Date Published: July 01, 2017
Publisher: International Union of Crystallography
Author(s): Martin Krenkel, Mareike Toepperwien, Frauke Alves, Tim Salditt.
Phase-contrast X-ray imaging of biological cells in two and three dimensions can be carried out with a low dose, based on free propagation and a setting of optimized wavefronts in cone-beam geometry. In order to reach the required contrast level, images have to be recorded in the holographic regime. The main result of this work is holographic recordings of a quality that is fully amenable to quantitative phase retrieval, beyond previous approximations. Different approaches to sample preparations, data recording and phase retrieval are compared.
Imaging and tomography of biological cells with hard X-rays are associated with considerable challenges. For these weakly diffracting objects, it is much more difficult to reach high resolution and sufficient contrast than for most other samples. At the same time, hard X-rays can in principle probe the cell’s native electron-density distribution at subcellular resolution with quantitative contrast (Wilke et al., 2015 ▸), provided that the signal-to-noise ratio is sufficient. Furthermore, penetration power and depth of focus enable studies of cells embedded in complex environments as well as cells enclosed deeply within tissue (Krenkel et al., 2015 ▸). In this way, hard X-ray imaging can complement established imaging techniques such as fluorescence light microscopy, electron microscopy and soft X-ray microscopy (Larabell & Nugent, 2010 ▸).
Before turning to the experimental details we will hence first address the phase-retrieval approaches used in this paper. Note that these approaches are applicable to any propagation-based phase-contrast experiment, independent of the specific recording geometry, such as plane-wave or cone-beam illumination. For example, they could equally well be applied to propagation imaging with visible light, neutrons or electrons.
In this work we have benchmarked propagation-based phase-contrast imaging of biological cells, in particular mouse alveolar macrophages, in two and three dimensions, by comparison and optimization of different phase-retrieval approaches. In particular, we have demonstrated an improved holo-TIE algorithm, based on a four-distance recording, and a weighted phase map for low and high spatial frequencies. Importantly, the range of applicability of this approach is extremely high, as neither restrictive assumptions on the object’s optical properties (such as homogeneity or weakly varying phase) nor restrictive prior information such as support or sparsity is required. Further, the algorithm can be applied over the full range of Fresnel numbers and not only in the direct-contrast regime. Fig. 4 ▸ nicely demonstrates that the image quality of the holo-TIE approach surpasses conventional CTF and almost reaches that of iterative algorithms (mHIO, IRGN), which are computationally more complex and also require prior information.