Compressed sensing for STEM tomography

Laurène Donati, Masih Nilchian, Sylvain Trépout, Cédric Messaoudi, Sergio Marco, Michael Unser
2017 Ultramicroscopy  
A central challenge in scanning transmission electron microscopy (STEM) is to reduce the electron radiation dosage required for accurate imaging of 3D biological nano-structures. Methods that permit tomographic reconstruction from a reduced number of STEM acquisitions without introducing significant degradation in the final volume are thus of particular importance. In random-beam STEM (RB-STEM), the projection measurements are acquired by randomly scanning a subset of pixels at every tilt view.
more » ... In this work, we present a tailored RB-STEM acquisition-reconstruction framework that fully exploits the compressed sensing principles. We first demonstrate that RB-STEM acquisition fulfills the "incoherence" condition when the image is expressed in terms of wavelets. We then propose a regularized tomographic reconstruction framework to recover volumes from RB-STEM measurements. We demonstrate through simulations on synthetic and real projection measurements that the proposed framework reconstructs high-quality volumes from strongly downsampled RB-STEM data and outperforms existing techniques at doing so. This application of compressed sensing principles to STEM paves the way for a practical implementation of RB-STEM and opens new perspectives for high-quality reconstructions in STEM tomography. (L. Donati). 1 Both authors contributed equally to this work. rified biological specimens (cryo-STEM), yielding improved resolving power and broadening the scope of acceptable biological specimens [8] . Yet, despite its promise, cryo-STEM is subject to the same experimental limitation as other EM techniques -high-resolution imaging requires dense sampling with large electron radiation dosage, yet biological samples are extremely sensitive to electroninduced irradiation damages. This dosage constraint is even more critical in electron tomography (ET), which requires a series of projection images to be taken covering a large range of tilt angles [9] . Moreover, the geometry of conventional tomographic STEM imaging systems constrains the imaging of samples to a limited angular range. As a result, artifacts consequent to a missing wedge of information in the Fourier space may appear on the reconstructed image if the angular coverage is insufficient [10] . A trade-off between the reconstruction quality ( i.e. , wide and numerous high-SNR acquisitions) and the sample integrity ( i.e. , low electron dosage) must thus be considered when optimizing 3D STEM imaging. Several researches have therefore focused on reconstruction methods that address the limited-angle problem and permit lowerhttp://dx.
doi:10.1016/j.ultramic.2017.04.003 pmid:28411510 fatcat:3fsyvuecnnh27h32vlfjs3q63e