Title: Leveraging Dual Datasets for Enhanced 3-D Cryo-EM Reconstruction via the Double Moments Method

Abstract: Cryo-electron microscopy (cryo-EM) serves as a potent imaging modality for deriving three-dimensional molecular architectures from noisy tomographic projections of randomly oriented particles. This study presents a novel data fusion approach, known as the Method of Double Moments (MoDM), designed to reconstruct molecular structures using two sets of second-order moments derived from projection images subjected to different orientation distributions: one uniform and the other non-uniform with an unknown distribution. We establish that these moments typically yield a unique determination of the underlying structure, modulo a global rotation and reflection. Furthermore, we have devised an algorithm based on convex relaxation that ensures precise recovery utilizing solely second-order statistics. Our findings highlight the benefits of acquiring and modeling multiple datasets under varied experimental conditions, demonstrating that exploiting dataset diversity can significantly improve reconstruction fidelity in computational imaging applications.

Source: arXiv Generated at: 2026-06-02 00:00:00 UTC