Reconstruction of scalar and vectorial components in X-ray dark-field tomography.

Grating-based X-ray dark-field imaging is a novel technique for obtaining image contrast for object structures at size scales below setup resolution. Such an approach appears particularly beneficial for medical imaging and nondestructive testing. It has already been shown that the dark-field signal depends on the direction of observation. However, up to now, algorithms for fully recovering the orientation dependence in a tomographic volume are still unexplored. In this publication, we propose a reconstruction method for grating-based X-ray dark-field tomography, which models the orientation-dependent signal as an additional observable from a standard tomographic scan. In detail, we extend the tomographic volume to a tensorial set of voxel data, containing the local orientation and contributions to dark-field scattering. In our experiments, we present the first results of several test specimens exhibiting a heterogeneous composition in microstructure, which demonstrates the diagnostic potential of the method.

Reliable Out-of-Distribution Recognition of Synthetic Images.

Generative adversarial networks (GANs) and diffusion models (DMs) have revolutionized the creation of synthetically generated but realistic-looking images. Distinguishing such generated images from real camera captures is one of the key tasks in current multimedia forensics research. One particular challenge is the generalization to unseen generators or post-processing. This can be viewed as an issue of handling out-of-distribution inputs. Forensic detectors can be hardened by the extensive augmentation of the training data or specifically tailored networks. Nevertheless, such precautions only manage but do not remove the risk of prediction failures on inputs that look reasonable to an analyst but in fact are out of the training distribution of the network. With this work, we aim to close this gap with a Bayesian Neural Network (BNN) that provides an additional uncertainty measure to warn an analyst of difficult decisions. More specifically, the BNN learns the task at hand and also detects potential confusion between post-processing and image generator artifacts. Our experiments show that the BNN achieves on-par performance with the state-of-the-art detectors while producing more reliable predictions on out-of-distribution examples.

Comparison of methods for sensitivity correction in Talbot-Lau computed tomography.

PURPOSE: In Talbot-Lau X-ray phase contrast imaging, the measured phase value depends on the position of the object in the measurement setup. When imaging large objects, this may lead to inhomogeneous phase contributions within the object. These inhomogeneities introduce artifacts in tomographic reconstructions of the object. METHODS: In this work, we compare recently proposed approaches to correct such reconstruction artifacts. We compare an iterative reconstruction algorithm, a known operator network and a U-net. The methods are qualitatively and quantitatively compared on the Shepp-Logan phantom and on the anatomy of a human abdomen. We also perform a dedicated experiment on the noise behavior of the methods. RESULTS: All methods were able to reduce the specific artifacts in the reconstructions for the simulated and virtual real anatomy data. The results show method-specific residual errors that are indicative for the inherently different correction approaches. While all methods were able to correct the artifacts, we report a different noise behavior. CONCLUSION: The iterative reconstruction performs very well, but at the cost of a high runtime. The known operator network shows consistently a very competitive performance. The U-net performs slightly worse, but has the benefit that it is a general-purpose network that does not require special application knowledge.

Toward Bridging the Simulated-to-Real Gap: Benchmarking Super-Resolution on Real Data.

Capturing ground truth data to benchmark super-resolution (SR) is challenging. Therefore, current quantitative studies are mainly evaluated on simulated data artificially sampled from ground truth images. We argue that such evaluations overestimate the actual performance of SR methods compared to their behavior on real images. Toward bridging this simulated-to-real gap, we introduce the Super-Resolution Erlangen (SupER) database, the first comprehensive laboratory SR database of all-real acquisitions with pixel-wise ground truth. It consists of more than 80k images of 14 scenes combining different facets: CMOS sensor noise, real sampling at four resolution levels, nine scene motion types, two photometric conditions, and lossy video coding at five levels. As such, the database exceeds existing benchmarks by an order of magnitude in quality and quantity. This paper also benchmarks 19 popular single-image and multi-frame algorithms on our data. The benchmark comprises a quantitative study by exploiting ground truth data and qualitative evaluations in a large-scale observer study. We also rigorously investigate agreements between both evaluations from a statistical perspective. One interesting result is that top-performing methods on simulated data may be surpassed by others on real data. Our insights can spur further algorithm development, and the publicy available dataset can foster future evaluations.

A gentle introduction to deep learning in medical image processing.

This paper tries to give a gentle introduction to deep learning in medical image processing, proceeding from theoretical foundations to applications. We first discuss general reasons for the popularity of deep learning, including several major breakthroughs in computer science. Next, we start reviewing the fundamental basics of the perceptron and neural networks, along with some fundamental theory that is often omitted. Doing so allows us to understand the reasons for the rise of deep learning in many application domains. Obviously medical image processing is one of these areas which has been largely affected by this rapid progress, in particular in image detection and recognition, image segmentation, image registration, and computer-aided diagnosis. There are also recent trends in physical simulation, modeling, and reconstruction that have led to astonishing results. Yet, some of these approaches neglect prior knowledge and hence bear the risk of producing implausible results. These apparent weaknesses highlight current limitations of deep ()learning. However, we also briefly discuss promising approaches that might be able to resolve these problems in the future.

A 3-D Projection Model for X-ray Dark-field Imaging.

The X-ray dark-field signal can be measured with a grating-based Talbot-Lau interferometer. It measures small angle scattering of micrometer-sized oriented structures. Interestingly, the signal is a function not only of the material, but also of the relative orientation of the sample, the X-ray beam direction, and the direction of the interferometer sensitivity. This property is very interesting for potential tomographically reconstructing structures below the imaging resolution. However, tomographic reconstruction itself is a substantial challenge. A key step of the reconstruction algorithm is the inversion of a forward projection model. In this work, we propose a very general 3-D projection model. We derive the projection model under the assumption that the observed scatter distribution has a Gaussian shape. We theoretically show the consistency of our model with existing, more constrained 2-D models. Furthermore, we experimentally show the compatibility of our model with simulations and real dark-field measurements. We believe that this 3-D projection model is an important step towards more flexible trajectories and, by extension, dark-field imaging protocols that are much better applicable in practice.

Simulation study on X-ray phase contrast imaging with dual-phase gratings.

PURPOSE: Two phase gratings in an X-ray grating interferometers can solve several technical challenges for clinical use of X-ray phase contrast. In this work, we adapt and evaluate this setup design to clinical X-ray sources and detectors in a simulation study. METHODS: For a given set of gratings, we optimize the remaining parameter space of a dual-phase grating setup using a numerical wave front simulation. The simulation results are validated with experimentally obtained visibility measurements on a setup with a microfocus tube and a clinical X-ray detector. We then confirm by simulation that the Lau condition for the [Formula: see text] grating also holds for two phase gratings. Furthermore, we use a [Formula: see text] grating with a fixed period to search for periods of matching phase grating configurations. RESULTS: Simulated and experimental visibilities agree very well. We show that the Lau condition for a dual-phase grating setup requires the interference patterns of the first phase grating to constructively overlay at the second phase grating. Furthermore, a total of three setup variants for given [Formula: see text] periods were designed with the simulation, resulting in visibilities between 4.5 and 9.1%. CONCLUSION: Dual-phase gratings can be used and optimized for a medical X-ray source and detector. The obtained visibilities are somewhat lower than for other Talbot-Lau interferometers and are a tradeoff between setup length and spatial resolution (or additional phase stepping, respectively). However, these disadvantage appears minor compared to the overall better photon statistics, and the fact that dual-phase grating setups can be expected to scale to higher X-ray energies.

Talbot-Lau x-ray phase-contrast setup for fast scanning of large samples.

Compared to conventional attenuation x-ray radiographic imaging, the x-ray Talbot-Lau technique provides further information about the scattering and the refractive properties of the object in the beam path. Hence, this additional information should improve the diagnostic process concerning medical applications and non-destructive testing. Nevertheless, until now, due to grating fabrication process, Talbot-Lau imaging suffers from small grating sizes (70 mm diameter). This leads to long acquisition times for imaging large objects. Stitching the gratings is one solution. Another one consists of scanning Talbot-Lau setups. In this publication, we present a compact and very fast scanning setup which enables imaging of large samples. With this setup a maximal scanning velocity of 71.7 mm/s is possible. A resolution of 4.1 lines/mm can be achieved. No complex alignment procedures are necessary while the field of view comprises 17.5 × 150 cm2. An improved reconstruction algorithm concerning the scanning approach, which increases robustness with respect to mechanical instabilities, has been developed and is presented. The resolution of the setup in dependence of the scanning velocity is evaluated. The setup imaging qualities are demonstrated using a human knee ex-vivo as an example for a high absorbing human sample.

Implementation of a Talbot-Lau interferometer in a clinical-like c-arm setup: A feasibility study.

X-ray grating-based phase-contrast imaging has raised interest regarding a variety of potential clinical applications, whereas the method is feasible using a medical x-ray tube. Yet, the transition towards a clinical setup remains challenging due to the requirement of mechanical robustness of the interferometer and high demands applying to medical equipment in clinical use. We demonstrate the successful implementation of a Talbot-Lau interferometer in an interventional c-arm setup. The consequence of vibrations induced by the rotating anode of the tube is discussed and the prototype is shown to provide a visibility of 21.4% at a tube voltage of 60 kV despite the vibrations. Regarding clinical application, the prototype is mainly set back due to the limited size of the field of view covering an area of 17 mm × 46 mm. A c-arm offers the possibility to change the optical axis according to the requirements of the medical examination. We provide a method to correct for artifacts that result from the angulation of the c-arm. Finally, the images of a series of measurements with the c-arm in different angulated positions are shown. Thereby, it is sufficient to perform a single reference measurement in parking position that is valid for the complete series despite angulation.

Talbot-Lau X-ray phase contrast for tiling-based acquisitions without reference scanning.

PURPOSE: Grating-based Talbot-Lau interferometers are a popular choice for phase-contrast X-ray acquisitions. Here, an air reference scan has to be acquired prior to an object scan. This particularly complicates acquisition of large objects: large objects are tiled into multiple scans due to the small field of view of current gratings. However, phase reference drifts occurring between these scans may require to repeatedly move the object in and out of the X-ray beam to update the reference information. METHODS: We developed an image processing technique that completely removes the need for phase reference scans in tiled acquisitions. We estimate the reference from object scans using a tailored iterated robust regression, using a novel efficient optimizer. RESULTS: Our evaluation indicates that the estimated reference is not only close to the acquired reference but also improves the final image quality. We hypothesize that this is because we mitigate errors that are introduced when actually acquiring the reference phase. CONCLUSION: Phase-contrast imaging of larger objects may benefit from computational estimation of phase reference data due to reduced scanning complexity and improved image quality.

Improved reconstruction of phase-stepping data for Talbot-Lau x-ray imaging.

Grating-based Talbot-Lau x-ray interferometry is a popular method for measuring absorption, phase shift, and small-angle scattering. The standard acquisition method for this modality is phase stepping, where the Talbot pattern is reconstructed from multiple images acquired at different grating positions. We review the implicit assumptions in phase-stepping reconstruction, and find that the assumptions of perfectly known grating positions and homoscedastic noise variance are violated in some scenarios. Additionally, we investigate a recently reported estimation bias in the visibility and dark-field signal. To adapt the phase-stepping reconstruction to these findings, we propose three improvements to the reconstruction. These improvements are (a) to use prior knowledge to compute more accurate grating positions to reduce moiré artifacts, (b) to utilize noise variance information to reduce dark-field and phase noise in high-visibility acquisitions, and (c) to perform correction of an estimation bias in the interferometer visibility, leading to more quantitative dark-field imaging in acquisitions with a low signal-to-noise ratio. We demonstrate the benefit of our methods on simulated data, as well as on images acquired with a Talbot-Lau interferometer.

Kinect-Based Correction of Overexposure Artifacts in Knee Imaging with C-Arm CT Systems.

Objective. To demonstrate a novel approach of compensating overexposure artifacts in CT scans of the knees without attaching any supporting appliances to the patient. C-Arm CT systems offer the opportunity to perform weight-bearing knee scans on standing patients to diagnose diseases like osteoarthritis. However, one serious issue is overexposure of the detector in regions close to the patella, which can not be tackled with common techniques. Methods. A Kinect camera is used to algorithmically remove overexposure artifacts close to the knee surface. Overexposed near-surface knee regions are corrected by extrapolating the absorption values from more reliable projection data. To achieve this, we develop a cross-calibration procedure to transform surface points from the Kinect to CT voxel coordinates. Results. Artifacts at both knee phantoms are reduced significantly in the reconstructed data and a major part of the truncated regions is restored. Conclusion. The results emphasize the feasibility of the proposed approach. The accuracy of the cross-calibration procedure can be increased to further improve correction results. Significance. The correction method can be extended to a multi-Kinect setup for use in real-world scenarios. Using depth cameras does not require prior scans and offers the possibility of a temporally synchronized correction of overexposure artifacts. To achieve this, we develop a cross-calibration procedure to transform surface points from the Kinect to CT voxel coordinates.

Multi-illuminant estimation with conditional random fields.

Most existing color constancy algorithms assume uniform illumination. However, in real-world scenes, this is not often the case. Thus, we propose a novel framework for estimating the colors of multiple illuminants and their spatial distribution in the scene. We formulate this problem as an energy minimization task within a conditional random field over a set of local illuminant estimates. In order to quantitatively evaluate the proposed method, we created a novel data set of two-dominant-illuminant images comprised of laboratory, indoor, and outdoor scenes. Unlike prior work, our database includes accurate pixel-wise ground truth illuminant information. The performance of our method is evaluated on multiple data sets. Experimental results show that our framework clearly outperforms single illuminant estimators as well as a recently proposed multi-illuminant estimation approach.

STXM goes 3D: digital reconstruction of focal stacks as novel approach towards confocal soft x-ray microscopy.

Fresnel zone plate based soft x-ray transmission microspectroscopy has developed into a routine technique for high-resolution elemental or chemical 2D imaging of thin film specimens. The availability of high resolution Fresnel lenses with short depth of focus offers the possibility of optical slicing (in the third dimension) by focus series with resolutions in the submicron regime. We introduce a 3D reconstruction algorithm that uses a variance-based metric to assign a focus measure as basis for volume rendering. The algorithm is applied to simulated geometries and opaque soft matter specimens thus enabling 3D visualization. These studies with z-resolution of few 100nm serve as important step towards the vision of a confocal transmission x-ray microscope.

Signal decomposition for X-ray dark-field imaging.

Grating-based X-ray dark-field imaging is a new imaging modality. It allows the visualization of structures at micrometer scale due to small-angle scattering of the X-ray beam. However, reading darkfield images is challenging as absorption and edge-diffraction effects also contribute to the dark-field signal, without adding diagnostic value. In this paper, we present a novel--and to our knowledge the first--algorithm for isolating small-angle scattering in dark-field images, which greatly improves their interpretability. To this end, our algorithm utilizes the information available from the absorption and differential phase images to identify clinically irrelevant contributions to the dark-field image. Experimental results on phantom and ex-vivo breast data promise a greatly enhanced diagnostic value of dark-field images.

CONRAD--a software framework for cone-beam imaging in radiology.

PURPOSE: In the community of x-ray imaging, there is a multitude of tools and applications that are used in scientific practice. Many of these tools are proprietary and can only be used within a certain lab. Often the same algorithm is implemented multiple times by different groups in order to enable comparison. In an effort to tackle this problem, the authors created CONRAD, a software framework that provides many of the tools that are required to simulate basic processes in x-ray imaging and perform image reconstruction with consideration of nonlinear physical effects. METHODS: CONRAD is a Java-based state-of-the-art software platform with extensive documentation. It is based on platform-independent technologies. Special libraries offer access to hardware acceleration such as OpenCL. There is an easy-to-use interface for parallel processing. The software package includes different simulation tools that are able to generate up to 4D projection and volume data and respective vector motion fields. Well known reconstruction algorithms such as FBP, DBP, and ART are included. All algorithms in the package are referenced to a scientific source. RESULTS: A total of 13 different phantoms and 30 processing steps have already been integrated into the platform at the time of writing. The platform comprises 74.000 nonblank lines of code out of which 19% are used for documentation. The software package is available for download at http://conrad.stanford.edu. To demonstrate the use of the package, the authors reconstructed images from two different scanners, a table top system and a clinical C-arm system. Runtimes were evaluated using the RabbitCT platform and demonstrate state-of-the-art runtimes with 2.5 s for the 256 problem size and 12.4 s for the 512 problem size. CONCLUSIONS: As a common software framework, CONRAD enables the medical physics community to share algorithms and develop new ideas. In particular this offers new opportunities for scientific collaboration and quantitative performance comparison between the methods of different groups.