Metal objects in the field of view cause artifacts in the image, which manifest as dark and bright streaks and degrade the diagnostic value of the image. Standard approaches for metal artifact reduction are often unable to correct these artifacts sufficiently or introduce new artifacts. We propose a new data-based method to reduce metal artifacts in CT images applying conditional Generative Adversarial Networks to the corrupted data. A generator network is applied directly to the corrupted projections by the metal objects to learn the corrected sinogram data. Further, two discriminator networks are used to evaluate the image quality of the enhanced data from the generator. The method was initially developed based on a supervised approach. However, there is usually no ground truth for actual clinical data without artifacts, which is needed to train the networks. Therefore, the method was further improved to train an unsupervised network, i.e., without the ground truth. In addition the input data, the neighboring slices and the stochastic components of the image are included using the latent space representation of the data. The results show that the trained generator network can reasonably replace the missing projection data and reduce the artifacts in the reconstructed image.
KEYWORDS: Magnetic particle imaging, In vivo imaging, Scanners, Visualization, Heart, Reconstruction algorithms, Data acquisition, Signal to noise ratio, 3D image processing, Magnetism
Magnetic particle imaging (MPI) is a highly sensitive imaging method that enables the visualization of magnetic tracer materials with a temporal resolution of more than 46 volumes per second. In MPI, the size of the field of view (FoV) scales with the strengths of the applied magnetic fields. In clinical applications, those strengths are limited by peripheral nerve stimulation, specific absorption rates, and the requirement to acquire images of high spatial resolution. Therefore, the size of the FoV is usually a few cubic centimeters. To bypass this limitation, additional focus fields and/or external object movements can be applied. The latter approach is investigated. An object is moved through the scanner bore one step at a time, whereas the MPI scanner continuously acquires data from its static FoV. Using a 3-D phantom and dynamic 3-D in vivo data, it is shown that the data from such a moving table experiment can be jointly reconstructed after reordering the data with respect to the stepwise object shifts and heart beat phases.
Magnetic Particle Imaging is capable of three-dimensional real-time imaging. Due to high spatial and temporal resolution, the method offers a great potential to be used in interventional scenarios. In this contribution, a design study integrating a single-sided coil assembly into a patient table is presented. An elliptical and an approximated elliptical coil topology are compared and proposed as alternatives to the commonly used circular shaped coils. Through this, the size of the field of view can be extended while not exceeding the lateral width of the patient table.
In Magnetic Particle Imaging the spatial distribution of superparamagnetic iron-oxide nanoparticles is determined using oscillating magnetic fields. The change of particle magnetization is recorded with receive coils. Spatial encoding is achieved with a superimposed gradient field featuring a field-free point. Particles not located in the vicinity of this point are in saturation and therefore do not induce a signal in the coils. Image reconstruction based on a system matrix is accurate, but time consuming. Recently, a method was introduced that images several small patches instead of one large field of view. This contribution applies this approach and additionally suggests to reusing the system matrix of one patch for the reconstruction of all patches. We will motivate this idea with symmetry characteristics of the magnetic fields applied in Magnetic Particle Imaging and perform a simulation study on homogeneous as well as inhomogeneous fields to show the potential of the approach.
Magnetic Particle Imaging (MPI) has been presented by Gleich and Weizenecker in 2005. Since then, a number of innovations have been introduced by many di erent research groups. In 2009, for instance, Sattel et al. presented a novel single-sided MPI scanner geometry. The major advantage of this particular scanner geometry is the unlimited measurement eld. For the imaging process in MPI, super-paramagnetic iron oxide nanoparticles (SPIONs) are applied as tracer material. The tracer is excited by sinusoidally varying magnetic elds. In this contribution, simulated magnetic elds were evaluated based on the measured eld distribution of a single-sided scanner realization. It is of particular importance to know the quality of the gradient elds, since image resolution in MPI is directly linked to the gradient strength.
In computed tomography, star shape artifacts are introduced by metal objects, which are inside a patient's
body. The quality of the reconstructed image can be enhanced by applying a metal artifact reduction method. Unfortunately, a method that removes all such artifacts in order to make the images valuable for medical diagnosis remains to be found. In this study, the influence of metal segmentation is investigated. A thresholding technique, which is the state of the art in the field, is compared with a manual segmentation. Results indicate that a more accurate segmentation can lead to a preservation of important anatomical details, which are of high value for medical diagnosis.
Photon counting detectors are expected to bring along various clinical benefits in CT imaging. Among the benefits of these detectors is their intrinsic spectral sensitivity that allows to resolve the incident X-ray spectrum. Their capability for multi-energy imaging enables material segmentation, but it is also possible to use the spectral information to create fused gray-scale CT images with improved imaging properties.
We have developed and investigated an optimization method that maximizes the image contrast-to-noise ratio, making use of the spectral information in data recorded with a counting detector with up to six energy thresholds.
The resulting merged gray-scale CT images exhibit significantly improved CNR2 for a number of clinically
established, potentially novel and hypothetical contrast agents in the thin absorber approximation.
In this work we motivate and describe the optimization method, provide the deduced optimal sets of threshold energies and mixing weights, and summarize the maximally achievable gain in CNR2 for each contrast agent under study.
Beside the original scanner geometry for Magnetic Particle Imaging (MPI) introduced by Gleich et. al. in 2005,1
alternative scanner geometries have been introduced.2-4 In excess of the opportunities in medical application
offered by MPI itself, these new scanner geometries permit additional medical application scenarios. Here, the
single-sided scanner geometry is implemented as imaging device for supporting the sentinel lymph node biopsy
concept. In this contribution, the medical application is outlined, and the geometry of the scanner device is
presented together with first simulation results providing information about the achievable image quality.
Digital Tomosynthesis (DT) is an x-ray limited-angle imaging technique. An accurate image reconstruction in
tomosynthesis is a challenging task due to the violation of the tomographic sufficiency conditions. A classical
"shift-and-add" algorithm (or simple backprojection) suffers from blurring artifacts, produced by structures
located above and below the plane of interest. The artifact problem becomes even more prominent in the presence
of materials and tissues with a high x-ray attenuation, such as bones, microcalcifications or metal. The focus of
the current work is on reduction of ghosting artifacts produced by bones in the musculoskeletal tomosynthesis.
A novel dissimilarity concept and a modified backprojection with an adaptive spatially dependent weighting
scheme (ωBP) are proposed. Simulated data of software phantom, a structured hardware phantom and a
human hand raw-data acquired with a Siemens Mammomat Inspiration tomosynthesis system were reconstructed
using conventional backprojection algorithm and the new ωBP-algorithm. The comparison of the results to the
non-weighted case demonstrates the potential of the proposed weighted backprojection to reduce the blurring
artifacts in musculoskeletal DT. The proposed weighting scheme is not limited to the tomosynthesis limitedangle
geometry. It can also be adapted for Computed Tomography (CT) and included in iterative reconstruction
algorithms (e.g. SART).
In Computed Tomography (CT) metal objects in the region of interest introduce data inconsistencies during acquisition.
The reconstruction process results in an image with star shaped artifacts. To enhance image quality the influence of
metal objects can be reduced by different metal artifact reduction (MAR) strategies. For an adequate evaluation of new
MAR approaches a ground truth reference data set is needed. In technical evaluations, where phantoms are available
with and without metal inserts, ground truth data can easily be acquired by a reference scan. Obviously, this is not
possible for clinical data.
In this work, three different evaluation methods for metal artifacts as well as comparison of MAR methods without the
need of an acquired reference data set will be presented and compared. The first metric is based on image contrast; a
second approach involves the filtered gradient information of the image, and the third method uses a forward projection
of the reconstructed image followed by a comparison with the actually measured projection data.
All evaluation techniques are performed on phantom and on clinical CT data with and without MAR and compared with
reference-based evaluation methods as well as expert-based classifications.
In clinical computed tomography (CT), images from patient examinations taken with conventional scanners
exhibit noise characteristics governed by electronics noise, when scanning strongly attenuating obese patients
or with an ultra-low X-ray dose. Unlike CT systems based on energy integrating detectors, a system with a
quantum counting detector does not suffer from this drawback. Instead, the noise from the electronics mainly
affects the spectral resolution of these detectors. Therefore, it does not contribute to the image noise in spectrally
non-resolved CT images. This promises improved image quality due to image noise reduction in scans obtained
from clinical CT examinations with lowest X-ray tube currents or obese patients. To quantify the benefits of
quantum counting detectors in clinical CT we have carried out an extensive simulation study of the complete
scanning and reconstruction process for both kinds of detectors. The simulation chain encompasses modeling
of the X-ray source, beam attenuation in the patient, and calculation of the detector response. Moreover,
in each case the subsequent image preprocessing and reconstruction is modeled as well. The simulation-based,
theoretical evaluation is validated by experiments with a novel prototype quantum counting system and a Siemens
Definition Flash scanner with a conventional energy integrating CT detector. We demonstrate and quantify the
improvement from image noise reduction achievable with quantum counting techniques in CT examinations with
ultra-low X-ray dose and strong attenuation.
KEYWORDS: Magnetism, Scanners, Magnetic particle imaging, Nanoparticles, 3D image processing, Particles, Computed tomography, 3D acquisition, Imaging systems, Magnetic resonance imaging
Magnetic Particle Imaging (MPI) has been introduced as a modality that allows for the acquisition of three-dimensional
functional images with high sensitivity in real time. Here, alternative coil topologies are presented that differ
significantly from the original set-up. Two novel coil topologies will be presented. Beside an asymmetric coil topology,
where all field generating coils are arranged on a single side, an effective coil assembly has been accomplished that
creates a field-free line for spatial coding. The alternative coil topologies may overcome the problem of a confined
measurement field or lead to an increase of the sensitivity of MPI.
Renal lesion detection and characterization using Computed Tomography is an important application in genitourinary
radiology. Although in general the detection of renal lesions has been shown to be exceedingly accuratce, the detection of
benign renal cysts is still problematic. Under certain circumstances, the attenuation values inside a cyst increase
incorrectly with an increase in the iodine concentration in the surrounding soft tissue. This so called pseudoenhancement
complicates the classification of cysts and creates severe difficulties to distinguish a benign nonenhancing lesion from an
enhancing mass.
In the present study, the standard procedure based on a single energy 120 kV mode is compared to three dual energy
modes available on the Siemens Somatom Definition Flash scanner.
In order to simulate the kidney and the lesions, several plastic rods were placed inside a small container filled with
different iodine concentrations. This phantom is then positioned inside water tanks of different sizes. The rods simulating
the lesions are made out of a special plastic with constant HU value throughout the relevant X-ray energy range.
During the project, three important aspects have been discovered: 1) for normal situations, a 100/140 Sn kV mode on the
Siemens Flash scanner is similar to the traditional single energy 120 kV mode. 2) For small patient sizes, all dual energy
modes show a reduction of pseudoenhancement. 3) For larger patients, only the 100/140 Sn kV mode results in a
reduction of pseudoenhancement. Both the 80/140 kV and the 80/140 Sn kV mode show a worse performance than the
120 kV single energy mode in a very large phantom size.
A novel approach for coupling brain tumor mass effect with a continuous model of cancer progression is proposed.
The purpose of the present work is to devise an efficient approximate model for the mechanical interaction of
the tumor with its surroundings in order to aid registration of brain tumor images with statistical atlases as well
as the generation of atlases of brain tumor disease.
To model tumor progression a deterministic reaction-diffusion formalism, which describes the spatio-temporal
dynamics of a coarse-grained population density of cancerous cells, is discretized on a regular grid. Tensor
information obtained from a probabilistic atlas is used to model the anisotropy of the diffusion of malignant cells
within white matter. To account for the expansive nature of the tumor a parametric deformation model is linked
to the computed net cell density of cancerous cells. To this end, we formulate a constrained optimization problem
using an inhomogeneous regularization that in turn allows for approximating physical properties of brain tissue.
The described coupling model can in general be applied to estimate mass effect of non-convex, diffusive as well
as multifocal tumors so that no simplification of the growth model has to be stipulated.
The present work has to be considered as a proof-of-concept. Visual assessment of the computed results
demonstrates the potential of the described method. We conclude that the analogy to the problem formulation in
image registration potentially allows for a sensible integration of the described approach into a unified framework
of image registration and tumor modeling.
Magnetic particle imaging (MPI) is a new tomographic imaging technique capable of determining the spatial
distribution of superparamagnetic iron oxide particles at high temporal and spatial resolution. Reconstruction
of the particle distribution requires the system function to be known. In almost all other tomographic imaging
techniques, a basic mathematical model of the system function exists, so that for reconstruction of an image,
only measured data from the object under examination have to be provided. Due to the complex behavior of
the particle dynamics, this is more complicated in MPI. Therefore, to date, the system function is measured in a
tedious calibration procedure. To this end, a small delta sample is moved to each position inside the measuring
field, while the magnetization response is acquired consecutively. However, although this measurement-based
approach provides a good estimate of the system function, it has several drawbacks. Most important, the
measured system function contains noise, which limits the size of the delta sample and in turn the resolution
of the sampling grid. In this work, the noise induced limitations of the measurement-based system function are
investigated in a simulation study. More precisely, the influence of the system function noise and the size of the
delta sample on the resulting image quality after reconstruction are analyzed.
Under normal circumstances the quality of images reconstructed with the classic FBP CT reconstruction algorithm is adequate for medical diagnosis. However, in some special cases the assumptions made by this method are not applicable because of non-linearities in the underlying physical imaging processes. Especially in the presence of metal implants in the field of view, effects like beam hardening, scatter and photon starvation result in serious streaking and banding artifacts around and between these objects. In order to reduce the artifacts, several different types of correction methods were introduced during the last two decades. In one of the most often used approaches, an interpolation scheme is used to replace all corrupted beam data in the shadow of the metal with artificially generated values. Although this leads to a reduction of the most severe artifacts, typically the results are far from being perfect. Instead of removing all artifacts, in most cases new streak artifacts are introduced. In the present work it is shown that the origin of these new artifacts is related to the loss of edge information of the objects by using surrogate data. The application of a more sophisticated artifact reduction method based on a segmentation of a preliminary reconstructed image decreases the number of newly introduced artifacts to a large degree. This is possible, because edge information between air and tissue recovered from the preliminary reconstruction can be included into the correction scheme. It is concluded that a restoration scheme without additionally information is not sufficient for a successful metal artifact reduction method.
In this work different surrogate data strategies to reduce metal artifacts in reconstructed CT images are tested.
Inconsistent sinogram projection data caused by e.g. beam hardening are the origin of metal artifacts in the
reconstructed images. The goal of this work is to replace this inconsistent projection data by artificially generated data.
Therefore, here, two 1D interpolation strategies, a directional interpolation based upon the sinogram 'flow' and a 1D
interpolation by means of the non-equispaced fast Fourier transform are compared to a fully 2D method based upon the
idea of image inpainting. Due to the fact that the artificially generated data never perfectly fit the gap inside the
projection data caused by the inconsistencies, those repaired sinogram data are reconstructed using a weighted
Maximum Likelihood Expectation Maximization algorithm called λ-MLEM algorithm. In this way, the artificially
generated data, still contaminated with residual inconsistencies, are weighted less during reconstruction.
Today's digital radiography systems mostly use unsharp masking-like image enhancement techniques based on splitting input images into two or three frequency channels. This method allows to enhance very small structures (edge enhancement) as well as enhancement of global contrast (harmonization). However, structures of medium size are not accessible by such enhancement. We develop and test a nonlinear enhancement algorithm based on hierarchically repeated unsharp masking, resulting in a multiscale architecture allowing consistent access to structures of all sizes. The algorithm is noise- resistant in the sense that it prevents unacceptable noise amplification. Clinical tests performed in the radiology departments of two major German hospitals so far strongly indicate the superior performance and high acceptance of the new processing.
2D/3D registration makes it possible to use pre-operative CT scans for navigation purposes during X-ray fluoroscopy guided interventions. We present a fast voxel-based method for this registration task, which uses a recently introduced similarity measure (pattern intensity). This measure is especially suitable for 2D/3D registration, because it is robust with respect to structures such as a stent visible in the X-ray fluoroscopy image but not in the CT scan. The method uses only a part of the CT scan for the generation of digitally reconstructed radiographs (DRRs) to accelerate their computation. Nevertheless, computation time is crucial for intra-operative application and a further speed-up is required, because numerous DRRs must be computed. For that reason, the suitability of different volume rendering methods for 2D/3D registration has been investigated. A method based on the shear-warp factorization of the viewing transformation turned out to be especially suitable and builds the basis of the registration algorithm. The algorithm has been applied to images of a spine phantom and to clinical images. For comparison, registration results have been calculated using ray-casting. The shear-warp factorization based rendering method accelerates registration by a factor of up to seven compared to ray-casting without degrading registration accuracy. Using a vertebra as feature for registration, computation time is in the range of 3-4s (Sun UltraSparc, 300 MHz) which is acceptable for intra-operative application.
Four main problems have to be solved for template matching based motion compensation in digital subtraction angiography. All the problems are concerned with the similarity measure that is the objective function to be optimized within the template matching procedure: (1) Due to the injection of contrast agent, mask and contrast image are dissimilar, which degrades the quality of some similarity measures. (2) Homogeneous areas in the fluoroscopic images lead to an insufficient quality of the similarity measure. (3) Shift invariant structures in fluoroscopic images (e.g. straight lines or edges) lead to a ridge-like objective function that potentially gives wrong results from the optimization procedure: If a ridge-like structure of the objective function is present, movements along this direction cannot be detected. Therefore, the local accuracy of the estimated motion component parallel to these directions must be ranked low while the corresponding orthogonal direction must be ranked high. We here present a technique to obtain local, directional rankings from the shape of the objective function which especially improves the quality of DSA images obtained from peripheral areas like the shinbone. (4) Inhomogeneous movements inside a single template lead to ambiguous or even irrelevant optima of the objective function: This problem is out of the scope of the present paper and will therefore not be addressed here. The performance of the point-based registration using different weightings in the least squares procedure (equally, isotropic and anisotropic weighting) has been compared. The isotropic and anisotropic weighting turned out to be superior to the equally weighted least squares procedure.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.