-expert users, facilitating practical implementation across various applications.