A comparison of absorption and phase contrast for X-ray imaging of biological cells

Deep learning to overcome Zernike phase-contrast nanoCT artifacts for automated micro-nano porosity segmentation in bone

Bone material contains a hierarchical network of micro- and nano-cavities and channels, known as the lacuna-canalicular network (LCN), that is thought to play an important role in mechanobiology and turnover. The LCN comprises micrometer-sized lacunae, voids that house osteocytes, and submicrometer-sized canaliculi that connect bone cells. Characterization of this network in three dimensions is crucial for many bone studies. To quantify X-ray Zernike phase-contrast nanotomography data, deep learning is used to isolate and assess porosity in artifact-laden tomographies of zebrafish bones. A technical solution is proposed to overcome the halo and shade-off domains in order to reliably obtain the distribution and morphology of the LCN in the tomographic data. Convolutional neural network (CNN) models are utilized with increasing numbers of images, repeatedly validated by `error loss' and `accuracy' metrics. U-Net and Sensor3D CNN models were trained on data obtained from two different synchrotron Zernike phase-contrast transmission X-ray microscopes, the ANATOMIX beamline at SOLEIL (Paris, France) and the P05 beamline at PETRA III (Hamburg, Germany). The Sensor3D CNN model with a smaller batch size of 32 and a training data size of 70 images showed the best performance (accuracy 0.983 and error loss 0.032). The analysis procedures, validated by comparison with human-identified ground-truth images, correctly identified the voids within the bone matrix. This proposed approach may have further application to classify structures in volumetric images that contain non-linear artifacts that degrade image quality and hinder feature identification.

Which resolution?

The relationship between the contrast to noise ratio and intensity based cross-correlation coefficients for both protein crystallography and X-ray imaging are compared. It is concluded that, for protein crystallography at near atomic resolution, the intensity based cross-correlation coefficients give a reasonable indication of the quality of the corresponding electron density. For X-ray imaging of biological materials such as cells and soft tissue, the wide range of contrast of the features means that intensity based correlation coefficients can give a poor indication of the interpretability of an image. Rather than the term resolution, it is the contrast to noise ratio for a feature of interest at the relevant spatial frequency that is more relevant. Additional metrics are required to describe the quality of an image, and these are discussed.

Water-window x-ray emission from laser-produced Au plasma under optimal target thickness and focus conditions

Optimal laser irradiation conditions for water-window (WW) x-ray emission (2.3-4.4 nm) from an Au plasma are investigated to develop a laboratory-scale WW x-ray source. A minimum Au target thickness of 1 µm is obtained for a laser intensity of ∼10^{13} W/cm^{2} by observing the intensity drop in the WW spectra. Au targets produced by thermal evaporation are found to have a higher conversion efficiency than commercial foil targets for WW x-ray radiation. In addition, optimal laser spots for fixed laser energies (240 and 650 mJ) are found for an Au target ∼1 mm in front of the focal point, where suitable conditions for plasma temperature and plume volume coupling are achieved. The mechanism of the optimal target thickness and spot size can be well explained using a radiation hydrodynamic simulation code.

Coherent and incoherent imaging of biological specimens with electrons and X-rays

Since radiation damage is proportional to fluence, radiation damage limits the spatial resolution of biological structures determined by either X-ray or electron scattering. If only elastic scattering is used for structural information then electrons are superior as the ratio of elastic to inelastic scattering is higher than for X-rays. For soft X-rays in the water window below the O K edge photoabsorption contrast might be better than elastic scattering for distinguishing different biological materials. Phase contrast elastic scattering is most effective in the hard X-ray region up to about 10 keV. Radiation damage limits spatial resolution for most X-ray imaging to 10-20 nm. Local molar concentrations of Na,K and Ca ions can be determined at somewhat lower spatial resolutions using relevant absorption edges. At higher energies resolution res is only limited by the fluence available from the light source, since energy deposition is small.

The achievable resolution for X-ray imaging of cells and other soft biological material

X-ray imaging of soft materials is often difficult because of the low contrast of the components. This particularly applies to frozen hydrated biological cells where the feature of interest can have a similar density to the surroundings. As a consequence, a high dose is often required to achieve the desired resolution. However, the maximum dose that a specimen can tolerate is limited by radiation damage. Results from 3D coherent diffraction imaging (CDI) of frozen hydrated specimens have given resolutions of ∼80 nm compared with the expected resolution of 10 nm predicted from theoretical considerations for identifying a protein embedded in water. Possible explanations for this include the inapplicability of the dose-fractionation theorem, the difficulty of phase determination, an overall object-size dependence on the required fluence and dose, a low contrast within the biological cell, insufficient exposure, and a variety of practical difficulties such as scattering from surrounding material. A recent article [Villaneuva-Perez et al. (2018), Optica, 5, 450-457] concluded that imaging by Compton scattering gave a large dose advantage compared with CDI because of the object-size dependence for CDI. An object-size dependence would severely limit the applicability of CDI and perhaps related coherence-based methods for structural studies. This article specifically includes the overall object size in the analysis of the fluence and dose requirements for coherent imaging in order to investigate whether there is a dependence on object size. The applicability of the dose-fractionation theorem is also discussed. The analysis is extended to absorption-based imaging and imaging by incoherent scattering (Compton) and fluorescence. This article includes analysis of the dose required for imaging specific low-contrast cellular organelles as well as for protein against water. This article concludes that for both absorption-based and coherent diffraction imaging, the dose-fractionation theorem applies and the required dose is independent of the overall size of the object. For incoherent-imaging methods such as Compton scattering, the required dose depends on the X-ray path length through the specimen. For all three types of imaging, the dependence of fluence and dose on a resolution d goes as 1/d 4 when imaging uniform-density voxels. The independence of CDI on object size means that there is no advantage for Compton scattering over coherent-based imaging methods. The most optimistic estimate of achievable resolution is 3 nm for imaging protein molecules in water/ice using lensless imaging methods in the water window. However, the attainable resolution depends on a variety of assumptions including the model for radiation damage as a function of resolution, the efficiency of any phase-retrieval process, the actual contrast of the feature of interest within the cell and the definition of resolution itself. There is insufficient observational information available regarding the most appropriate model for radiation damage in frozen hydrated biological material. It is advocated that, in order to compare theory with experiment, standard methods of reporting results covering parameters such as the feature examined (e.g. which cellular organelle), resolution, contrast, depth of the material (for 2D), estimate of noise and dose should be adopted.

Comparison of high-resolution synchrotron-radiation-based phase-contrast imaging and absorption-contrast imaging for evaluating microstructure of vascular networks in rat brain: from 2D to 3D views

Conventional imaging methods such as magnetic resonance imaging, computed tomography and digital subtraction angiography have limited temporospatial resolutions and shortcomings like invasive angiography, potential allergy to contrast agents, and image deformation, that restrict their application in high-resolution visualization of the structure of microvessels. In this study, through comparing synchrotron radiation (SR) absorption-contrast imaging to absorption phase-contrast imaging, it was found that SR-based phase-contrast imaging could provide more detailed ultra-high-pixel images of microvascular networks than absorption phase-contrast imaging. Simultaneously, SR-based phase-contrast imaging was used to perform high-quality, multi-dimensional and multi-scale imaging of rat brain angioarchitecture. With the aid of image post-processing, high-pixel-size two-dimensional virtual slices can be obtained without sectioning. The distribution of blood supply is in accordance with the results of traditional tissue staining. Three-dimensional anatomical maps of cerebral angioarchitecture can also be acquired. Functional partitions of regions of interest are reproduced in the reconstructed rat cerebral vascular networks. Imaging analysis of the same sample can also be displayed simultaneously in two- and three-dimensional views, which provides abundant anatomical information together with parenchyma and vessels. In conclusion, SR-based phase-contrast imaging holds great promise for visualizing microstructure of microvascular networks in two- and three-dimensional perspectives during the development of neurovascular diseases.

3D Quantitative and Ultrastructural Analysis of Mitochondria in a Model of Doxorubicin Sensitive and Resistant Human Colon Carcinoma Cells

Drug resistance remains a major obstacle in cancer treatment. Because mitochondria mediate metabolic reprogramming in cancer drug resistance, we focused on these organelles in doxorubicin sensitive and resistant colon carcinoma cells. We employed soft X-ray cryo nano-tomography to map three-dimensionally these cells at nanometer-resolution and investigate the correlation between mitochondrial morphology and drug resistance phenotype. We have identified significant structural differences in the morphology of mitochondria in the two strains of cancer cells, as well as lower amounts of Reactive oxygen species (ROS) in resistant than in sensitive cells. We speculate that these features could elicit an impaired mitochondrial communication in resistant cells, thus preventing the formation of the interconnected mitochondrial network as clearly detected in the sensitive cells. In fact, the qualitative and quantitative three-dimensional assessment of the mitochondrial morphology highlights a different structural organization in resistant cells, which reflects a metabolic cellular adaptation functional to survive to the offense exerted by the antineoplastic treatment.

A comparison of absorption and phase contrast for X-ray imaging of biological cells. Erratum

An error in the calculation for X-ray absorption imaging has been identified in the paper by Nave (2018) [J. Synchrotron Rad. 25, 1490-1504]. The required fluence and dose in the paper are a factor of ten too low for this mode of imaging.