Efficient misalignment correction for annular LED arrays in intensity diffraction tomography.

Abstract

Intensity diffraction tomography (IDT) is a powerful label-free 3D microscopy technique capable of reconstructing the 3D refractive index (RI) of biological samples using angled programmable illumination and computational algorithms. Despite its potential for high spatiotemporal resolution imaging, its practical performance is highly sensitive to precise alignment between experimental setups and algorithmic models. In this Letter, we present an efficient misalignment correction method for annular LED arrays in IDT (mcIDT), incorporating an improved Fourier-Mellin transform (FMT) algorithm with enhanced noise resistance for wavelength and positional corrections. Furthermore, a global positional misalignment model is optimized using the least squares method, enabling robust correction even in the presence of significant misalignments. By integrating algorithmic "calibration" with physical position "correction," mcIDT significantly enhances both the resolution and accuracy of 3D RI reconstruction. Experimental results on resolution targets and biological samples, such as HepG2 and C166 cells, demonstrate its superior resolution and robustness in label-free, high spatiotemporal resolution 3D volumetric imaging. The proposed mcIDT offers a flexible and efficient solution for label-free computational microscopy across diverse biological and industrial applications.