This PDF file contains the front matter associated with SPIE Proceedings Volume 9412, including the Title Page, Copyright information, Table of Contents, Invited Panel Discussion, and Conference Committee listing.

Image artifacts generated by metal implants have been a problem associated with CT since its introduction. Recent techniques to mitigate this problem have included the utilization of certain Dual-Energy CT (DECT) features. DECT can produce virtual monochromatic spectral (VMS) images, simulating how the data would appear if scanned at a single x-ray energy (keV). High-keV VMS images can greatly reduce the severity of metal artifacts. A problem with these high-keV images is that contrast enhancement provided by all commercially-available contrast media is severely reduced. It is therefore impossible to generate VMS images with simultaneous high contrast and minimized metal artifact severity. Novel contrast agents based on higher atomic number elements can maintain contrast enhancement at the higher energy levels where artifacts are reduced. This study evaluated three such candidate elements: bismuth, tantalum, and tungsten, as well as two conventional contrast elements: iodine and barium. A water-based phantom with vials containing these five elements in solution, as well as different artifact-producing metal structures, was scanned with a DECT scanner capable of rapid operating voltage switching. In the VMS datasets, substantial reductions in the contrast were observed for iodine and barium, which suffered from contrast reduction of 97 and 91% respectively at 140 versus 40 keV. In comparison under the same conditions, the novel candidate agents demonstrated contrast enhancement reductions of only 20, 29 and 32% for tungsten, tantalum and bismuth respectively. At 140 versus 40 keV, metal artifact severity was reduced by 57-85% depending on the phantom configuration.

The implementation of multimodal imaging in medicine is highly beneficial as different physical properties may provide complementary information, augmented detection ability, and diagnosis verification. Nanoparticles have been recently used as contrast agents for various imaging modalities. Their significant advantage over conventional large-scale contrast agents is the ability of detection at early stages of the disease, being less prone to obstacles on their path to the target region, and possible conjunction to therapeutics. Copper ions play essential role in human health. They are used as a cofactor for multiple key enzymes involved in various fundamental biochemistry processes. Extremely small size copper oxide nanoparticles (CuO-NPs) are readily soluble in water with high colloidal stability yielding high bioavailability. The goal of this study was to examine the magnetic and acoustic characteristics of CuO-NPs in order to evaluate their potential to serve as contrast imaging agent for both MRI and ultrasound. CuO-NPs 7nm in diameter were synthesized by hot solution method. The particles were scanned using a 9.4T MRI and demonstrated a concentration dependent T1 relaxation time shortening phenomenon. In addition, it was revealed that CuO-NPs can be detected using the ultrasonic B-scan imaging. Finally, speed of sound based ultrasonic computed tomography was applied and showed that CuO-NPs can be clearly imaged. In conclusion, the preliminary results obtained, positively indicate that CuO-NPs may be imaged by both MRI and ultrasound. The results motivate additional in-vivo studies, in which the clinical utility of fused images derived from both modalities for diagnosis improvement will be studied.

Contrast enhancement is a key component of CT imaging and offer opportunities for optimization. The design and optimization of new techniques however requires orchestration with the scan parameters and further a methodology to relate contrast enhancement and injection function. In this study, we used such a methodology to develop a method, analytical inverse method, to predict the required injection function to achieve a desired contrast enhancement in a given organ by incorporation of a physiologically based compartmental model. The method was evaluated across 32 different target contrast enhancement functions for aorta, kidney, stomach, small intestine, and liver. The results exhibited that the analytical inverse method offers accurate performance with error in the range of 10% deviation between the predicted and desired organ enhancement curves. However, this method is incapable of predicting the injection function based on the liver enhancement. The findings of this study can be useful in optimizing contrast medium injection function as well as the scan timing to provide more consistency in the way that the contrast enhanced CT examinations are performed. To our knowledge, this work is one of the first attempts to predict the contrast material injection function for a desired organ enhancement curve.

In this work, we apply non-convex, sparsity exploiting regularization techniques to image reconstruction in computed tomography (CT).We modify the well-known total variation (TV) penalty to use a non-convex smooth hard threshold (SHT) penalty as opposed to the typical ℓ₁ norm. The SHT penalty is different from the p <1 norms in that it is bounded above and has bounded gradient as its argument approaches the zero vector. We propose a re-weighting scheme utilizing the Chambolle-Pock (CP) algorithm in an attempt to solve a data-error constrained optimization problem utilizing the SHT penalty and call the resulting algorithm SHTCP. We then demonstrate the algorithm on sparse-view reconstruction of a simulated breast phantom with noiseless and noisy data and compare the converged images to those generated by a CP algorithm solving the analogous data-error constrained problem utilizing the TV. We demonstrate that SHTCP allows for more accurate reconstruction in the case of sparse-view noisy data and, in the case of noiseless data, allows for accurate reconstruction from fewer views than its TV counterpart.

Traumatic brain injury (TBI) is a major cause of death and disability. The current front-line imaging modality for TBI detection is CT, which reliably detects intracranial hemorrhage (fresh blood contrast 30-50 HU, size down to 1 mm) in non-contrast-enhanced exams. Compared to CT, flat-panel detector (FPD) cone-beam CT (CBCT) systems offer lower cost, greater portability, and smaller footprint suitable for point-of-care deployment. We are developing FPD-CBCT to facilitate TBI detection at the point-of-care such as in emergent, ambulance, sports, and military applications. However, current FPD-CBCT systems generally face challenges in low-contrast, soft-tissue imaging. Model-based reconstruction can improve image quality in soft-tissue imaging compared to conventional filtered back-projection (FBP) by leveraging high-fidelity forward model and sophisticated regularization. In FPD-CBCT TBI imaging, measurement noise characteristics undergo substantial change following artifact correction, resulting in non-negligible noise amplification. In this work, we extend the penalized weighted least-squares (PWLS) image reconstruction to include the two dominant artifact corrections (scatter and beam hardening) in FPD-CBCT TBI imaging by correctly modeling the variance change following each correction. Experiments were performed on a CBCT test-bench using an anthropomorphic phantom emulating intra-parenchymal hemorrhage in acute TBI, and the proposed method demonstrated an improvement in blood-brain contrast-to-noise ratio (CNR = 14.2) compared to FBP (CNR = 9.6) and PWLS using conventional weights (CNR = 11.6) at fixed spatial resolution (1 mm edge-spread width at the target contrast). The results support the hypothesis that FPD-CBCT can fulfill the image quality requirements for reliable TBI detection, using high-fidelity artifact correction and statistical reconstruction with accurate post-artifact-correction noise models.

Combined PET/MR imaging allows to incorporate the high-resolution anatomical information delivered by MRI into the PET reconstruction algorithm for improvement of PET accuracy beyond standard corrections. We used the working hypothesis that glucose uptake in adipose tissue is low. Thus, our aim was to shift ¹⁸F-FDG PET signal into image regions with a low fat content. Dixon MR imaging can be used to generate fat-only images via the water/fat chemical shift difference. On the other hand, the Origin Ensemble (OE) algorithm, a novel Markov chain Monte Carlo method, allows to reconstruct PET data without the use of forward- and back projection operations. By adequate modifications to the Markov chain transition kernel, it is possible to include anatomical a priori knowledge into the OE algorithm. In this work, we used the OE algorithm to reconstruct PET data of a modified IEC/NEMA Body Phantom simulating body water/fat composition. Reconstruction was performed 1) natively, 2) informed with the Dixon MR fat image to down-weight ¹⁸F-FDG signal in fatty tissue compartments in favor of adjacent regions, and 3) informed with the fat image to up-weight ¹⁸F-FDG signal in fatty tissue compartments, for control purposes. Image intensity profiles confirmed the visibly improved contrast and reduced partial volume effect at water/fat interfaces. We observed a 17±2% increased SNR of hot lesions surrounded by fat, while image quality was almost completely retained in fat-free image regions. An additional in vivo experiment proved the applicability of the presented technique in practice, and again verified the beneficial impact of fat-constrained OE reconstruction on PET image quality.

This study investigates the feasibility of obtaining CT-derived 3D surfaces from data provided by the scanning-beam digital x-ray (SBDX) system. Simulated SBDX short-scan acquisitions of a Shepp-Logan and a thorax phantom containing a high contrast spherical volume were generated. 3D reconstructions were performed using a penalized weighted least squares method with total variation regularization (PWLS-TV), as well as a more efficient variant employing gridding of projection data to parallel rays (gPWLS-TV). Voxel noise, edge blurring, and surface accuracy were compared to gridded filtered back projection (gFBP). PWLS reconstruction of a noise-free reduced-size Shepp-Logan phantom had 1.4% rRMSE. In noisy gPWLS-TV reconstructions of a reduced-size thorax phantom, 99% of points on the segmented sphere perimeter were within 0.33, 0.47, and 0.70 mm of the ground truth, respectively, for fluences comparable to imaging through 18.0, 27.2, and 34.6 cm acrylic. Surface accuracies of gFBP and gPWLS-TV were similar at high fluences, while gPWLS-TV offered improvement at the lowest fluence. The gPWLS-TV voxel noise was reduced by 60% relative to gFBP, on average. High-contrast linespread functions measured 1.25 mm and 0.96 mm (FWHM) for gPWLS-TV and gFBP. In a simulation of gated and truncated projection data from a full-sized thorax, gPWLS-TV reconstruction yielded segmented surface points which were within 1.41 mm of ground truth. Results support the feasibility of 3D surface segmentation with SBDX. Further investigation of artifacts caused by data truncation and patient motion is warranted.

Model-based iterative reconstruction (MBIR) is implemented to process full clinical data sets of dedicated breast tomosynthesis (DBT) in a low dose condition and achieves less spreading of anatomical structure between slices. MBIR is a statistical based reconstruction which can control the trade-off between data fitting and image regularization. In this study, regularization is formulated with anisotropic prior weighting that independently controls the image regularization between in-plane and out-of-plane voxel neighbors. Studies at complete and partial convergence show that the appropriate formulation of data-fit and regularization terms along with anisotropic prior weighting leads to a solution with improved localization of objects within a more narrow range of slices. This result is compared with the solutions using simultaneous iterative reconstruction technique (SIRT), which is one of the state of art reconstruction in DBT. MBIR yields higher contrast-to-noise for medium and large size microcalcifications and diagnostic structures in volumetric breast images and supports opportunity for dose reduction for 3D breast imaging.

Relative to prospective projection gating, retrospective projection gating for dynamic CT applications allows fast imaging times, minimizing the potential for physiological and anatomic variability. Preclinically, fast imaging is attractive due to the rapid clearance of low molecular weight contrast agents and the rapid heart rate of rodents. Clinically, retrospective gating is relevant for intraoperative C-arm CT. More generally, retrospective sampling provides an opportunity for significant reduction in x-ray dose within the framework of compressive sensing theory and sparsity-constrained iterative reconstruction. Even so, CT reconstruction from projections with random temporal sampling is a very poorly conditioned inverse problem, requiring high fidelity regularization to minimize variability in the reconstructed results. Here, we introduce a highly novel data acquisition and regularization strategy for spectro-temporal (5D) CT reconstruction from retrospectively gated projections. We show that by taking advantage of the rank-sparse structure and separability of the temporal and spectral reconstruction sub-problems, being able to solve each sub-problem independently effectively guarantees that we can solve both problems together. In this paper, we show 4D simulation results (2D + 2 energies + time) using the proposed technique and compare them with two competing techniques— spatio-temporal total variation minimization and prior image constrained compressed sensing. We also show in vivo, 5D (3D + 2 energies + time) myocardial injury data acquired in a mouse, reconstructing 20 data sets (10 phases, 2 energies) and performing material decomposition from data acquired over a single rotation (360°, dose: ~60 mGy).

Compared to published amorphous-silicon (TFT) based X-ray detectors, crystalline silicon CMOS-based active-pixel detectors exploit the benefits of low noise, high speed, on-chip integration and featuring offered by CMOS technology. This presentation focuses on the specific advantage of high image quality at very low dose levels. The measurement of very low dose performance parameters like Detective Quantum Efficiency (DQE) and Noise Equivalent Dose (NED) is a challenge by itself. Second-order effects like defect pixel behavior, temporal and quantization noise effects, dose measurement accuracy and limitation of the x-ray source settings will influence the measurements at very low dose conditions. Using an analytical model to predict the low dose behavior of a detector from parameters extracted from shot-noise limited dose levels is presented. These models can also provide input for a simulation environment for optimizing the performance of future detectors. In this paper, models for predicting NED and the DQE at very low dose are compared to measurements on different CMOS detectors. Their validity for different sensor and optical stack combinations as well as for different x-ray beam conditions was validated.

The detective quantum efficiency (DQE) of an x-ray detector, expressed as a function of spatial frequency, describes the ability to produce high-quality images relative to an ideal detector. While the DQE normally decreases substantially with increasing frequency, we describe an approach that can be used to improve the DQE response by increasing the DQE at high spatial frequencies. The approach makes use of an apodized-aperture pixel (AAP) design that requires use of a high-resolution x-ray converter such as selenium coupled to a sensor array with very small physical sensor elements, such as CMOS sensors. While sensors with elements of 10 - 25 μm are too small for most practical applications in medical radiography, we describe how larger image pixels of a practical size can be synthesized to provide a better DQE than simple binning or using physical pixels of the same size. A theoretical cascaded-systems analysis shows the DQE at the image sampling cut-off frequency can be improved by up to a factor of 2.5x. The AAP approach was validated experimentally using a CMOS/CsI-based detector having 0.05-mm sensor elements. Using AAP images with 0.2-mm pixels, the high-frequency DQE value was increased from 0.2 to 0.4 compared to simple 4x4 binning. It is concluded that ultra-high-resolution sensors can be used to optimize the high-frequency performance of x-ray detectors and make substantial improvements in image quality for visualization of small stuctures and fine image detail in comparison to current imaging systems.

Active Matrix Flat Panel Imagers (AMFPI) based on an array of thin film transistors (TFT) have become the dominant technology for digital x-ray imaging. In low dose applications, the performance of both direct and indirect conversion detectors are limited by the electronic noise associated with the TFT array. New concepts of direct and indirect detectors have been proposed using avalanche amorphous selenium (a-Se), referred to as high gain avalanche rushing photoconductor (HARP). The indirect detector utilizes a planar layer of HARP to detect light from an x-ray scintillator and amplify the photogenerated charge. The direct detector utilizes separate interaction (non-avalanche) and amplification (avalanche) regions within the a-Se to achieve depth-independent signal gain. Both detectors require the development of large area, solid state HARP. We have previously reported the first avalanche gain in a-Se with deposition techniques scalable to large area detectors. The goal of the present work is to demonstrate the feasibility of large area HARP fabrication in an a-Se deposition facility established for commercial large area AMFPI. We also examine the effect of alternative pixel electrode materials on avalanche gain. The results show that avalanche gain > 50 is achievable in the HARP layers developed in large area coaters, which is sufficient to achieve x-ray quantum noise limited performance down to a single x-ray photon per pixel. Both chromium (Cr) and indium tin oxide (ITO) have been successfully tested as pixel electrodes.

Over the last ~15 years, the central goal in external beam radiotherapy of maximizing dose to the tumor while minimizing dose to surrounding normal tissues has been greatly facilitated by the development and clinical implementation of many innovations. These include megavoltage active matrix flat-panel imagers (MV AMFPIs) designed to image the treatment beam, and separate kilovoltage (kV) AMFPIs and x-ray sources designed to provide high-contrast projection and cone-beam CT images in the treatment room. While these systems provide clinically valuable information, a variety of advantages would accrue through introduction of the capability to produce clinically useful, high quality imaging information at multiple energies (e.g., kV and MV) from a single detector along the treatment beam direction. One possible approach for achieving this goal involves substitution of the x-ray converters used in conventional MV AMFPIs with thick, segmented crystalline scintillators designed for dual-energy operation, coupled with the addition of x-ray imaging beams that contain a significant diagnostic component. A second approach involves introduction of a large area, monolithic array of photon counting pixels with multiple energy thresholds and event counters, which could provide multi-spectral views of the treatment beam with improved contrast. In this paper, the motivations behind, and the merits of each approach are described. In addition, prospects for such dual-energy imagers and photon counting array designs are discussed in the context of the radiotherapy environment.

Flat panel imagers (FPI) are becoming the dominant detector technology for digital x-ray imaging. In indirect FPI, the scintillator that provides the highest image quality is Thallium (Tl) doped Cesium Iodide (CsI) with columnar structure. The maximum CsI thickness used in existing FPI is ~600 microns, due to concerns of loss in spatial resolution and light output with further increase in thickness. The goal of the present work is to investigate the screen-optics for CsI with thicknesses much larger than that used in existing FPI, so that the knowledge can be used to improve imaging performance in dose sensitive and higher energy applications, such as cone-beam CT (CBCT). Columnar CsI(Tl) scintillators up to 1 mm in thickness with different screen-optical design were investigated experimentally. Pulse height spectra (PHS) were measured to determine the Swank factor at x-ray energies between 25 and 75 keV, and to derive depth-dependent light escape efficiency i.e. gain. Detector presampling MTF, NPS and DQE were measured using a high-resolution CMOS optical sensor. Optical Monte Carlo simulation was performed to estimate optical parameters for each screen design and derive depth-dependent gain and MTF, from which overall MTF and DQE were calculated and compared with measured results. The depth-dependent imaging performance parameters were then used in a cascaded linear system model (CLSM) to investigate detector performance under screen- and sensor-side irradiation conditions. The methodology developed for understanding the optics of thick CsI(Tl) will lead to detector optimization in CBCT.

In a grating interferometer-based x-ray differential phase contrast (DPC) imaging system, an analyzer grating (i.e. a G2 grating) is typically used to help record the important refraction information obtained with the phase stepping technique. Such a method requires the sequential movement of the G2 grating as well as multiple x-ray exposures to perform phase stepping, and thus conventional DPC imaging is very time-consuming. Additionally, it also has some mechanical instability issues due to the movement of the G2 grating. To accelerate the data acquisition speed and achieve single shot x-ray DPC imaging with a collinear type G2 grating, in this study, a new signal extraction method had been investigated. With this alternative approach, a non-zero angle of rotation between the diffraction pattern (generated by the G1 grating) and the collinear G2 grating is used during the entire data acquisition. Due to this deliberate grating misalignment, a visible moiré pattern with a certain period shall be detected. Initial experiments have demonstrated that this new signal extraction method is able to provide us with three different types of signal: absorption, differential phase, and the dark field image signals. Although the spatial resolution for both the differential phase and the dark field images is blurred by several pixel length due to the used interpolation operation, the absorption image maintains the same spatial resolution as in the conventional x-ray imaging. This developed novel signal analysis method enables single shot DPC imaging and can greatly reduce the data acquisition time, thus facilitating the implementation of DPC imaging in the medical field.

Both the laser-plasma wakefield accelerator (LWFA) and X-ray phase-contrast imaging (XPCi) are promising technologies that are attracting the attention of the scientific community. Conventional X-ray absorption imaging cannot be used as a means of imaging biological material because of low contrast. XPCi overcomes this limitation by exploiting the variation of the refraction index of materials. The contrast obtained is higher than for conventional absorption imaging and requires a lower dose. The LWFA is a new concept of acceleration where electrons are accelerated to very high energy (~150 MeV) in very short distances (mm scale) by surfing plasma waves excited by the passage of an ultra-intense laser pulse (~10¹⁸ Wcm^-2) through plasma. Electrons in the LWFA can undergo transverse oscillation and emit synchrotron-like (betatron) radiation in a narrow cone around the propagation axis. The properties of the betatron radiation produced by LWFA, such as source size and spectrum, make it an excellent candidate for XPCi. In this work we present the characterization of betatron radiation produced by the LWFA in the ALPHA-X laboratory (University of Strathclyde). We show how phase contrast images can be obtained using the betatron radiation in a free-space propagation configuration and we discuss the potential and limitation of the LWFA driven XPCi.

Edge illumination (EI) X-ray phase-contrast imaging (XPCI) has potential for applications in different fields of research, including materials science, non-destructive industrial testing, small-animal imaging, and medical imaging. One of its main advantages is the compatibility with laboratory equipment, in particular with conventional non-microfocal sources, which makes its exploitation in normal research laboratories possible. In this work, we demonstrate that the signal in laboratory implementations of EI can be correctly described with the use of the simplified geometrical optics. Besides enabling the derivation of simple expressions for the sensitivity and spatial resolution of a given EI setup, this model also highlights the EI’s achromaticity. With the aim of improving image quality, as well as to take advantage of the fact that all energies in the spectrum contribute to the image contrast, we carried out EI acquisitions using a photon-counting energy-resolved detector. The obtained results demonstrate that this approach has great potential for future laboratory implementations of EI.

We present the results from a benchtop X-ray phase-contrast (XPC) method for lung imaging that represents a paradigm shift in the way small animal lung imaging is performed. In our method, information regarding airway microstructure that is encoded within speckle texture of a single XPC radiograph is decoded to spatially resolve changes in lung properties such as microstructure sizes, air volumes, and compliance, to name a few. Such functional information cannot be derived from conventional lung radiography or any other 2D imaging modality. By computing these images at different time points within a breathing cycle, dynamic functional imaging can be potentially achieved without the need for tomography.

Phase-contrast x-ray computed tomography (PCCT) is currently investigated and developed as a potentially very interesting extension of conventional CT, because it promises to provide high soft-tissue contrast for weakly absorbing samples. For data acquisition several images at different grating positions are combined to obtain a phase-contrast projection. For short exposure times, which are necessary for lower radiation dose, the photon counts in a single stepping position are very low. In this case, the currently used phase-retrieval does not provide reliable results for some pixels. This uncertainty results in statistical phase wrapping, which leads to a higher standard deviation in the phase-contrast projections than theoretically expected. For even lower statistics, the phase retrieval breaks down completely and the phase information is lost. New measurement procedures rely on a linear approximation of the sinusoidal phase stepping curve around the zero crossings. In this case only two images are acquired to obtain the phase-contrast projection. The approximation is only valid for small phase values. However, typically nearly all pixels are within this regime due to the differential nature of the signal. We examine the statistical properties of a linear approximation method and illustrate by simulation and experiment that the lower statistical limit can be redefined using this method. That means that the phase signal can be retrieved even with very low photon counts and statistical phase wrapping can be avoided. This is an important step towards enhanced image quality in PCCT with very low photon counts.

There are two basic sources of uncertainty in cancer chemotherapy: how much of the therapeutic agent reaches the cancer cells, and how effective it is in reducing or controlling the tumor when it gets there. There is also a concern about adverse effects of the therapy drug. Similarly in external-beam radiation therapy or radionuclide therapy, there are two sources of uncertainty: delivery and efficacy of the radiation absorbed dose, and again there is a concern about radiation damage to normal tissues. The therapy operating characteristic (TOC) curve, developed in the context of radiation therapy, is a plot of the probability of tumor control vs. the probability of normal-tissue complications as the overall radiation dose level is varied, e.g. by varying the beam current in external-beam radiotherapy or the total injected activity in radionuclide therapy. The TOC can be applied to chemotherapy with the administered drug dosage as the variable. The area under a TOC curve (AUTOC) can be used as a figure of merit for therapeutic efficacy, analogous to the area under an ROC curve (AUROC), which is a figure of merit for diagnostic efficacy. In radiation therapy AUTOC can be computed for a single patient by using image data along with radiobiological models for tumor response and adverse side effects. In this paper we discuss the potential of using mathematical models of drug delivery and tumor response with imaging data to estimate AUTOC for chemotherapy, again for a single patient. This approach provides a basis for truly personalized therapy and for rigorously assessing and optimizing the therapy regimen for the particular patient. A key role is played by Emission Computed Tomography (PET or SPECT) of radiolabeled chemotherapy drugs.

The reconstruction of CT images with low noise and highest spatial resolution is a challenging task. Usually, a trade-off between at least these two demands has to be found or several reconstructions with mutually exclusive properties, i.e. either low noise or high spatial resolution, have to be performed. Iterative reconstruction methods might be suitable tools to overcome these limitations and provide images of highest diagnostic quality with formerly mutually exclusive image properties. While image quality metrics like the modulation transfer function (MTF) or the point spread function (PSF) are well-defined in case of standard reconstructions, e.g. filtered backprojection, the iterative algorithms lack these metrics. To overcome this issue alternate methodologies like the model observers have been proposed recently to allow a quantification of a usually task-dependent image quality metric.¹ As an alternative we recently proposed an iterative reconstruction method, the alpha-image reconstruction (AIR), providing well-defined image quality metrics on a per-voxel basis.² In particular, the AIR algorithm seeks to find weighting images, the alpha-images, that are used to blend between basis images with mutually exclusive image properties. The result is an image with highest diagnostic quality that provides a high spatial resolution and a low noise level. As the estimation of the alpha-images is computationally demanding we herein aim at optimizing this process and highlight the favorable properties of AIR using patient measurements.

The simulation of magnetic resonance (MR) images plays an important role in the validation of image analysis algorithms such as image segmentation, due to lack of sufficient ground truth in real MR images. Previous work on MRI simulation has focused on explicitly modeling the MR image formation process. However, because of the overwhelming complexity of MR acquisition these simulations must involve simplifications and approximations that can result in visually unrealistic simulated images. In this work, we describe an example-based simulation framework, which uses an “atlas” consisting of an MR image and its anatomical models derived from the hard segmentation. The relationships between the MR image intensities and its anatomical models are learned using a patch-based regression that implicitly models the physics of the MR image formation. Given the anatomical models of a new brain, a new MR image can be simulated using the learned regression. This approach has been extended to also simulate intensity inhomogeneity artifacts based on the statistical model of training data. Results show that the example based MRI simulation method is capable of simulating different image contrasts and is robust to different choices of atlas. The simulated images resemble real MR images more than simulations produced by a physics-based model.

Motion blurring is still a challenge for cardiac CT imaging. A new motion estimation (ME) and motion compensation method is developed for cardiac CT. The proposed method estimates motion of entire heart, and then applies motion compensation. Therefore, the proposed method reduces motion artifacts not only in coronary artery region as most other methods did, but also reduces motion blurring in myocardium region. In motion compensated reconstruction, we use the Fourier transfer method proposed by Pack et al to obtain a series of partial images, and then warp and sum together to obtain final motion compensated images. The robustness and performance of the proposed method was verified with data from 10 patients and improvements in sharpness of both coronary arteries and myocardium were obtained.

Recently a new class of imaging systems, referred to as photon-processing (PP) systems, are being developed that uses real-time maximum-likelihood (ML) methods to estimate multiple attributes per detected photon and store these attributes in a list format. PP systems could have a number of potential advantages compared to systems that bin photons based on attributes such as energy, projection angle, and position, referred to as photon-counting (PC) systems. For example, PP systems do not suffer from binning-related information loss and provide the potential to extract information from attributes such as energy deposited by the detected photon. To quantify the effects of this advantage on task performance, objective evaluation studies are required. We performed this study in the context of quantitative 2-dimensional single-photon emission computed tomography (SPECT) imaging with the end task of estimating the mean activity concentration within a region of interest (ROI). We first theoretically outline the effect of null space on estimating the mean activity concentration, and argue that due to this effect, PP systems could have better estimation performance compared to PC systems with noise-free data. To evaluate the performance of PP and PC systems with noisy data, we developed a singular value decomposition (SVD)-based analytic method to estimate the activity concentration from PP systems. Using simulations, we studied the accuracy and precision of this technique in estimating the activity concentration. We used this framework to objectively compare PP and PC systems on the activity concentration estimation task. We investigated the effects of varying the size of the ROI and varying the number of bins for the attribute corresponding to the angular orientation of the detector in a continuously rotating SPECT system. The results indicate that in several cases, PP systems offer improved estimation performance compared to PC systems.

Scanning-Beam Digital X-ray (SBDX) is a technology for low-dose fluoroscopy that employs inverse geometry x-ray beam scanning. To assist with rapid modeling of inverse geometry x-ray systems, we have developed a Monte Carlo (MC) simulation tool based on the MC-GPU framework. MC-GPU version 1.3 was modified to implement a 2D array of focal spot positions on a plane, with individually adjustable x-ray outputs, each producing a narrow x-ray beam directed toward a stationary photon-counting detector array. Geometric accuracy and blurring behavior in tomosynthesis reconstructions were evaluated from simulated images of a 3D arrangement of spheres. The artifact spread function from simulation agreed with experiment to within 1.6% (rRMSD). Detected x-ray scatter fraction was simulated for two SBDX detector geometries and compared to experiments. For the current SBDX prototype (10.6 cm wide by 5.3 cm tall detector), x-ray scatter fraction measured 2.8-6.4% (18.6-31.5 cm acrylic, 100 kV), versus 2.2-5.0% in MC simulation. Experimental trends in scatter versus detector size and phantom thickness were observed in simulation. For dose evaluation, an anthropomorphic phantom was imaged using regular and regional adaptive exposure (RAE) scanning. The reduction in kerma-area-product resulting from RAE scanning was 45% in radiochromic film measurements, versus 46% in simulation. The integral kerma calculated from TLD measurement points within the phantom was 57% lower when using RAE, versus 61% lower in simulation. This MC tool may be used to estimate tomographic blur, detected scatter, and dose distributions when developing inverse geometry x-ray systems.

Photon-counting CT (PCCT) potentially offers both improved dose efficiency and material decomposition capabilities relative to CT systems using energy integrating detectors. With respect to material decomposition, both projection-based and image-based methods have been proposed, most of which require accurate a priori information regarding the shape of the x-ray spectra and the response of the detectors. Additionally, projection-based methods require access to projection data. These data can be difficult to obtain, since spectra, detector response, and projection data formats are proprietary information. Further, some published image-based, 3-material decomposition methods require a volume conservation assumption, which is often violated in solutions. We have developed an image-based material decomposition method that can overcome those limitations. We introduced a general condition on volume constraint that does not require the volume to be conserved in a mixture. An empirical calibration can be performed with various concentrations of basis materials. The material decomposition method was applied to images acquired from a prototype whole-body PCCT scanner. The results showed good agreement between the estimation and known mass concentration values. Factors affecting the performance of material decomposition, such as energy threshold configuration and volume conservation constraint, were also investigated. Changes in accuracy of the mass concentration estimates were demonstrated for four different energy configurations and when volume conservation was assumed.

Purpose: The multi-leaf collimator (MLC) assembly present on TomoTherapy (Accuray, Madison WI) radiation therapy (RT) and mega voltage CT machines is well suited to perform fluence field modulated CT (FFMCT). In addition, there is a demand in the RT environment for FFMCT imaging techniques, specifically volume of interest (VOI) imaging. Methods: A clinical TomoTherapy machine was programmed to deliver 30% imaging dose outside predefined VOIs. Four different size ROIs were placed at varying distances from isocenter. Projections intersecting the VOI received "full dose" while those not intersecting the VOI received 30% of the dose (i.e. the incident fluence for non VOI projections was 30% of the incident fluence for projections intersecting the VOI). Additional scans without fluence field modulation were acquired at "full" and 30% dose. The noise (pixel standard deviation) was measured inside the VOI region and compared between the three scans. Results: The VOI-FFMCT technique produced an image noise 1.09, 1.05, 1.05, and 1.21 times higher than the "full dose" scan for ROI sizes of 10 cm, 13 cm, 10 cm, and 6 cm respectively within the VOI region. Conclusions: Noise levels can be almost unchanged within clinically relevant VOIs sizes for RT applications while the integral imaging dose to the patient can be decreased, and/or the image quality in RT can be dramatically increased with no change in dose relative to non-FFMCT RT imaging. The ability to shift dose away from regions unimportant for clinical evaluation in order to improve image quality or reduce imaging dose has been demonstrated. This paper demonstrates that FFMCT can be performed using the MLC on a clinical TomoTherapy machine for the first time.

Purpose: Arthritis and bone trauma are often accompanied by bone marrow edema (BME). BME is challenging to detect in CT due to the overlaying trabecular structure but can be visualized using dual-energy (DE) techniques to discriminate water and fat. We investigate the feasibility of DE imaging of BME on a dedicated flat-panel detector (FPD) extremities cone-beam CT (CBCT) with a unique x-ray tube with three longitudinally mounted sources. Methods: Simulations involved a digital BME knee phantom imaged with a 60 kVp low-energy beam (LE) and 105 kVp high-energy beam (HE) (+0.25 mm Ag filter). Experiments were also performed on a test-bench with a Varian 4030CB FPD using the same beam energies as the simulation study. A three-source configuration was implemented with x-ray sources distributed along the longitudinal axis and DE CBCT acquisition in which the superior and inferior sources operate at HE (and collect half of the projection angles each) and the central source operates at LE. Three-source DE CBCT was compared to a double-scan, single-source orbit. Experiments were performed with a wrist phantom containing a 50 mg/ml densitometry insert submerged in alcohol (simulating fat) with drilled trabeculae down to ~1 mm to emulate the trabecular matrix. Reconstruction-based three-material decomposition of fat, soft tissue, and bone was performed. Results: For a low-dose scan (36 mAs in the HE and LE data), DE CBCT achieved combined accuracy of ~0.80 for a pattern of BME spherical lesions ranging 2.5 - 10 mm diameter in the knee phantom. The accuracy increased to ~0.90 for a 360 mAs scan. Excellent DE discrimination of the base materials was achieved in the experiments. Approximately 80% of the alcohol (fat) voxels in the trabecular phantom was properly identified both for single and 3-source acquisitions, indicating the ability to detect edemous tissue (water-equivalent plastic in the body of the densitometry insert) from the fat inside the trabecular matrix (emulating normal trabecular bone with significant fraction of yellow marrow). Conclusion: Detection of BME and quantification of water and fat content were achieved in extremities DE CBCT with a longitudinal configuration of sources providing DE imaging in a single gantry rotation. The findings support the development of DE imaging capability for CBCT of the extremities in areas conventionally in the domain of MRI.

X-ray computed tomography (CT) with energy-discriminating capabilities presents exciting opportunities for increased dose efficiency and improved material decomposition analyses. However, due to constraints imposed by the inability of photon-counting detectors (PCD) to respond accurately at high photon flux, to date there has been no clinical application of PCD-CT. Recently, our lab installed a research prototype system consisting of two x-ray sources and two corresponding detectors, one using an energy-integrating detector (EID) and the other using a PCD. In this work, we report the first third-party evaluation of this prototype CT system using both phantoms and a cadaver head. The phantom studies demonstrated several promising characteristics of the PCD sub-system, including improved longitudinal spatial resolution and reduced beam hardening artifacts, relative to the EID sub-system. More importantly, we found that the PCD sub-system offers excellent pulse pileup control in cases of x-ray flux up to 550 mA at 140 kV, which corresponds to approximately 2.5×10¹¹ photons per cm² per second. In an anthropomorphic phantom and a cadaver head, the PCD sub-system provided image quality comparable to the EID sub-system for the same dose level. Our results demonstrate the potential of the prototype system to produce clinically-acceptable images in vivo.

Fluence field modulated CT allows for improvements in image quality and dose reduction. To date, only 1-D modulators have been proposed, the extension to 2-D modulation is difficult with solid-metal attenuation-based modulators. This work proposes to use liquids and gas to attenuate the x-ray beam which can be arrayed allowing for 2-D fluence modulation. The thickness of liquid and the pressure for a given path length of gas were determined that provided the same attenuation as 30 cm of soft tissue at 80, 100, 120, and 140 kV. Gaseous Xenon and liquid Iodine, Zinc Chloride, and Cerium Chloride were studied. Additionally, we performed some proof-of-concept experiments in which (1) a single cell of liquid was connected to a reservoir which allowed the liquid thickness to be modulated and (2) a 96 cell array was constructed in which the liquid thickness in each cell was adjusted manually. Liquid thickness varied as a function of kV and chemical composition, with Zinc Chloride allowing for the smallest thickness; 1.8, 2.25, 3, and 3.6 cm compensated for 30 cm of soft tissue at 80, 100, 120, and 140 kV respectively. The 96 cell Iodine attenuator allowed for a reduction in both dynamic range to the detector and scatter to primary ratio. Successful modulation of a single cell was performed at 0, 90, and 130 degrees using a simple piston/actuator. The thickness of liquids and the Xenon gas pressure seem logistically implementable within the constraints of CBCT and diagnostic CT systems.

A new scalable CT system architecture is introduced with the potential to achieve much higher temporal resolution than is possible with current CT designs while maintaining the flux per rotation near today’s levels. Higher effective rotation speeds can be achieved leveraging today’s x-ray tube designs and capabilities. The new CT architecture comprises the following elements: (1) decoupling of the source rotation from the detector rotation through the provision of two independent, coaxial and coplanar rotating gantries (drums); (2) observation of a source at a range of azimuthal angles with respect to a given detector cell; (3) utilization of a multiplicity of x-ray sources; (4) use of a wide-angle iso-centered detector mounted on the independent detector drum; (5) the detector drum presents a wide angular aperture allowing x-rays from the various sources to pass through, with the active detector cells occupying about 240-degrees in one configuration, and the wide aperture the complementary 120-degrees; (6) anti-scatter grids with absorbing lamellas oriented substantially parallel to the main gantry plane; (7) optional sparse view acquisition in “bunches,” a unique sparse sampling pattern potentially enabling further data acquisition speed-up for specific applications. Temporal resolution gains are achieved when multiple sources are simultaneously in view of the extended detector. Accurate data acquisition then relies on multiplexing in space, time, or spectra. Thus the use of an energy-discriminating detector, such as a photon-counting detector, and of tube pulsing will be advantageous. Volume-based scatter correction methods have the potential to apply when space multiplexing is used.

Purpose: Photon counting x-ray detectors (PCXDs) are an important emerging technology for spectral imaging and material differentiation with numerous potential applications in diagnostic imaging. We report development of a Si-strip PCXD system originally developed for mammography with potential application to spectral CT of musculoskeletal extremities, including challenges associated with sparse sampling, spectral calibration, and optimization for higher energy x-ray beams. Methods: A bench-top CT system was developed incorporating a Si-strip PCXD, fixed anode x-ray source, and rotational and translational motions to execute complex acquisition trajectories. Trajectories involving rotation and translation combined with iterative reconstruction were investigated, including single and multiple axial scans and longitudinal helical scans. The system was calibrated to provide accurate spectral separation in dual-energy three-material decomposition of soft-tissue, bone, and iodine. Image quality and decomposition accuracy were assessed in experiments using a phantom with pairs of bone and iodine inserts (3, 5, 15 and 20 mm) and an anthropomorphic wrist. Results: The designed trajectories improved the sampling distribution from 56% minimum sampling of voxels to 75%. Use of iterative reconstruction (viz., penalized likelihood with edge preserving regularization) in combination with such trajectories resulted in a very low level of artifacts in images of the wrist. For large bone or iodine inserts (>5 mm diameter), the error in the estimated material concentration was <16% for (50 mg/mL) bone and <8% for (5 mg/mL) iodine with strong regularization. For smaller inserts, errors of 20-40% were observed and motivate improved methods for spectral calibration and optimization of the edge-preserving regularizer. Conclusion: Use of PCXDs for three-material decomposition in joint imaging proved feasible through a combination of rotation-translation acquisition trajectories and iterative reconstruction with optimized regularization.

Energy-discriminating, photon counting (EDPC) detectors have been proposed for CT systems for their spectral imaging capabilities, improved dose efficiency and higher spatial resolution. However, these advantages disappear at high flux because of the damaging effects of pulse pileup. From an information theoretic standpoint, spectral information is lost. The information loss is particularly high when we assume that the EDPC detector extracts information using a bank of comparators, as current EDPC detectors do. We analyze the use of alternative pulse detection logic which could preserve information in the presence of pileup. For example, the peak-only detector counts only a single event at the peak energy of multiple pulses which are piled up. We describe and evaluate five of these alternatives in simulation by numerically estimating the Cramer-Rao lower bound of the variance. At high flux, alternative mechanisms outperform comparators. In spectral imaging tasks, the variance reduction can be as high as an order of magnitude.

Recently, photon counting detectors capable of extracting spectral information have received much attention in CT, with promise of using spectral information to construct material basis images, correct beam-hardening artifacts, or provide improved imaging of K-edge contrast agents.^{1, 2} In this work, we focus on the goal of constructing images of basis material maps, and investigate the feasibility of analytically computing pixel variance maps for these images, so that alternative data acquisition and reconstruction methods can be compared and evaluated with respect to their noise properties. Our approach is based on linearization of the basis material decomposition and reconstruction operations, and we therefore demonstrate the method using the ubiquitous filtered back-projection algorithm, which is linear. We then performed preliminary investigation of the method by comparing basis material variance maps for two data acquisition methods that were previously found to have different noise properties:³ two-sided bin measurements acquired from separate, independent data realizations and two-sided bin measurements acquired from a single data realization.

Spectral imaging with photon counting detectors has recently attracted a lot of interest in X-ray and CT imaging due to its potential to enable ultra low radiation dose x-ray imaging. However, when radiation exposure level is low, quantum noise may be prohibitively high to hinder applications. Therefore, it is desirable to develop new methods to reduce quantum noise in the acquired data from photon counting detectors. In this paper, we propose a new denoising algorithm to reduce quantum noise in data acquired using an ideal photon counting detector. The proposed method exploits the intrinsic low dimensionality of acquired spectral data to decompose the acquired data in a series of orthonormal spectral bases. The first few spectral bases contain object information while the rest of the bases contain primarily quantum noise. The separation of image content and noise in these orthogonal spatial bases provides a means to reject noise without losing image content. Numerical simulations were conducted to validate and evaluate the proposed noise reduction algorithm. The results demonstrated that the proposed method can effectively reduce quantum noise while maintaining both spatial and spectral fidelity.

Purpose: In-Depth photon counting detectors (PCXD) use an edge-on configuration and have multi-layer segmentations. The benefit of this configuration for additional spectral information depends on the energy response. Also, inter-layer cross-talk introduces correlation to the signal collected from each layer, which makes the independent Poisson model no longer valid for estimating the Cramér-Rao lower bound (CRLB) of the material decomposition variance. We proposed to use a multivariate Gaussian model as the substitute address the data correlation. Methods: A 120 kVp incident spectrum was simulated and transmitted through 25cm of water and 1cm of calcium. 5- layer In-Depth and 1-layer Edge-On PCXDs with full energy resolution were simulated using Monte Carlo methods. We selected Si, GaAs and CdTe as detector materials. The detectors were defined to have 1mm wide pixels and thickness of 70mm (Si), 10.5mm (GaAs) and 3mm (CdTe). Geant4 was used and energy response functions (ERF) capturing secondary events were obtained, together with the Gaussian parameter estimates. We evaluated the CRLBs of the In-Depth and Edge-On detectors for each material and the systematic variance bounds were compared. Results: For uncorrelated data, the CRLB can assume Poisson statistics. As the data becomes more correlated, the Poisson CRLB fails to capture the cross-talk effect, but a Gaussian model can, and is accurate if the number of photons is not small. The CRLB analysis shows that the effects of the ERF and the noise correlation are significant. If cross-talk can be corrected, the depth information proves to be beneficial and can reduce the variance lower bound by 3% to 10% depending on the detector material. Conclusions: The multivariate Gaussian model was validated to be a good substitute to the Poisson model for PCXD CRLB estimation. It can avoid the errors that would otherwise be caused by correlated measurements.

Photon counting spectral detectors (PCSD) with smaller pixels and efficient sensors are desirable in applications like material decomposition and phase contrast x-ray imaging where discrimination of small signals and fine structure may be desired. Charge sharing in PCSD increases with decreasing pixel sizes and increasing sensor thickness such that the energy calibration or utility of spectral information can become a major hurdle. Utility of a combination of high Z sensors and small pixel sizes in PCSD is limited without efficient threshold calibration and charge sharing mitigation. Here we explore the utility of x-ray tube kVp as a reference to achieve efficient and fast calibration of PCSDs. This calibration method itself does not require rearranging the imaging setup and is not impacted by charge sharing. Our preliminary results indicate that this method can be useful even in scenarios where metal fluorescence and radioactive source based calibration techniques may be practically impossible. Our results are validated using x-ray fluorescence based calibration for a Silicon detector with moderate charge sharing. Calibration of a particularly challenging case of a Medipix2 detector (55 μm pixel size) with a 1 mm thick CdTe sensor and a Medipix3 detector with CdTe sensor is also demonstrated. A cross validation with K-edge identification of Gd is also presented here.

Variations among detector channels in CT very sensitively lead to ring artefacts in the reconstructed images. For material decomposition in the projection domain, the variations can result in intolerable biases in the material line integral estimates. A typical way to overcome these effects is to apply calibration methods that try to unify spectral responses from different detector channels to an ideal response from a detector model. However, the calibration procedure can be rather complex and require excessive calibration measurements for a multitude of combinations of x-ray shapes, tissue combinations and thicknesses. In this paper, we propose a channel-wise model for a multibin photon-counting detector for spectral CT. Predictions of this channel-wise model match well with their physical performances, which can thus be used to eliminate ring artefacts in CT images and achieve projection-basis material decomposition. In an experimental validation, image data show significant improvement with respect to ring artefacts compared to images calibrated with flat-fielding data. Projection-based material decomposition gives basis material images showing good separation among individual materials and good quantification of iodine and gadolinium contrast agents. The work indicates that the channel-wise model can be used for quantitative CT with this detector.

The ability to assay tumor biologic features and the impact of drugs on tumor biology is fundamental to drug development. Advances in our ability to measure genomics, gene expression, protein expression, and cellular biology have led to a host of new targets for anticancer drug therapy. In translating new drugs into clinical trials and clinical practice, these same assays serve to identify patients most likely to benefit from specific anticancer treatments. As cancer therapy becomes more individualized and targeted, there is an increasing need to characterize tumors and identify therapeutic targets to select therapy most likely to be successful in treating the individual patient’s cancer. Thus far assays to identify cancer therapeutic targets or anticancer drug pharmacodynamics have been based upon in vitro assay of tissue or blood samples. Advances in molecular imaging, particularly PET, have led to the ability to perform quantitative non-invasive molecular assays. Imaging has traditionally relied on structural and anatomic features to detect cancer and determine its extent. More recently, imaging has expanded to include the ability to image regional biochemistry and molecular biology, often termed molecular imaging. Molecular imaging can be considered an in vivo assay technique, capable of measuring regional tumor biology without perturbing it. This makes molecular imaging a unique tool for cancer drug development, complementary to traditional assay methods, and a potentially powerful method for guiding targeted therapy in clinical trials and clinical practice. The ability to quantify, in absolute measures, regional in vivo biologic parameters strongly supports the use of molecular imaging as a tool to guide therapy. This review summarizes current and future applications of quantitative molecular imaging as a biomarker for cancer therapy, including the use of imaging to (1) identify patients whose tumors express a specific therapeutic target; (2) determine whether the drug reaches the target; (3) identify an early response to treatment; and (4) predict the impact of therapy on long-term outcomes such as survival. The manuscript reviews basic concepts important in the application of molecular imaging to cancer drug therapy, in general, and will discuss specific examples of studies in humans, and highlight future directions, including ongoing multi-center clinical trials using molecular imaging as a cancer biomarker.

We propose a new method for PET/MR respiratory motion compensation, which is based on a 3D-2D registration of strongly undersampled MR data and a) runs in parallel with the PET acquisition, b) can be interlaced with clinical MR sequences, and c) requires less than one minute of the total MR acquisition time per bed position. In our simulation study, we applied a 3D encoded radial stack-of-stars sampling scheme with 160 radial spokes per slice and an acquisition time of 38 s. Gated 4D MR images were reconstructed using a 4D iterative reconstruction algorithm. Based on these images, motion vector fields were estimated using our newly-developed 3D-2D registration framework. A 4D PET volume of a patient with eight hot lesions in the lungs and upper abdomen was simulated and MoCo 4D PET images were reconstructed based on the motion vector fields derived from MR. For evaluation, average SUV_mean values of the artificial lesions were determined for a 3D, a gated 4D, a MoCo 4D and a reference (with ten-fold measurement time) gated 4D reconstruction. Compared to the reference, 3D reconstructions yielded an underestimation of SUV_mean values due to motion blurring. In contrast, gated 4D reconstructions showed the highest variation of SUV_mean due to low statistics. MoCo 4D reconstructions were only slightly affected by these two sources of uncertainty resulting in a significant visual and quantitative improvement in terms of SUV_mean values. Whereas temporal resolution was comparable to the gated 4D images, signal-to-noise ratio and contrast-to-noise ratio were close to the 3D reconstructions.

Cardiovascular disease is currently the leading cause of mortality worldwide. Digital subtraction angiography (DSA) is widely used to enhance the visibility of small vessels and vasculature obscurred by overlying bone and lung fields by subtracting a mask and contrast image. However, motion between these mask and contrast images can introduce artifacts that can render a study non-diagnostic. This makes DSA particularly unsuccessful for cardiac imaging. A method called dual-energy, or energy subtraction angiography (ESA), was proposed in the past as an alternative for vascular imaging, however it was not pursued because experimental results suggested that image quality was deemed as poor and inferior to DSA. Image quality for angiography comes down to iodine signal and noise. In this paper we investigate the fundamental iodine signal and noise analysis of ESA and compare it to DSA. Method: We developed a polyenergetic and monoenergetic theoretical model for iodine signal and noise for both ESA and DSA. We validated our polyenergetic model by experiment where ESA and DSA images of a vascular phantom were acquired using an x-ray system with a flat panel CsI Xmaru1215CF-MPTM (Rayence Co., Ltd., Republic of Korea) detector. For ESA low and high applied tube voltages of 50 kV and 120 kV (2.5 mmCu), respectively, and for DSA the applied tube voltage was 80 kV. Iodine signal-to-noise ratio (SNR) per entrance exposure was calculated for each iodine concentration for both ESA and DSA. Results: Our measured iodine SNR agreed well with theoretical calculations. Iodine SNR for ESA was relatively higher than DSA for low iodine mass loadings, and as iodine mass loading increases iodine SNR decreases. Conclusions: We have developed a model for iodine SNR for both DSA and ESA. Our model was validated with experiment and showed excellent agreement. We have shown that there is potential for obtaining iodine-specific images using ESA that are similar to DSA.

Purpose: Common methods for assessing image quality of digital breast tomosynthesis (DBT) devices currently utilize simplified or otherwise unrealistic phantoms, which use inserts in a uniform background and gauge performance based on a subjective evaluation of insert visibility. This study proposes a different methodology to assess system performance using a three-dimensional clinically-informed anthropomorphic breast phantom. Methods: The system performance is assessed by imaging the phantom and computationally characterizing the resultant images in terms of several new metrics. These include a contrast index (reflective of local difference between adipose and glandular material), a contrast to noise ratio index (reflective of contrast against local background noise), and a nonuniformity index (reflective of contributions of noise and artifacts within uniform adipose regions). Indices were measured at ROI sizes of 10mm and 37 mm, respectively. The method was evaluated at fixed dose of 1.5 mGy AGD. Results: Results indicated notable differences between systems. At 10 mm, vendor A had the highest contrast index, followed by B and C in that. The performance ranking was identical at the largest ROI size. The non-uniformity index similarly exhibited system-dependencies correlated with visual appearance of clutter from out-of-plane artifacts. Vendor A had the greatest NI at all ROI sizes, B had the second greatest, and C the least. Conclusions: The findings illustrate that the anthropomorphic phantom can be used as a quality control tool with results that are targeted to be more reflective of clinical performance of breast tomosynthesis systems of multiple manufacturers.

Static C-Arm CT 3D FDK baseline reconstructions (3D-DSA) are unable to provide temporal information to radiologists. 4D-DSA provides a time series of 3D volumes implementing a constrained image, thresholded 3D-DSA, reconstruction utilizing temporal dynamics in the 2D projections. Volumetric limiting spatial resolution (VLSR) of 4DDSA is quantified and compared to a 3D-DSA reconstruction using the same 3D-DSA parameters. Investigated were the effects of varying over significant ranges the 4D-DSA parameters of 2D blurring kernel size applied to the projection and threshold applied to the 3D-DSA when generating the constraining image of a scanned phantom (SPH) and an electronic phantom (EPH). The SPH consisted of a 76 micron tungsten wire encased in a 47 mm O.D. plastic radially concentric thin walled support structure. An 8-second/248-frame/198^° scan protocol acquired the raw projection data. VLSR was determined from averaged MTF curves generated from each 2D transverse slice of every (248) 4D temporal frame (3D). 4D results for SPH and EPH were compared to the 3D-DSA. Analysis of the 3D-DSA resulted in a VLSR of 2.28 and 1.69 lp/mm for the EPH and SPH respectively. Kernel (2D) sizes of either 10x10 or 20x20 pixels with a threshold of 10% of the 3D-DSA as a constraining image provided 4D-DSA VLSR nearest to the 3D-DSA. 4D-DSA algorithms yielded 2.21 and 1.67 lp/mm with a percent error of 3.1% and 1.2% for the EPH and SPH respectively as compared to the 3D-DSA. This research indicates 4D-DSA is capable of retaining the resolution of the 3D-DSA.

A family of imaging task-specific metrics designated Relative Object Detectability (ROD) metrics was developed to enable objective, quantitative comparisons of different x-ray systems. Previously, ROD was defined as the integral over spatial frequencies of the Fourier Transform of the object function, weighted by the detector DQE for one detector, divided by the comparable integral for another detector. When effects of scatter and focal spot unsharpness are included, the generalized metric, GDQE, is substituted for the DQE, resulting in the G-ROD metric. The G-ROD was calculated for two different detectors with two focal spot sizes using various-sized simulated objects to quantify the improved performance of new high-resolution CMOS detector systems. When a measured image is used as the object, a Generalized Measured Relative Object Detectability (GM-ROD) value can be generated. A neuro-vascular stent (Wingspan) was imaged with the high-resolution Micro-Angiographic Fluoroscope (MAF) and a standard flat panel detector (FPD) for comparison using the GM-ROD calculation. As the lower integration bound increased from 0 toward the detector Nyquist frequency, increasingly superior performance of the MAF was evidenced. Another new metric, the R-ROD, enables comparing detectors to a reference detector of given imaging ability. R-RODs for the MAF, a new CMOS detector and an FPD will be presented. The ROD family of metrics can provide quantitative more understandable comparisons for different systems where the detector, focal spot, scatter, object, techniques or dose are varied and can be used to optimize system selection for given imaging tasks.

Statistical iterative reconstruction (IR) techniques have demonstrated many advantages in X-ray CT reconstruction. The statistical iterative reconstruction approach is often modeled as an optimization problem including a data fitting function and a penalty function. The tuning parameter values that regulate the strength of the penalty function are critical for achieving good reconstruction results. However, appropriate tuning parameter values that are suitable for the scan protocols and imaging tasks are often difficult to choose. In this work, we propose a path seeking algorithm that is capable of generating a series of IR images with different strengths of the penalty function. The path seeking algorithm uses the ratio of the gradients of the data fitting function and the penalty function to select pixels for small fixed size updates. We describe the path seeking algorithm for penalized weighted least squares (PWLS) with a Huber penalty function in both the directions of increasing and decreasing tuning parameter value. Simulations using the XCAT phantom show the proposed method produces path images that are very similar to the IR images that are computed via direct optimization. The root-mean- squared-error of one path image generated by the proposed method relative to full iterative reconstruction is about 6 HU for the entire image and 10 HU for a small region. Different path seeking directions, increment sizes and updating percentages of the path seeking algorithm are compared in simulations. The proposed method may reduce the dependence on selection of good tuning parameter values by instead generating multiple IR images, without significantly increasing the computational load.

To date, cardiovascular surgery enables the treatment of a wide range of aortic pathologies. One of the current challenges in this field is given by the detection of high-risk patients for adverse aortic events, who should be treated electively. Reliable diagnostic parameters, which indicate the urge of treatment, have to be determined. Functional imaging by means of 4D phase contrast-magnetic resonance imaging (PC-MRI) enables the time-resolved measurement of blood flow velocity in 3D. Applied to aortic phantoms, three dimensional blood flow properties and their relation to adverse dynamics can be investigated in vitro. Emerging ”in silico” methods of numerical simulation can supplement these measurements in computing additional information on crucial parameters. We propose a framework that complements 4D PC-MRI imaging by means of numerical simulation based on the Finite Element Method (FEM). The framework is developed on the basis of a prototypic aortic phantom and validated by 4D PC-MRI measurements of the phantom. Based on physical principles of biomechanics, the derived simulation depicts aortic blood flow properties and characteristics. The framework might help identifying factors that induce aortic pathologies such as aortic dilatation or aortic dissection. Alarming thresholds of parameters such as wall shear stress distribution can be evaluated. The combined techniques of 4D PC-MRI and numerical simulation can be used as complementary tools for risk-stratification of aortic pathology.

A fast and accurate scatter imaging technique to differentiate cancerous and healthy breast tissue is introduced in this work. Such a technique would have wide-ranging clinical applications from intra-operative margin assessment to breast cancer screening. Coherent Scatter Computed Tomography (CSCT) has been shown to differentiate cancerous from healthy tissue, but the need to raster scan a pencil beam at a series of angles and slices in order to reconstruct 3D images makes it prohibitively time consuming. In this work we apply the coded aperture coherent scatter spectral imaging technique to reconstruct 3D images of breast tissue samples from experimental data taken without the rotation usually required in CSCT. We present our experimental implementation of coded aperture scatter imaging, the reconstructed images of the breast tissue samples and segmentations of the 3D images in order to identify the cancerous and healthy tissue inside of the samples. We find that coded aperture scatter imaging is able to reconstruct images of the samples and identify the distribution of cancerous and healthy tissues (i.e., fibroglandular, adipose, or a mix of the two) inside of them. Coded aperture scatter imaging has the potential to provide scatter images that automatically differentiate cancerous and healthy tissue inside of ex vivo samples within a time on the order of a minute.

The objective of this work was to evaluate the advantages and shortcomings associated with Coherent Scatter Computed Tomography (CSCT) systems for breast imaging and study possible alternative configurations. The relationship between dose in a breast phantom and a simple surrogate of image quality in pencil-beam and fan-beam CSCT geometries was evaluated via Monte Carlo simulation, and an improved pencil-beam setup was proposed for faster CSCT data acquisition. CSCT projection datasets of a simple breast phantom have been simulated using a new version of the MC-GPU code that includes an improved model of x-ray coherent scattering using experimentally measured molecular interference functions. The breast phantom was composed of an 8 cm diameter cylinder of 50/50 glandular/adipose material and nine rods with different diameters of cancerous, adipose and glandular tissues. The system performance has been assessed in terms of the contrast-to-noise ratio (CNR) in multiple regions of interest within the reconstructed images, for a range of exposure levels. The enhanced pencil-beam setup consisted of multiplexed pencil beams and specific post-processing of the projection data to calculate the scatter intensity coming from each beam separately. At reconstruction spatial resolution of 1×1×1 mm³ and from 1 to 10 mGy of received breast dose, fan-beam geometry showed higher statistical noise and lower CNR than pencil-beam geometry. Conventional CT acquisition had the highest CNR per dose. However, the CNR figure of merit did not combine yet all the information available at different scattering angles in the CSCT, which has potential for increased discrimination of materials with similar attenuation properties. Preliminary evaluation of the multiplexed pencil-beam geometry showed that the scattering profiles simulated with the new approach are similar to those of the single pencil-beam geometry. Conclusion: It has been shown that the GPU-accelerated MC-GPU code is a practical tool to simulate complete CSCT scans with different acquisition geometries and exposure levels. The simulation showed better performance in terms of the received dose and CNR with pencil-beam geometry in comparison to the fan-beam geometry. Finally, we demonstrated that the proposed multiplexed-beam geometry might be useful for faster acquisition of CSCT while providing comparable image quality as the pencil-beam geometry.

Microcalcifications are one feature of interest in mammography and breast tomosynthesis (BT). To achieve optimal conditions for detection of microcalcifications in BT imaging, different acquisition geometries should be evaluated. The purpose of this work was to investigate the influence of acquisition schemes with different angular ranges, projection distributions and dose distributions on the visibility of microcalcifications in reconstructed BT volumes.

Microcalcifications were inserted randomly in a high resolution software phantom and a simulation procedure was used to model a MAMMOMAT Inspiration BT system. The simulation procedure was based on analytical ray tracing to produce primary images, Monte Carlo to simulate scatter contributions and flatfield image acquisitions to model system characteristics. Image volumes were reconstructed using the novel method super-resolution reconstruction with statistical artifact reduction (SRSAR). For comparison purposes, the volume of the standard acquisition scheme (50^° angular range and uniform projection and dose distribution) was also reconstructed using standard filtered backprojection (FBP).

To compare the visibility and depth resolution of the microcalcifications, signal difference to noise ratio (SDNR) and artifact spread function width (ASFW) were calculated. The acquisition schemes with very high central dose yielded significantly lower SDNR than the schemes with more uniform dose distributions. The ASFW was found to decrease (meaning an increase in depth resolution) with wider angular range. In conclusion, none of the evaluated acquisition schemes were found to yield higher SDNR or depth resolution for the simulated microcalcifications than the standard acquisition scheme.

Scattered radiation remains one of the primary challenges for digital mammography, resulting in decreased image contrast and visualization of key features. While anti-scatter grids are commonly used to reduce scattered radiation in digital mammography, they are an incomplete solution that can add radiation dose, cost, and complexity. Instead, a software-based scatter correction method utilizing asymmetric scatter kernels is developed and evaluated in this work, which improves upon conventional symmetric kernels by adapting to local variations in object thickness and attenuation that result from the heterogeneous nature of breast tissue. This fast adaptive scatter kernel superposition (fASKS) method was applied to mammography by generating scatter kernels specific to the object size, x-ray energy, and system geometry of the projection data. The method was first validated with Monte Carlo simulation of a statistically-defined digital breast phantom, which was followed by initial validation on phantom studies conducted on a clinical mammography system. Results from the Monte Carlo simulation demonstrate excellent agreement between the estimated and true scatter signal, resulting in accurate scatter correction and recovery of 87% of the image contrast originally lost to scatter. Additionally, the asymmetric kernel provided more accurate scatter correction than the conventional symmetric kernel, especially at the edge of the breast. Results from the phantom studies on a clinical system further validate the ability of the asymmetric kernel correction method to accurately subtract the scatter signal and improve image quality. In conclusion, software-based scatter correction for mammography is a promising alternative to hardware-based approaches such as anti-scatter grids.

In x-ray breast imaging, the anatomical noise background of the breast has a significant impact on the detection of lesions and other features of interest. This anatomical noise is typically characterized by a parameter, β, which describes a power law dependence of anatomical noise on spatial frequency (the shape of the anatomical noise power spectrum). Large values of β have been shown to reduce human detection performance, and in conventional mammography typical values of β are around 3.2. Recently, x-ray differential phase contrast (DPC) and the associated dark field imaging methods have received considerable attention as possible supplements to absorption imaging for breast cancer diagnosis. However, the impact of these additional contrast mechanisms on lesion detection is not yet well understood. In order to better understand the utility of these new methods, we measured the β indices for absorption, DPC, and dark field images in 15 cadaver breast specimens using a benchtop DPC imaging system. We found that the measured β value for absorption was consistent with the literature for mammographic acquisitions (β = 3.61±0.49), but that both DPC and dark field images had much lower values of β (β = 2.54±0.75 for DPC and β = 1.44±0.49 for dark field). In addition, visual inspection showed greatly reduced anatomical background in both DPC and dark field images. These promising results suggest that DPC and dark field imaging may help provide improved lesion detection in breast imaging, particularly for those patients with dense breasts, in whom anatomical noise is a major limiting factor in identifying malignancies.

A novel breast CT system capable of traversing non-traditional 3D trajectories was developed to address cone beam sampling insufficiency for pendant breast imaging. The purpose of this study was to characterize differences in three dimensional x-ray dose distributions in a target volume due to the acquisition trajectory. Three cylindrical phantoms of different diameters and an anthropomorphic breast phantom were scanned in a pendant geometry with two orbits- azimuthal orbit with no polar tilt, and a saddle orbit with ±15° contiguous polar tilts. The phantoms were initially filled with water and then with a 75:25% water: methanol mixture, to simulate different density breast tissues. Fully-3D CT scans were performed using a tungsten anode x-ray source. Ionization chamber calibrated radiochromic film was used to determine average dose delivered to the central sagittal slice of a volume, as well as to visualize the 2D dose distribution across the slice. Results indicated that the mean glandular dose for normal imaging exposures, measured at the central slice across different diameters ranged from 3.93-5.28 mGy, with the lowest average dose measured on the largest diameter cylinder. In all cases, the dose delivered by the saddle was consistently 1-3% lower than the no-tilt scans. These results corroborate previous cylinder Monte Carlo studies which showed a 1% reduction in saddle dose. The average dose measured in the breast phantom filled with 75:25 mixture was slightly higher for saddle. Non-traditional 3D breast CT scans have slightly better dose performance for equal image noise compared with simple, under sampled circular orbits.

By varying its thickness to compensate for the different path length through the patient as a function of fan angle, a pre-patient bowtie filter modulates flux distribution to reduce patient dose, scatter, and detector dynamic range, and to improve image quality. A dynamic bowtie filter is superior to its traditional, static counterpart in its ability to adjust its thickness along different fan and view angles to suit a specific patient and task. Among the proposed dynamic bowtie designs, the piecewise-linear and the digital beam attenuators offer more flexibility than conventional filters, but rely on analog positioning of a limited number of wedges. In this work, we introduce a new approach with digital control, called the fluid-filled dynamic bowtie filter. It is a two-dimensional array of small binary elements (channels filled or unfilled with attenuating liquid) in which the cumulative thickness along the x-ray path contributes to the bowtie’s total attenuation. Using simulated data from a pelvic scan, the performance is compared with the piecewise-linear attenuator. The fluid-filled design better matches the desired target attenuation profile and delivers a 4.2x reduction in dynamic range. The variance of the reconstruction (or noise map) can also be more homogeneous. In minimizing peak variance, the fluid-filled attenuator shows a 3% improvement. From the initial simulation results, the proposed design has more control over the flux distribution as a function of both fan and view angles.

Optimal selections of tube potential (kV) and tube current (mA) are essential in maximizing the diagnostic potential of a given CT technology while minimizing radiation dose. The use of a lower tube potential may improve image contrast, but may also require a significantly higher tube current to compensate for the rapid decrease of tube output at lower tube potentials. Therefore, the selection of kV and mA should take those kinds of constraints as well as the specific diagnostic imaging task in to consideration. For conventional quasi-linear CT systems employing linear filtered back-projection (FBP) image reconstruction algorithm, the optimization of kV-mA combinations are relatively straightforward, as neither spatial resolution nor noise texture has significant dependence on kV and mA settings. In these cases, zero-frequency analysis such as contrast-to-noise ratio (CNR) or normalized CNR by dose (CNRD) can be used for optimal kV-mA selection. The recently introduced statistical model-based iterative reconstruction (MBIR) method, however, has introduced new challenges to optimal kV and mA selection, as both spatial resolution and noise texture become closely correlated with kV and mA. In this work, a task-based approach based on modern signal detection theory and the corresponding frequency-dependent analysis has been proposed to perform the kV and mA optimization for both FBP and MBIR. By performing exhaustive measurements of task-based detectability index through the technically accessible kV-mA parameter space, iso-detectability contours were generated and overlaid on top of iso-dose contours, from which the kV-mA pair that minimize dose yet still achieving the desired detectability level can be identified.

The dynamic, piecewise-linear attenuator has been proposed as a concept which can shape the radiation flux incident on the patient. By reducing the signal to photon-rich measurements and increasing the signal to photon-starved measurements, the piecewise-linear attenuator has been shown to improve dynamic range, scatter, and variance and dose metrics in simulation. The piecewise-linear nature of the proposed attenuator has been hypothesized to mitigate artifacts at transitions by eliminating jump discontinuities in attenuator thickness at these points. We report the results of a prototype implementation of this concept. The attenuator was constructed using rapid prototyping technologies and was affixed to a tabletop x-ray system. Images of several sections of an anthropormophic pediatric phantom were produced and compared to those of the same system with uniform illumination. The thickness of the illuminated slab was limited by beam collimation and an analytic water beam hardening correction was used for both systems. Initial results are encouraging and show improved image quality, reduced dose and low artifact levels.

In this work we developed a novel denoising algorithm for DSA image series. This algorithm takes advantage of the low rank nature of the DSA image sequences to enable a dramatic reduction in radiation and/or contrast doses in DSA imaging. Both spatial and temporal regularizers were introduced in the optimization algorithm to further reduce noise. To validate the method, in vivo animal studies were conducted with a Siemens Artis Zee biplane system using different radiation dose levels and contrast concentrations. Both conventionally processed DSA images and the DSA images generated using the novel denoising method were compared using absolute noise standard deviation and the contrast to noise ratio (CNR). With the application of the novel denoising algorithm for DSA, image quality can be maintained with a radiation dose reduction by a factor of 20 and/or a factor of 2 reduction in contrast dose. Image processing is completed on a GPU within a second for a 10s DSA data acquisition.

Unintentional head motion during Positron Emission Tomography (PET) data acquisition can degrade PET image quality and lead to artifacts. Poor patient compliance, head tremor, and coughing are examples of movement sources. Head motion due to patient non-compliance can be an issue with the rise of amyloid brain PET in dementia patients. To preserve PET image resolution and quantitative accuracy, head motion can be tracked and corrected in the image reconstruction algorithm. While fiducial markers can be used, a contactless approach is preferable. A video-based head motion tracking system for a dedicated portable brain PET scanner was developed. Four wide-angle cameras organized in two stereo pairs are used for capturing video of the patient’s head during the PET data acquisition. Facial points are automatically tracked and used to determine the six degree of freedom head pose as a function of time. The presented work evaluated the newly designed tracking system using a head phantom and a moving American College of Radiology (ACR) phantom. The mean video-tracking error was 0.99±0.90 mm relative to the magnetic tracking device used as ground truth. Qualitative evaluation with the ACR phantom shows the advantage of the motion tracking application. The developed system is able to perform tracking with accuracy close to millimeter and can help to preserve resolution of brain PET images in presence of movements.

The promising increase in cancer detection rates^{1, 2} makes digital breast tomosynthesis (DBT) an interesting alternative to full-field digital mammography (FFDM) in breast cancer screening. However, this benefit comes at the cost of an increased average glandular dose in a combined DBT plus FFDM acquisition protocol. Synthetic mammograms, which are computed from the reconstructed tomosynthesis volume data, have demonstrated to be an alternative to a regular FFDM exposure in a DBT plus synthetic 2D reading mode.³ Besides weighted averaging and modified maximum intensity projection (MIP) methods,^{4, 5} the integration of CAD techniques for computing a weighting function in the forward projection step of the synthetic mammogram generation has been recently proposed.^{6, 7} In this work, a novel and computationally efficient method is presented based on an edge-retaining algorithm, which directly computes the weighting function by an edge-detection filter.

To perform task-based image quality assessment in CT, it is desirable to have a large number of realistic patient images with known diagnostic truth. One effective way to achieve this objective is to create hybrid images that combine patient images with simulated lesions. Because conventional hybrid images generated in the image-domain fails to reflect the impact of scan and reconstruction parameters on lesion appearance, this study explored a projection-domain approach. Liver lesion models were forward projected according to the geometry of a commercial CT scanner to acquire lesion projections. The lesion projections were then inserted into patient projections (decoded from commercial CT raw data with the assistance of the vendor) and reconstructed to acquire hybrid images. To validate the accuracy of the forward projection geometry, simulated images reconstructed from the forward projections of a digital ACR phantom were compared to physically acquired ACR phantom images. To validate the hybrid images, lesion models were inserted into patient images and visually assessed. Results showed that the simulated phantom images and the physically acquired phantom images had great similarity in terms of HU accuracy and high-contrast resolution. The lesions in the hybrid image had a realistic appearance and merged naturally into the liver background. In addition, the inserted lesion demonstrated reconstruction-parameter-dependent appearance. Compared to conventional image-domain approach, our method enables more realistic hybrid images for image quality assessment.

Mathematical model observers are expected to assist in preclinical optimization of image acquisition and reconstruction parameters. A clinically realistic and robust model observer platform could help in multiparameter optimizations without requiring frequent human-observer validations. We are developing search-capable visual-search (VS) model observers with this potential. In this work, we present initial results on optimization of DBT scan angle and the number of projection views for low-contrast mass detection. Comparison with human-observer results shows very good agreement. These results point towards the benefits of using relatively wider arcs and low projection angles per arc degree for improved mass detection. These results are particularly interesting considering that the FDA-approved DBT systems like Hologic Selenia Dimensions uses a narrow (15-degree) acquisition arc and one projection per arc degree.

This work compares the visibility of spheres and simulated masses in 2D-mammography and tomosynthesis systems using human observer studies. Performing comparison studies between breast imaging systems poses a number of practical challenges within a clinical environment. We therefore adopted a simulation approach which included synthetic breast blocks, a validated lesion simulation model and a set of validated image modelling tools as a viable alternative to clinical trials. A series of 4-alternative forced choice (4AFC) human observer experiments has been conducted for signal detection tasks using masses and spheres as targets. Five physicists participated in the study viewing images with a 5mm target at a range of contrast levels and 60 trials per experimental condition. The results showed that tomosynthesis has a lower threshold contrast than 2D-mammography for masses and spheres, and that detection studies using spheres may produce overly-optimistic threshold contrast values.

The development of new x-ray imaging techniques often requires prior knowledge of tissue attenuation, but the sources of such information are sparse. We have measured the attenuation of adipose breast tissue using spectral imaging, in vitro and in vivo. For the in-vitro measurement, fixed samples of adipose breast tissue were imaged on a spectral mammography system, and the energy-dependent x-ray attenuation was measured in terms of equivalent thicknesses of aluminum and poly-methyl methacrylate (PMMA). For the in-vivo measurement, a similar procedure was applied on a number of spectral screening mammograms. The results of the two measurements agreed well and were consistent with published attenuation data and with measurements on tissue-equivalent material.

Scanning-beam digital x-ray (SBDX) is an inverse geometry fluoroscopy system for low dose cardiac imaging. The use of a narrow scanned x-ray beam in SBDX reduces detected x-ray scatter and improves dose efficiency, however the tight beam collimation also limits the maximum achievable x-ray fluence. To increase the fluence available for imaging, we have constructed a new SBDX prototype with a wider x-ray beam, larger-area detector, and new real-time image reconstructor. Imaging is performed with a scanning source that generates 40,328 narrow overlapping projections from 71 x 71 focal spot positions for every 1/15 s scan period. A high speed 2-mm thick CdTe photon counting detector was constructed with 320x160 elements and 10.6 cm x 5.3 cm area (full readout every 1.28 s), providing an 86% increase in area over the previous SBDX prototype. A matching multihole collimator was fabricated from layers of tungsten, brass, and lead, and a multi-GPU reconstructor was assembled to reconstruct the stream of captured detector images into full field-of-view images in real time. Thirty-two tomosynthetic planes spaced by 5 mm plus a multiplane composite image are produced for each scan frame. Noise equivalent quanta on the new SBDX prototype measured 63%-71% higher than the previous prototype. X-ray scatter fraction was 3.9-7.8% when imaging 23.3-32.6 cm acrylic phantoms, versus 2.3- 4.2% with the previous prototype. Coronary angiographic imaging at 15 frame/s was successfully performed on the new SBDX prototype, with live display of either a multiplane composite or single plane image.

Breast compression is utilized in mammography to improve image quality and reduce radiation dose. Lesion conspicuity is improved by reducing scatter effects on contrast and by reducing the superposition of tissue structures. However, patient discomfort due to breast compression has been cited as a potential cause of noncompliance with recommended screening practices. Further, compression may also occlude blood flow in the breast, complicating imaging with intravenous contrast agents and preventing accurate quantification of contrast enhancement and kinetics. Previous studies have investigated reducing breast compression in planar mammography and digital breast tomosynthesis (DBT), though this typically comes at the expense of degradation in image quality or increase in mean glandular dose (MGD). We propose to optimize the image acquisition technique for reduced compression in DBT without compromising image quality or increasing MGD. A zero-frequency signal-difference-to-noise ratio model is employed to investigate the relationship between tube potential, SDNR and MGD. Phantom and patient images are acquired on a prototype DBT system using the optimized imaging parameters and are assessed for image quality and lesion conspicuity. A preliminary assessment of patient motion during DBT with minimal compression is presented.

Purpose: Conventional workflow in interventional imaging often ignores a wealth of prior information of the patient anatomy and the imaging task. This work introduces a task-driven imaging framework that utilizes such information to prospectively design acquisition and reconstruction techniques for cone-beam CT (CBCT) in a manner that maximizes task-based performance in subsequent imaging procedures. Methods: The framework is employed in jointly optimizing tube current modulation, orbital tilt, and reconstruction parameters in filtered back-projection reconstruction for interventional imaging. Theoretical predictors of noise and resolution relates acquisition and reconstruction parameters to task-based detectability. Given a patient-specific prior image and specification of the imaging task, an optimization algorithm prospectively identifies the combination of imaging parameters that maximizes task-based detectability. Initial investigations were performed for a variety of imaging tasks in an elliptical phantom and an anthropomorphic head phantom. Results: Optimization of tube current modulation and view-dependent reconstruction kernel was shown to have greatest benefits for a directional task (e.g., identification of device or tissue orientation). The task-driven approach yielded techniques in which the dose and sharp kernels were concentrated in views contributing the most to the signal power associated with the imaging task. For example, detectability of a line pair detection task was improved by at least three fold compared to conventional approaches. For radially symmetric tasks, the task-driven strategy yielded results similar to a minimum variance strategy in the absence of kernel modulation. Optimization of the orbital tilt successfully avoided highly attenuating structures that can confound the imaging task by introducing noise correlations masquerading at spatial frequencies of interest. Conclusions: This work demonstrated the potential of a task-driven imaging framework to improve image quality and reduce dose beyond that achievable with conventional imaging approaches.

In the last decade C–arm–based cone–beam CT became a widely used modality for intraoperative imaging. Typically a C–arm scan is performed using a circle–like trajectory around a region of interest. Therefor an angular range of at least 180° plus fan–angle must be covered to ensure a completely sampled data set. This fact defines some constraints on the geometry and technical specifications of a C–arm system, for example a larger C radius or a smaller C opening respectively. These technical modifications are usually not beneficial in terms of handling and usability of the C–arm during classical 2D applications like fluoroscopy. The method proposed in this paper relaxes the constraint of 180◦ plus fan–angle rotation to acquire a complete data set. The proposed C–arm trajectory requires a motorization of the orbital axis of the C and of ideally two orthogonal axis in the C plane. The trajectory consists of three parts: A rotation of the C around a defined iso–center and two translational movements parallel to the detector plane at the begin and at the end of the rotation. Combining these three parts to one trajectory enables for the acquisition of a completely sampled dataset using only 180° minus fan–angle of rotation. To evaluate the method we show animal and cadaver scans acquired with a mobile C-arm prototype. We expect that the transition of this method into clinical routine will lead to a much broader use of intraoperative 3D imaging in a wide field of clinical applications.

One of the major challenges for in vivo, micro-computed tomography (CT) imaging is poor soft tissue contrast. To increase contrast, exogenous contrast agents can be used as imaging probes. Combining these probes with a photon counting x-ray detector (PCXD) allows energy-sensitive CT and probe material decomposition from a series of images associated with different x-ray energies. We have implemented full-spectrum micro-CT using a PCXD and 2 keV energy sampling. We then decomposed multiple k-edge contrast materials present in an object (iodine, barium, and gadolinium) from water. Since the energy bins were quite narrow, the projection data was very noisy. This noise and further spectral distortions amplify errors in post-reconstruction material decompositions. Here, we propose and demonstrate a novel post-reconstruction denoising scheme which jointly enforces local and global gradient sparsity constraints, improving the contrast-to-noise ratio in full-spectrum micro-CT data and resultant material decompositions. We performed experiments using both calibration phantoms and ex vivo mouse data. Denoising increased the material contrast-to-noise ratio by an average of 13 times relative to filtered backprojection reconstructions. The relative decomposition error after denoising was 21%. To further improve material decomposition accuracy in future work, we also developed a model of the spectral distortions caused by PCXD imaging using known spectra from radioactive isotopes (¹⁰⁹Cd, ¹³³Ba). In future work, we plan to combine this model with the proposed denoising algorithm, enabling material decomposition with higher sensitivity and accuracy.

We are developing a hybrid, dual-source micro-CT system based on the combined use of an energy integrating (EID) x-ray detector and a photon counting x-ray detector (PCXD). Due to their superior spectral resolving power, PCXDs have the potential to reduce radiation dose and to enable functional and molecular imaging with CT. In most current PCXDs, however, spatial resolution and field of view are limited by hardware development and charge sharing effects. To address these problems, we propose spectral deblurring—a relatively simple algorithm for increasing the spatial resolution of hybrid, spectral CT data. At the heart of the algorithm is the assumption that the underlying CT data is piecewise constant, enabling robust recovery in the presence of noise and spatial blur by enforcing gradient sparsity. After describing the proposed algorithm, we summarize simulation experiments which assess the trade-offs between spatial resolution, contrast, and material decomposition accuracy given realistic levels of noise. When the spatial resolution between imaging chains has a ratio of 5:1, spectral deblurring results in a 52% increase in the material decomposition accuracy of iodine, gadolinium, barium, and water vs. linear interpolation. For a ratio of 10:1, a realistic representation of our hybrid imaging system, a 52% improvement was also seen. Overall, we conclude that the performance breaks down around high frequency and low contrast structures. Following the simulation experiments, we apply the algorithm to ex vivo data acquired in a mouse injected with an iodinated contrast agent and surrounded by vials of iodine, gadolinium, barium, and water.

Cardiac CT acquisitions for perfusion assessment can be performed in a dynamic or static mode. In this simulation study, we evaluate the relative classification and quantification performance of these modes for assessing myocardial blood flow (MBF). In the dynamic method, a series of low dose cardiac CT acquisitions yields data on contrast bolus dynamics over time; these data are fit with a model to give a quantitative MBF estimate. In the static method, a single CT acquisition is obtained, and the relative CT numbers in the myocardium are used to infer perfusion states. The static method does not directly yield a quantitative estimate of MBF, but these estimates can be roughly approximated by introducing assumed linear relationships between CT number and MBF, consistent with the ways such images are typically visually interpreted. Data obtained by either method may be used for a variety of clinical tasks, including 1) stratifying patients into differing categories of ischemia and 2) using the quantitative MBF estimate directly to evaluate ischemic disease severity. Through simulations, we evaluate the performance on each of these tasks. The dynamic method has very low bias in MBF estimates, making it particularly suitable for quantitative estimation. At matched radiation dose levels, ROC analysis demonstrated that the static method, with its high bias but generally lower variance, has superior performance in stratifying patients, especially for larger patients.

Differential phase contrast imaging is a promising new image modality that utilizes the refraction rather than the absorption of x-rays to image an object. A Talbot-Lau interferometer may be used to permit differential phase contrast imaging with a conventional medical x-ray source and detector. However, the current size of the gratings fabricated for these interferometers are often relatively small. As a result, data truncation image artifacts are often observed in a tomographic acquisition and reconstruction. When data are truncated in x-ray absorption imaging, the methods have been introduced to mitigate the truncation artifacts. However, the same strategy to mitigate absorption truncation artifacts may not be appropriate for differential phase contrast or dark field tomographic imaging. In this work, several new methods to mitigate data truncation artifacts in a multi-contrast imaging system have been proposed and evaluated for tomosynthesis data acquisitions. The proposed methods were validated using experimental data acquired for a bovine udder as well as several cadaver breast specimens using a benchtop system at our facility.

Digital chest tomosynthesis (DCT) is a 3D imaging modality which has been shown to approach the diagnostic capability of CT, but uses only one-tenth the radiation dose of CT. One limitation of current commercial DCT is the mechanical motion of the x-ray source which prolongs image acquisition time and introduces motion blurring in images. By using a carbon nanotube (CNT) x-ray source array, we have developed a stationary digital chest tomosynthesis (s- DCT) system which can acquire tomosynthesis images without mechanical motion, thus enhancing the image quality. The low dose and high quality 3D image makes the s-DCT system a viable imaging tool for monitoring cystic fibrosis (CF) patients. The low dose is especially important in pediatric patients who are both more radiosensitive and have a longer lifespan for radiation symptoms to develop. The purpose of this research is to evaluate the feasibility of using s-DCT as a faster, lower dose means for diagnosis and monitoring of CF in pediatric patients. We have created an imaging phantom by injecting a gelatinous mucus substitute into porcine lungs and imaging the lungs from within an anthropomorphic hollow chest phantom in order to mimic the human conditions of a CF patient in the laboratory setting. We have found that our s-DCT images show evidence of mucus plugging in the lungs and provide a clear picture of the airways in the lung, allowing for the possibility of using s- DCT to supplement or replace CT as the imaging modality for CF patients.

Dual-energy contrast-enhanced digital breast tomosynthesis (DE CE-DBT) uses an iodinated contrast agent to image the three-dimensional breast vasculature. The University of Pennsylvania has an ongoing DE CE-DBT clinical study in patients with known breast cancers. The breast is compressed continuously and imaged at four time points (1 pre-contrast; 3 post-contrast). DE images are obtained by a weighted logarithmic subtraction of the high-energy (HE) and low-energy (LE) image pairs. Temporal subtraction of the post-contrast DE images from the pre-contrast DE image is performed to analyze iodine uptake. Our previous work investigated image registration methods to correct for patient motion, enhancing the evaluation of vascular kinetics. In this project we investigate a segmentation algorithm which identifies blood vessels in the breast from our temporal DE subtraction images. Anisotropic diffusion filtering, Gabor filtering, and morphological filtering are used for the enhancement of vessel features. Vessel labeling methods are then used to distinguish vessel and background features successfully. Statistical and clinical evaluations of segmentation accuracy in DE-CBT images are ongoing.

Full field digital mammography (FFDM) has been the gold standard for mammography. It detects the presence, distribution, and morphology of microcalcifications (MCs), helping predict malignancy. Digital breast tomosynthesis (DBT) has overcome some limitations of FFDM such as poor sensitivity, specificity, and positive predictive values, due to superimposition of tissue, especially in dense breasts. Current DBT systems move an x-ray tube in either continuous (CM), or step-and-shoot motion (SSM). These systems are less effective than FFDM in MC detection due to lower spatial resolution. Motion of the x-ray source and system mechanical instability cause image blur. The image quality is further affected by patient motion due to the relatively long scan time. We developed a stationary DBT (s-DBT) system using a carbon nanotube (CNT) X-ray source array. The CNT array is electronically controlled, rapidly acquiring projection images over a large angular span, with zero tube motion. No source motion, coupled with a large angular span, results in improved in-plane and depth resolution. Using physical phantoms and human specimens, this system demonstrated higher spatial resolution than CM DBT. The objective of this study is to compare the diagnostic clinical performance of s-DBT to that of FFDM. Under UNC’s IRB regulations, 100 patients with breast lesions are being recruited and imaged with both modalities. A reader study will compare the diagnostic accuracy of the modalities. We have successfully imaged the first 30 patients. Initial results indicate that s-DBT alone produces comparable MC sharpness, and increased lesion conspicuity compared to FFDM.

Virtual clinical trials (VCT) can be carefully designed to inform, orient, or potentially replace clinical trials. The focus of this study was to demonstrate the capability of the sophisticated tools that can be used in the design, implementation, and performance analysis of VCTs, through characterization of the effect of background tissue density and heterogeneity on the detection of irregular masses in digital breast tomosynthesis. Twenty breast phantoms from the extended cardiactorso (XCAT) family, generated based on dedicated breast computed tomography of human subjects, were used to extract a total of 2173 volumes of interest (VOI) from simulated tomosynthesis images. Five different lesions, modeled after human subject tomosynthesis images, were embedded in the breasts, for a total of 6×2173 VOIs with and without lesions. Effects of background tissue density and heterogeneity on the detection of the lesions were studied by implementing a doubly composite hypothesis signal detection theory paradigm with location known exactly, lesion known exactly, and background known statistically. The results indicated that the detection performance as measured by the area under the receiver operating characteristic curve (ROC) deteriorated as density was increased, yielding findings consistent with clinical studies. The detection performance varied substantially across the twenty breasts. Furthermore, the log-likelihood ratio under H₀ and H₁ seemed to be affected by background tissue density and heterogeneity differently. Considering background tissue variability can change the outcomes of a VCT and is hence of crucial importance. The XCAT breast phantoms can address this concern by offering realistic modeling of background tissue variability based on a wide range of human subjects.

There is a clinical need to improve cerebral perfusion assessment during the treatment of ischemic stroke in the interventional suite. The clinician is able to determine whether the arterial blockage was successfully opened but is unable to sufficiently assess blood flow through the parenchyma. C-arm spin acquisitions can image the cerebral blood volume (CBV) but are challenged to capture the temporal dynamics of the iodinated contrast bolus, which is required to derive, e.g., cerebral blood flow (CBF) and mean transit time (MTT). Here we propose to utilize a circular tomosynthesis acquisition on the C-arm to achieve the necessary temporal sampling of the volume at the cost of incomplete data. We address the incomplete data problem by using tools from compressed sensing and incorporate temporal interpolation to improve our temporal resolution. A CT neuro perfusion data set is utilized for generating a dynamic (4D) volumetric model from which simulated tomo projections are generated. The 4D model is also used as a ground truth reference for performance evaluation. The performance that may be achieved with the tomo acquisition and 4D reconstruction (under simulation conditions, i.e., without considering data fidelity limitations due to imaging physics and imaging chain) is evaluated. In the considered scenario, good agreement between the ground truth and the tomo reconstruction in the parenchyma was achieved.

Neutron Stimulated Emission Computed Tomography (NSECT) is an emerging noninvasive imaging technique that measures the distribution of isotopes from biological tissue using fast-neutron inelastic scattering reaction. As a high-energy neutron beam illuminates the sample, the excited nuclei emit gamma rays whose energies are unique to the emitting nuclei. Tomographic images of each element in the spectrum can then be reconstructed to represent the spatial distribution of elements within the sample using a first generation tomographic scan. NSECT's high radiation dose deposition, however, requires a sampling strategy that can yield maximum image quality under a reasonable radiation dose. In this work, we introduce an NSECT sinogram sampling technique based on the Normalized Mutual Information (NMI) of the reconstructed images. By applying the Radon Transform on the ground-truth image obtained from a carbon-based synthetic phantom, different NSECT sinogram configurations were simulated and compared by using the NMI as a similarity measure. The proposed methodology was also applied on NSECT images acquired using MCNP5 Monte Carlo simulations of the same phantom to validate our strategy. Results show that NMI can be used to robustly predict the quality of the reconstructed NSECT images, leading to an optimal NSECT acquisition and a minimal absorbed dose by the patient.

The main purpose of this work is to study the ability of denoising algorithms to reduce the radiation dose in Digital Breast Tomosynthesis (DBT) examinations. Clinical use of DBT is normally performed in “combo-mode”, in which, in addition to DBT projections, a 2D mammogram is taken with the standard radiation dose. As a result, patients have been exposed to radiation doses higher than used in digital mammography. Thus, efforts to reduce the radiation dose in DBT examinations are of great interest. However, a decrease in dose leads to an increased quantum noise level, and related decrease in image quality. This work is aimed at addressing this problem by the use of denoising techniques, which could allow for dose reduction while keeping the image quality acceptable. We have studied two “state of the art” denoising techniques for filtering the quantum noise due to the reduced dose in DBT projections: Non-local Means (NLM) and Block-matching 3D (BM3D). We acquired DBT projections at different dose levels of an anthropomorphic physical breast phantom with inserted simulated microcalcifications. Then, we found the optimal filtering parameters where the denoising algorithms are capable of recovering the quality from the DBT images acquired with the standard radiation dose. Results using objective image quality assessment metrics showed that BM3D algorithm achieved better noise adjustment (mean difference in peak signal to noise ratio < 0.1dB) and less blurring (mean difference in image sharpness ~ 6%) than the NLM for the projections acquired with lower radiation doses.

Virtual clinical trials have been proposed as a viable alternative to clinical trials for testing and comparing the performance of breast imaging systems. One of the main simulation methodologies used in virtual trials employs clinical images of patients in which simulated models of cancer are inserted using a physics-based template multiplication technique. The purpose of this work is to investigate two assumptions commonly considered in this simulation approach: Firstly, given the absence of useful depth information in a clinical situation, an average measure of the local breast glandularity is commonly used as an estimate of the breast composition at the insertion site; secondly, it is also assumed that any change in the relative noise in the image at the insertion site, after insertion of a mass, is negligible. In order to test the validity of these assumptions, spheres representing idealised masses and anthropomorphic computational breast phantoms with perfect prior knowledge of local tissue composition and distribution were used. Results from several region of interest (ROI) insertions demonstrated a lack of variation obtained in contrast with insertion depth using the template multiplication insertion method as compared to the true depth-wise variation contrast values obtained from voxel replacement in a heterogeneous phantom. It was also found that the amount of noise is underestimated by insertion of spherical masses using template multiplication method by 8% - 29% compared to voxel replacement for the test conditions. This resulted in up to 12% variation in contrast-to-noise-ratio (CNR) values between template multiplication and voxel replacement methods.

The recent advancements in the graphics card technology raised the performance of parallel computing and contributed to the introduction of iterative reconstruction methods for x-ray computed tomography in clinical CT scanners. Iterative maximum likelihood (ML) based reconstruction methods are known to reduce image noise and to improve the diagnostic quality of low-dose CT. However, iterative reconstruction of a region of interest (ROI), especially ML based, is challenging. But for some clinical procedures, like cardiac CT, only a ROI is needed for diagnostics. A high-resolution reconstruction of the full field of view (FOV) consumes unnecessary computation effort that results in a slower reconstruction than clinically acceptable. In this work, we present an extension and evaluation of an existing ROI processing algorithm. Especially improvements for the equalization between regions inside and outside of a ROI are proposed. The evaluation was done on data collected from a clinical CT scanner. The performance of the different algorithms is qualitatively and quantitatively assessed. Our solution to the ROI problem provides an increase in signal-to-noise ratio and leads to visually less noise in the final reconstruction. The reconstruction speed of our technique was observed to be comparable with other previous proposed techniques. The development of ROI processing algorithms in combination with iterative reconstruction will provide higher diagnostic quality in the near future.

Cardiac computed tomography (CCT) has been widely used in diagnoses of coronary artery diseases due to the continuously improving temporal and spatial resolution. When helical CT with a lower pitch scanning mode is used, the effective radiation dose can be significant when compared to other radiological exams. Many methods have been developed to reduce radiation dose in coronary CT exams including high pitch scans using dual source CT scanners and step-and-shot scanning mode for both single source and dual source CT scanners. Additionally, software methods have also been proposed to reduce noise in the reconstructed CT images and thus offering the opportunity to reduce radiation dose while maintaining the desired diagnostic performance of a certain imaging task. In this paper, we propose that low-dose scans should be considered in order to avoid the harm from accumulating unnecessary X-ray radiation. However, low dose CT (LDCT) images tend to be degraded by quantum noise and streak artifacts. Accordingly, in this paper, a 3D dictionary representation based image processing method is proposed to reduce CT image noise. Information on both spatial and temporal structure continuity is utilized in sparse representation to improve the performance of the image processing method. Clinical cases were used to validate the proposed method.

We have developed a simulation tool for modeling the performance of CMOS-based medical x-ray detectors, based on the Monte Carlo toolkit GEANT4. Following the Fujita-Lubberts-Swank approach recently reported by Star-Lack et al., we calculate modulation transfer function MTF(f), noise power spectrum NPS(f) and detective quantum efficiency DQE(f) curves. The complete optical stack is modeled, including scintillator, fiber optic plate (FOP), optical adhesive and CMOS image sensor. For critical parts of the stack, detailed models have been developed, taking into account their respective microstructure. This includes two different scintillator types: Gd₂O₂S:Tb (GOS) and CsI:Tl. The granular structure of the former is modeled using anisotropic Mie scattering. The columnar structure of the latter is introduced into calculations directly, using the parameterization capabilities of GEANT4. The underlying homogeneous CsI layer is also incorporated into the model as well as the optional reflective layer on top of the scintillator screen or the protective polymer top coat. The FOP is modeled as an array of hexagonal bundles of fibers. The simulated CMOS stack consists of layers of Si₃N₄ and SiO₂ on top of a silicon pixel array. The model is validated against measurements of various test detector structures, using different x-ray spectra (RQA5 and RQA-M2), showing good match between calculated and measured MTF(f) and DQE(f) curves.

The skin dose tracking system (DTS) that we developed provides a color-coded illustration of the cumulative skin dose distribution on a 3D graphic of the patient during fluoroscopic procedures for immediate feedback to the interventionist. To improve the accuracy of dose calculation, we now have incorporated two additional important corrections (1) for the holder used to immobilize the head in neuro-interventions and (2) for the built-in compensation filters used for beam equalization. Both devices have been modeled in the DTS software so that beam intensity corrections can be made. The head-holder is modeled as two concentric hemi-cylindrical surfaces such that the path length between those surfaces can be determined for rays to individual points on the skin surface. The head-holder on the imaging system we used was measured to attenuate the primary x-rays by 10 to 20% for normal incidence, and up to 40% at non-normal incidence. In addition, three compensation filters of different shape are built into the collimator apparatus and were measured to have attenuation factors ranging from 58% to 99%, depending on kVp and beam filtration. These filters can translate and rotate in the beam and their motion is tracked by the DTS using the digital signal from the imaging system. When it is determined that a ray to a given point on the skin passes through the compensation filter, the appropriate attenuation correction is applied. These corrections have been successfully incorporated in the DTS software to provide a more accurate determination of skin dose.

In chest tomosynthesis (TS) the most commonly used reconstruction methods are based on Filtered Back Projection (FBP) algorithms. Due to the limited angular range of x-ray projections, FBP reconstructed data is typically associated with a low spatial resolution in the out-of-plane dimension. Lung nodule measures that depend on depth information such as 3D shape and volume are therefore difficult to estimate. In this paper the relation between features from FBP reconstructed lung nodules and the true out-of-plane nodule elongation is investigated and a method for estimating the out-of-plane nodule elongation is proposed. In order to study these relations a number of steps that include simulation of spheroidal-shaped nodules, insertion into synthetic data volumes, construction of TS-projections and FBP-reconstruction were performed. In addition, the same procedure was used to simulate nodules and insert them into clinical chest TS projection data. The reconstructed nodule data was then investigated with respect to in-plane diameter, out-of-plane elongation, and attenuation coefficient. It was found that the voxel value in each nodule increased linearly with nodule elongation, for nodules with a constant attenuation coefficient. Similarly, the voxel value increased linearly with in-plane diameter. These observations indicate the possibility to predict the nodule elongation from the reconstructed voxel intensity values. Such a method would represent a quantitative approach to chest tomosynthesis that may be useful in future work on volume and growth rate estimation of lung nodules.

This paper is about geometric calibration of the high resolution CT (Computed Tomography) system. Geometric calibration refers to the estimation of a set of parameters that describe the geometry of the CT system. Such parameters are so important that a little error of them will degrade the reconstruction images seriously, so more accurate geometric parameters are needed in the higher-resolution CT systems. But conventional calibration methods are not accurate enough for the current high resolution CT system whose resolution can reach sub-micrometer or even tens of nanometers. In this paper, we propose a new calibration method which has higher accuracy and it is based on the optimization theory. The superiority of this method is that we build a new cost function which sets up a relationship between the geometrical parameters and the binary reconstruction image of a thin wire. When the geometrical parameters are accurate, the cost function reaches its maximum value. In the experiment, we scanned a thin wire as the calibration data and a thin bamboo stick as the validation data to verify the correctness of the proposed method. Comparing with the image reconstructed with the geometric parameters calculated by using the conventional calibration method, the image reconstructed with the parameters calculated by our method has less geometric artifacts, so it can verify that our method can get more accurate geometric calibration parameters. Although we calculated only one geometric parameter in this paper, the geometric artifacts are still eliminated significantly. And this method can be easily generalized to all the geometrical parameters calibration in fan-beam or cone-beam CT systems.

In CT, the magnitude of enhancement is proportional to the amount of contrast medium (CM) injected. However, high doses of iodinated CM pose health risks, ranging from mild side effects to serious complications such as contrast-induced nephropathy (CIN). This work presents a method that enables the reduction of CM dosage, without affecting the diagnostic image quality. The technique proposed takes advantage of the additional spectral information provided by photon-counting CT systems. In the first step, we apply a material decomposition technique on the projection data to discriminate iodine from other materials. Then, we estimated the noise of the decomposed image by calculating the Cramér-Rao lower bound of the parameter estimator. Next, we iteratively reconstruct the iodine-only image by using the decomposed image and the estimation of noise as an input into a maximum-likelihood iterative reconstruction algorithm. Finally, we combine the iodine-only image with the original image to enhance the contrast of low iodine concentrations. The resulting reconstructions show a notably improved contrast in the final images. Quantitatively, the combined image has a significantly improved CNR, while the measured concentrations are closer to the actual concentrations of the iodine. The preliminary results from our technique show the possibility of reducing the clinical dosage of iodine, without affecting the diagnostic image quality.

We present a theoretical framework describing projections obtained from computed tomography systems considering physics of each component consisting of the systems. The projection model mainly consists of the attenuation of x-ray photons through objects including x-ray scatter and the detection of attenuated/scattered x-ray photons at pixel detector arrays. X-ray photons are attenuated by the Beers-Lambert law and scattered by using the Klein-Nishina formula. The cascaded signal-transfer model for the detector includes x-ray photon detection and light photon conversion/spreading in scintillators, light photon detection in photodiodes, and the addition of electronic noise quanta. On the other hand, image noise is considered by re-distributing the pixel signals in pixel-by-pixel ways at each image formation stage using the proper distribution functions. Instead of iterating the ray tracing over each energy bin in the x-ray spectrum, we first perform the ray tracing for an object only considering the thickness of each component. Then, we assign energy-dependent linear attenuation coefficients to each component in the projected images. This approach reduces the computation time by a factor of the number of energy bins in the x-ray spectrum divided by the number of components in the object compared with the conventional ray-tracing method. All the methods developed in this study are validated in comparisons with the measurements or the Monte Carlo simulations.

Non invasive fractional flow reserve derived from CT coronary angiography (CT-FFR) has to date been typically performed using the principles of fluid analysis in which a lumped parameter coronary vascular bed model is assigned to represent the impedance of the downstream coronary vascular networks absent in the computational domain for each coronary outlet. This approach may have a number of limitations. It may not account for the impact of the myocardial contraction and relaxation during the cardiac cycle, patient-specific boundary conditions for coronary artery outlets and vessel stiffness. We have developed a novel approach based on 4D-CT image tracking (registration) and structural and fluid analysis, to address these issues. In our approach, we analyzed the deformation variation of vessels and the volume variation of vessels, primarily from 70% to 100% of cardiac phase, to better define boundary conditions and stiffness of vessels. We used a statistical estimation method based on a hierarchical Bayes model to integrate 4D-CT measurements and structural and fluid analysis data. Under these analysis conditions, we performed structural and fluid analysis to determine pressure, flow rate and CT-FFR. The consistency of this method has been verified by a comparison of 4D-CTFFR analysis results derived from five clinical 4D-CT datasets with invasive measurements of FFR. Additionally, phantom experiments of flexible tubes with/without stenosis using pulsating pumps, flow sensors and pressure sensors were performed. Our results show that the proposed 4D-CT-FFR analysis method has the potential to accurately estimate the effect of coronary artery stenosis on blood flow.

The most accurate technique to model the X- and gamma radiation path through a numerically defined object is the Monte Carlo simulation which follows single photons according to their interaction probabilities. A simplified and much faster approach, which just integrates total interaction probabilities along selected paths, is known as ray tracing. Both techniques are used in medical imaging for simulating real imaging systems and as projectors required in iterative tomographic reconstruction algorithms. These approaches are ready for massive parallel implementation e.g. on Graphics Processing Units (GPU), which can greatly accelerate the computation time at a relatively low cost. In this paper we describe the application of the NVIDIA OptiX ray-tracing engine, popular in professional graphics and rendering applications, as a new powerful tool for X- and gamma ray-tracing in medical imaging. It allows the implementation of a variety of physical interactions of rays with pixel-, mesh- or nurbs-based objects, and recording any required quantities, like path integrals, interaction sites, deposited energies, and others. Using the OptiX engine we have implemented a code for rapid Monte Carlo simulations of Single Photon Emission Computed Tomography (SPECT) imaging, as well as the ray-tracing projector, which can be used in reconstruction algorithms. The engine generates efficient, scalable and optimized GPU code, ready to run on multi GPU heterogeneous systems. We have compared the results our simulations with the GATE package. With the OptiX engine the computation time of a Monte Carlo simulation can be reduced from days to minutes.

Algorithmic system modeling which includes a geometric motion of source, phantom, and detector for reconstructing the tomographic images is well-known in medical imaging field. Especially in a digital X-ray tomosynthesis system (DTS) which scans an object in limited angle, not a full 360-degree, an accurate system modeling should be derived to reconstruct an excellent cross sectional image. In this study, we analytically modeled a ray-driven forward projector and a pixel-driven back-projector. We firstly acquired forward projected images of a computerized Shepp-Logan phantom over an ±20° angular range using ray-driven projector. On top of that, we reconstructed the analytically scanned phantom using pixel-driven back-projector based on a conventional filtered back-projection (FBP) in tomosynthesis. We evaluated root-mean-square errors (RMSEs) and horizontal profiles of normalized pixel values in the reconstructed axial cross sectional images. The results indicated that pixel-driven back-projector combined with ray-driven projector showed low RMSEs of 0.25, 0.49, 0.80, 1.46, and 0.94 among five different regions-of-interests (ROIs). Illustrated horizontal profiles of normalized pixel values between the referenced phantom and reconstructed object showed similar values, which demonstrated that both ray- and pixel-driven projector/back-projector could be utilized in linear motion DTS.

Projection and back-projection are the most computational consuming parts in Computed Tomography (CT) reconstruction. Parallelization strategies using GPU computing techniques have been introduced. We in this paper present a new parallelization scheme for both projection and back-projection. The proposed method is based on CUDA technology carried out by NVIDIA Corporation. Instead of build complex model, we aimed on optimizing the existing algorithm and make it suitable for CUDA implementation so as to gain fast computation speed. Besides making use of texture fetching operation which helps gain faster interpolation speed, we fixed sampling numbers in the computation of projection, to ensure the synchronization of blocks and threads, thus prevents the latency caused by inconsistent computation complexity. Experiment results have proven the computational efficiency and imaging quality of the proposed method.

This work proposes a simulation model involving subtraction of digital mammography images obtained at different X-ray beam levels of energy to aid the detection of breast malignant lesions. Absorption coefficients behavior of 3 main structures of clinical interest – adipose tissue, fiber glandular tissue and the typical carcinoma – as a function of the beam energy from a Mo X-ray tube was the basis to develop a computer simulation of the possible acquired images. The simulation has considered a typical compressed breast with 4.5cm in thickness, and variations of the carcinoma and glandular tissues thicknesses - 0.4 up to 2.0cm and 4.1 to 2.5cm, respectively - were evaluated as a function of the photons mean energy - 14 up to 25 keV, in the typical mammography energy range. Results have shown that: (a) if the carcinoma thickness is over 0.4cm, its detection may be feasible even masked by fiber tissue with exposures in the range of 19 to 25 keV; (b) for masked carcinoma with thickness in the range of 0.4-2.0cm, the proposed procedure can enhance it in the image resulting from the digital subtraction between images obtained at 14 and at 22 keV. Therefore such results indicate that this simulation procedure can be a useful tool in aiding the identification of possible missed malignant lesions which could not be detected in the typical exam, mainly considering dense breasts.

Diffusion-weighted magnetic resonance imaging (DW-MRI) provides a novel insight into the brain to facilitate our understanding of the brain connectivity and microstructure. While in-vivo DW-MRI enables imaging of living patients and longitudinal studies of brain changes, post-mortem ex-vivo DW-MRI has numerous advantages. Ex-vivo imaging benefits from greater resolution and sensitivity due to the lack of imaging time constraints; the use of tighter fitting coils; and the lack of movement artifacts. This allows characterization of normal and abnormal tissues with unprecedented resolution and sensitivity, facilitating our ability to investigate anatomical structures that are inaccessible in-vivo. This also offers the opportunity to develop today novel imaging biomarkers that will, with tomorrow’s MR technology, enable improved in-vivo assessment of the risk of disease in an individual. Post-mortem studies, however, generally rely on the fixation of specimen to inhibit tissue decay which starts as soon as tissue is deprived from its blood supply. Unfortunately, fixation of tissues substantially alters tissue diffusivity profiles. In addition, ex-vivo DW-MRI requires particular care when packaging the specimen because the presence of microscopic air bubbles gives rise to geometric and intensity image distortion. In this work, we considered the specific requirements of post-mortem imaging and designed an optimized protocol for ex-vivo whole brain DW-MRI using a human clinical 3T scanner. Human clinical 3T scanners are available to a large number of researchers and, unlike most animal scanners, have a bore diameter large enough to image a whole human brain. Our optimized protocol will facilitate widespread ex-vivo investigations of large specimen.

Among the various metrics that quantify radiation dose in computed tomography (CT), organ dose is one of the most representative quantities reflecting patient-specific radiation burden.1 Accurate estimation of organ dose requires one to effectively model the patient anatomy and the irradiation field. As illustrated in previous studies, the patient anatomy factor can be modeled using a library of computational phantoms with representative body habitus.² However, the modeling of irradiation field can be practically challenging, especially for CT exams performed with tube current modulation. The central challenge is to effectively quantify the scatter irradiation field created by the dynamic change of tube current. In this study, we present a convolution-based technique to effectively quantify the primary and scatter irradiation field for TCM examinations. The organ dose for a given clinical patient can then be rapidly determined using the convolution-based method, a patient-matching technique, and a library of computational phantoms. 58 adult patients were included in this study (age range: 18–70 y.o., weight range: 60–180 kg). One computational phantom was created based on the clinical images of each patient. Each patient was optimally matched against one of the remaining 57 computational phantoms using a leave-one-out strategy. For each computational phantom, the organ dose coefficients (CTDI_vol-normalized organ dose) under fixed tube current were simulated using a validated Monte Carlo simulation program. Such organ dose coefficients were multiplied by a scaling factor, (CTDI_vol )_{organ, convolution} that quantifies the regional irradiation field. The convolution-based organ dose was compared with the organ dose simulated from Monte Carlo program with TCM profiles explicitly modeled on the original phantom created based on patient images. The estimation error was within 10% across all organs and modulation profiles for abdominopelvic examination. This strategy enables prospective and retrospective patient-specific dose estimation without the need of Monte Carlo simulation.

Computerized Tomography (CT) is a powerful radiographic imaging technology but the health risk due to the exposure of x-ray radiation has drawn wide concern. In this study, we propose to use kVp modulation to reduce the radiation dose and achieve the personalized low dose CT. Two sets of simulation are performed to demonstrate the effectiveness of kVp modulation and the corresponding calibration. The first simulation used the helical body phantom (HBP) that is an elliptical water cylinder with high density bone inserts. The second simulation uses the NCAT phantom to emulate the practical use of kVp modulation approach with region of interest (ROI) selected in the cardiac region. The kVp modulation profile could be optimized view by view based on the knowledge of patient attenuation. A second order correction is applied to eliminate the beam hardening artifacts. To simplify the calibration process, we first generate the calibration vectors for a few representative spectra and then acquire other calibration vectors with interpolation. The simulation results demonstrate the beam hardening artifacts in the images with kVp modulation can be eliminated with proper beam hardening correction. The results also show that the simplification of calibration did not impair the image quality: the calibration with the simplified and the complete vectors both eliminate the artifacts effectively and the results are comparable. In summary, this study demonstrates the feasibility of kVp modulation and gives a practical way to calibrate the high order beam hardening artifacts.

The Medipix All Resolution Scanner (MARS) spectral CT is intended for small animal, pre-clinical imaging and uses an x-ray detector (Medipix) operating in single photon counting mode. The MARS system provides spectrometric information to facilitate differentiation of tissue types and bio-markers. For longitudinal studies of disease models, it is desirable to characterise the system’s dosimetry. This dosimetry study is performed using three phantoms each consisting of a 30 mm diameter homogeneous PMMA cylinder simulating a mouse. The imaging parameters used for this study are derived from those used for gold nanoparticle identification in mouse kidneys. Dosimetry measurement are obtained with thermo-luminescent Lithium Fluoride (LiF:CuMgP) detectors, calibrated in terms of air kerma and placed at different depths and orientations in the phantoms. Central axis TLD air kerma rates of 17.2 (± 0.71) mGy/min and 18.2 (± 0.75) mGy/min were obtained for different phantoms and TLD orientations. Validation measurements were acquired with a pencil ionization chamber, giving an air-kerma rate of 20.3 (±1) mGy/min and an estimated total air kerma of 81.2 (± 4) mGy for a 720 projection acquisition. It is anticipated that scanner design improvements will significantly decrease future dose requirements. The procedures developed in this work will be used for further dosimetry calculations when optimizing image acquisition for the MARS system as it undergoes development towards human clinical applications.

Radiation exposure during CT imaging has drawn growing concern from academia, industry as well as the general public. Sinusoidal tube current modulation has been available in most commercial products and used routinely in clinical practice. To further exploit the potential of tube current modulation, Sperl et al. proposed a Computer-Assisted Scan Protocol and Reconstruction (CASPAR) scheme [6] that modulates the tube current based on the clinical applications and patient specific information. The purpose of this study is to accelerate the CASPAR scheme to make it more practical for clinical use and investigate its dose benefit for different clinical applications. The Monte Carlo simulation in the original CASPAR scheme was substituted by the dose reconstruction to accelerate the optimization process. To demonstrate the dose benefit, we used the CATSIM package generate the projection data and perform standard FDK reconstruction. The NCAT phantom at thorax position was used in the simulation. We chose three clinical cases (routine chest scan, coronary CT angiography with and without breast avoidance) and compared the dose level with different mA modulation schemes (patient specific, sinusoidal and constant mA) with matched image quality. The simulation study of three clinical cases demonstrated that the patient specific mA modulation could significantly reduce the radiation dose compared to sinusoidal modulation. The dose benefits depend on the clinical application and object shape. With matched image quality, for chest scan the patient specific mA profile reduced the dose by about 15% compared to the sinusoid mA modulation; for the organ avoidance scan the dose reduction to the breast was over 50% compared to the constant mA baseline.