This PDF file contains the front matter associated with SPIE Proceedings Volume 8668, including the Title Page, Copyright information, Table of Contents, and the Conference Committee listing.

The purpose of our study is to demonstrate the effectiveness of using smartphones for measuring ionizing radiation and to combine the findings with radiation exposure from medical imaging procedures for recording individual annual/lifetime radiation dose and provide improved radiation protection. We developed an application for smartphones which can use the properties of the video camera chip in such smartphones together with highly sufficient statistical evaluation of the signals to detect ionizing radiation. We could show that this application can be used in a large range of dose rates from natural backgrounds to high dose rate pulsed radiation like in fluoroscopic radiation or CT investigations. We could also show that these kinds of systems might help to provide better radiation protection by advising medical staff to use radiation protection material in best dose saving ways.

Physical phantoms are essential for the development, optimization, and clinical evaluation of x-ray systems. These phantoms are used for various tests such as quality assurance testing, system characterization, reconstruction evaluation, and dosimetry. They should ideally be capable of serving as ground truth for purposes such as virtual clinical trials. Currently, there is no anthropomorphic 3D physical phantom commercially available. We present our development of a new suite of physical breast phantoms based on real patient data. The phantoms were generated from the NURBS-based extended cardiac-torso (XCAT) breast phantoms, which were segmented from patient dedicated breast computed tomography data. High-resolution multi-material 3D printing technology was used to fabricate the physical models. Glandular tissue and skin were presented by the most radiographically dense photopolymer available to the printer, mimicking a 75% glandular tissue. Adipose tissue was presented by the least radiographically dense photopolymer, mimicking a 35% glandular tissue. The glandular equivalency was measured by comparing x-ray images of samples of the photopolymers available to the printer with those of breast tissue-equivalent materials. The mammographic projections and tomosynthesis reconstructed images of fabricated models showed great improvement over available phantoms, presenting a more realistic breast background.

The detectors that are used for endovascular image-guided interventions (EIGI), particularly for neurovascular interventions, do not provide clinicians with adequate visualization to ensure the best possible treatment outcomes. Developing an improved x-ray imaging detector requires the determination of estimated clinical x-ray entrance exposures to the detector. The range of exposures to the detector in clinical studies was found for the three modes of operation: fluoroscopic mode, high frame-rate digital angiographic mode (HD fluoroscopic mode), and DSA mode. Using these estimated detector exposure ranges and available CMOS detector technical specifications, design requirements were developed to pursue a quantum limited, high resolution, dynamic x-ray detector based on a CMOS sensor with 50 μm pixel size. For the proposed MAF-CMOS, the estimated charge collected within the full exposure range was found to be within the estimated full well capacity of the pixels. Expected instrumentation noise for the proposed detector was estimated to be 50-1,300 electrons. Adding a gain stage such as a light image intensifier would minimize the effect of the estimated instrumentation noise on total image noise but may not be necessary to ensure quantum limited detector operation at low exposure levels. A recursive temporal filter may decrease the effective total noise by 2 to 3 times, allowing for the improved signal to noise ratios at the lowest estimated exposures despite consequent loss in temporal resolution. This work can serve as a guide for further development of dynamic x-ray imaging prototypes or improvements for existing dynamic x-ray imaging systems.

The multi-resolution technology was developed for dynamic flat panel detectors for X-ray imaging. This multi-resolution technology allows us to switch between the 1 x 1 mode (150 μm square) and the 2 x 2 mode (300 μm square) instantaneously, using an external control. We developed a 17" x 17" dynamic detector using this multi-resolution technology. This novel dynamic detector has a high sensitivity and a high-speed readout and it can reduce the radiation exposure dose and deliver a smooth image. The key feature of our multi-resolution technology is capable of reading 4 pixel signals simultaneously. The sensitivity and the readout speed in the 2 x 2 mode were 4 times higher than those in the 1 x 1 mode. The multi-resolution technology was implemented using a unique thin film transistor structure; that is, one pixel has two switches, each of which are turned on/off depending on the readout mode. As a result, the dynamic detector with a large active area of 17" x 17" realized a high detective quantum efficiency value of 40% under the low radiation of RQA5 20 nGy and a high-speed readout of 30 frames/sec. This multi-resolution technology made it possible to reduce the radiation exposure dose in a variety of applications.

Field emission electron sources have been increasingly investigated and practiced as cold electron sources for many Xray generation mechanisms especially in certain medical imaging applications, such as computed tomography (CT). In a field emission electron source, the emission current and cathode life are the two key performance parameters of interests. Conventional field emission electron source in the form of a 2-dimensional single surface cathode design is often undergoing the bottleneck of limited emission current. Higher current can be obtained by increasing the strength of driving electric field. However this is at the expense of a reduced cathode life. In this paper we present a novel field emission electron source design based on a 3-dimensional semi-enclosed cavity structure, by utilizing the cavity’s multiple inner faces as electron emission surfaces. The cavity has one bottom inner face open for electron ejection, the area of which was kept the same as a conventional single surface cathode to maintain the same initial electron beam cross section. The extended emission area of the new cathode design due to its 3-dimensional structure provides a higher electron emission current while maintaining the same beam cross section, resulting in an improved effective emission current density. We have demonstrated a significantly increased electron emission current from this 3-dimensional cathode under a constant electric field, without over-driving the cathode field emitters. Alternatively this new cathode design can also achieve the same total emission current with a lower electric field compared with a conventional 2- dimensional cathode, which will significantly extend its life.

Preclinical imaging is a cornerstone of translational research, as all therapeutic drugs need to be tested for efficacy and toxicity on animals prior to human trials. Optical imaging techniques, such as bioluminescence and multispectral fluorescence imaging, currently dominate preclinical functional imaging despite their depth dependent limitations on quantitation and sensitivity. Translating drugs developed with these techniques to clinical models can therefore be difficult. Hence, clinically relevant nuclear imaging techniques, such as SPECT and PET, are therefore becoming increasingly used in preclinical imaging. Dedicated preclinical SPECT and PET systems are now available, but for many preclinical research groups this requires a significant investment in new equipment.

Active matrix flat-panel imagers (AMFPIs) have become ubiquitous in medical imaging environments. AMFPIs are based on two-dimensional pixelated arrays coupled to various x-ray converter materials that provide either indirect or direct detection of the incident x-ray radiation. However, the capabilities of this technology are severely constrained by the underlying solid-state properties of the amorphous silicon semiconductor material employed in the thin-film transistors present in each array pixel. The considerably higher electron and hole mobilities of polycrystalline silicon, a semiconductor material that (like amorphous silicon) is well suited to fabrication of transistors for large area electronics, provide the potential to overcome these constraints by increasing the overall gain of the system relative to the electronic additive noise. To explore this potential, a series of prototype arrays based on increasingly complex pixel designs employing polycrystalline silicon transistors is under development by our collaboration. The designs include several generations of active pixel arrays that incorporate sophisticated pixel-level amplifier circuits with the goal of improving imaging performance. In this paper, an initial analysis of the noise and DQE performance of selected prototype pixel circuit designs will be presented. The results are based on a combination of Monte Carlo -based circuit simulations and cascaded systems analysis, supplemented with information obtained from measurements performed on poly-Si transistors. The paper concludes with a brief discussion of the potential for, and challenges associated with, the creation of single photon counting arrays based on poly-Si TFTs.

In this work, we demonstrate the ability to determine the material composition of a sample by measuring coherent scatter diffraction patterns generated using a coded-aperture x-ray scatter imaging (CAXSI) system. Most materials are known to exhibit unique diffraction patterns through coherent scattering of low-energy x-rays. However, clinical x-ray imagers typically discard scatter radiation as noise that degrades image quality. Through the addition of a coded aperture, the system can be sensitized to coherent scattered photons that carry information about the identity and location of the scattering material. In this work, we demonstrate this process using a Monte-Carlo simulation of a CAXSI system. A simulation of a CAXSI system was developed in GEANT4 with modified physics libraries to model coherent scatter diffraction patterns in materials. Simulated images were generated from 10 materials including plastics, hydrocarbons, and biological tissue. The materials were irradiated using collimated pencil- and fan-beams with energies of 160 kVp. The diffraction patterns were imaged using a simulated 2D detector and mathematically deconstructed using an analytical projection model that accounted for the known x-ray source spectrum. The deconstructed diffraction patterns were then matched with a library of known coherent scatter form-factors of different materials to determine the identity of the scatterer at different locations in the object. The results showed good agreement between the measured and known scatter patterns from the materials, demonstrating the ability to image and identify materials at different 3D locations within an object using a projection-based CAXSI system.

Individual projection images in Digital Breast Tomosynthesis (DBT) must be acquired with low levels of radiation, which significantly increases image noise. This work investigates the influence of a denoising algorithm and the Anscombe transformation on the reduction of quantum noise in DBT images. The Anscombe transformation is a variance-stabilizing transformation that converts the signal-dependent quantum noise to an approximately signalindependent Gaussian additive noise. Thus, this transformation allows for the use of conventional denoising algorithms, designed for additive Gaussian noise, on the reduction of quantum noise, by working on the image in the Anscombe domain. In this work, denoising was performed by an adaptive Wiener filter, previously developed for 2D mammography, which was applied to a set of synthetic DBT images generated using a 3D anthropomorphic software breast phantom. Ideal images without noise were also generated in order to provide a ground-truth reference. Denoising was applied separately to DBT projections and to the reconstructed slices. The relative improvement in image quality was assessed using objective image quality metrics, such as peak signal-to-noise ratio (PSNR) and mean structural similarity index (SSIM). Results suggest that denoising works better for tomosynthesis when using the Anscombe transformation and when denoising was applied to each projection image before reconstruction; in this case, an average increase of 9.1 dB in PSNR and 58.3% in SSIM measurements was observed. No significant improvement was observed by using the Anscombe transformation when denoising was applied to reconstructed images, suggesting that the reconstruction algorithm modifies the noise properties of the DBT images.

Digital breast tomosynthesis (DBT) has been investigated as a promising alternative to conventional X-ray mammography for breast cancer screening. By reconstructing 3D volumetric images from multiple 2D projections measured over a limited angular range, it can offer depth-directional information and improve both sensitivity and specificity of cancer detection in dense breasts. The diagnostic performance of DBT can be affected by a number of imaging parameters. The angular range of scan orbit is one of the most crucial factors, since it determines the depth-directional resolution. Recently, we proposed the wide angle tomosynthesis based on voltage modulations of X-ray source. By using X-rays with large penetration power on exterior positions, it can acquire high-SNR projections over a wide angular range. In this paper, we present comparative studies on exposure conditions in DBT, including narrow and wide angle scan using an invariant tube voltage of X-ray source, and wide angle scan with the voltage modulation technique. In addition, we compared the conventional reconstruction methods with recently proposed IDIR algorithms. In preliminary studies, the wide-angle scheme with proposed IDIR algorithm showed superior performances in detecting abnormal lesions over conventional approaches.

Chest tomosynthesis is an imaging modality that provides 3D sectional information of a patients thoracic cavity using limited angle x-ray projections. Studies show that tomosynthesis can improve the detection of subtle lung nodules comparing to conventional radiography at a lower radiation dose than CT. In the conventional design, the projection images are collected by mechanically moving a single x-ray source to different viewing angles. We investigated the feasibility of stationary chest tomosynthesis using the distributed CNT x-ray source array technology, which can generate a scanning x-ray beam without any mechanical motion. A proof-of-concept system was constructed using a short linear source array and a at panel detector. The performance of the source including the flux was evaluated in the context of chest imaging. The bench-top system was characterized and images of a chest phantom were acquired and reconstructed. The preliminary results demonstrate the feasibility of stationary chest tomosynthesis using the CNT x-ray source array technology.

In digital breast tomosynthesis (DBT), a 3D image of the breast is generated from x-ray projections at various angles. There are two mechanisms for acquiring projection images in DBT, step-and-shoot motion and continuous tube motion. The benefit of continuous tube motion is shorter scan time and hence less patient motion; the trade-off is focal spot blurring. To minimize focal spot blurring in a system with continuous tube motion, this study proposes a new velocity profile for the x-ray tube during the scan. Unlike existing systems for which the x-ray tube has constant angular velocity, we investigate a smoothly-varying tube velocity that approaches zero during each projection and is larger between projections. With this unique design, the filtered backprojection reconstruction of a sinusoidal test object was calculated, and modulation was determined at various frequencies. It is shown that the newly proposed tube velocity yields increased modulation in the reconstruction relative to a conventional system with continuous tube motion. The modulation in the re-designed system differs minimally from an analogous step-and-shoot system operated with the same scan time. This improvement in image quality was validated with reconstructions of microcalcifications in computer breast phantoms. It is known that continuous tube motion reduces the contrast of microcalcifications relative to stepand-shoot systems; we show that the newly proposed tube motion increases the contrast of microcalcifications compared to conventional continuous tube motion. In conclusion, this work proposes a strategy for optimizing the velocity of tube motion in DBT.

Contrast enhanced (CE) breast imaging has been proposed as a method to increase the sensitivity and specificity of breast cancer detection. Because malignant lesions often exhibit angiogenesis, the uptake of radio-opaque contrast agents (e.g. iodine) results in increased attenuation compared to the background tissue. Both planar CE digital mammography (CE-DM) and digital breast tomosynthesis (CE-DBT) have been proposed, using temporal or dual energy (DE) subtraction to remove tissue backgrounds. In the current study, we apply a cascaded linear systems model approach to analyze CE techniques with DE subtraction for designing a diagnostic imaging study, including the effects of contrast dynamics. We apply the model for both CE-DM and CE-DBT to calculate the ideal observer signal-to-noise ratio (SNR) for the detection of I contrast objects of different sizes and concentrations. The calculation of this figure-of-merit (FOM) was be used to optimize CE clinical imaging protocols.

We have recently developed a method of using a distributed x-ray source array to obtain images with scatter correction for tomographic reconstruction of an object. The method consists of obtaining x-ray images of the object with and without the primary beam sampling apparatus. In this study, we report the results of applying the scatter correction method for breast tomosynthesis imaging using the carbon nanotube x-ray based stationary Digital Breast Tomosynthesis (s-DBT) system developed at UNC. The unique design of s-DBT system makes it possible to estimate the image of the scatter profile of the object with very low dose, and without significant increase in acquisition time. An anthropomorphic breast phantom was used for quantitative analysis of the change in contrast and scatter-to-primary ratio. Our results suggest that the scatter correction method is effective and can be used for enhanced contrast.

In digital breast tomosynthesis (DBT), the reconstructed image quality of small objects such as microcalcifications are limited by the detector resolution and the reconstruction voxel dimensions used, in addition to the inherent in-plane and inter-plane blurring due to limited-angle image acquisition. We are investigating the effects of subpixel reconstruction on the image quality of microcalcifications. In this study, subpixel projection images were generated by polynomial interpolation of the gray level values of the original projection images to reduce the pixel pitch by a factor of 2 to 4. The voxel dimensions in the reconstruction volume were varied. The in-plane voxel dimensions were chosen to match the pitch in the subpixel projection images while the voxel dimension in the depth-direction was varied systematically to study the relative effects. DBT scans of human subjects with microcalcifications were acquired with a GE prototype DBT system at 0.1 mm x 0.1 mm pixel pitch. In addition, computer modeling of the same system was used to generate simulated DBT projections of small spheres as a surrogate for microcalcifications. DBT volumes corresponding to the subpixel projections were reconstructed with SART. The FWHMs of the line profiles of the simulated microcalcifications on their in-focus DBT slices and FWHMs of the inter-plane artifact spread function (ASF) in the depth-direction were used for comparison of reconstruction quality. The results indicated that subpixel reconstruction increased the object contrast and reduced the inter-plane blurring to a certain extent. Further work is underway to study the dependence of the trade-off between image quality and reconstruction voxel dimensions on the image acquisition parameters (total scan angle, angular increment) of the DBT system and the properties of the objects of interest.

Present day treatment for neurovascular pathological conditions involves the use of devices with very small features such as stents, coils, and balloons; hence, these interventional procedures demand high resolution xray imaging under fluoroscopic conditions to provide the capability to guide the deployment of these fine endovascular devices. To address this issue, a high resolution x-ray detector based on EMCCD technology is being developed. The EMCCD field-of-view is enlarged using a fiber-optic taper so that the detector features an effective pixel size of 37 μm giving it a Nyquist frequency of 13.5 lp/mm, which is significantly higher than that of the state of the art Flat Panel Detectors (FPD). Quantitative analysis of the detector, including gain calibration, instrumentation noise equivalent exposure (INEE) and modulation transfer function (MTF) determination, are presented in this work. The gain of the detector is a function of the detector temperature; with the detector cooled to 50 C, the highest relative gain that could be achieved was calculated to be 116 times. At this gain setting, the lowest INEE was measured to be 0.6 μR/frame. The MTF, measured using the edge method, was over 2% up to 7 cycles/ mm. To evaluate the performance of the detector under clinical conditions, an aneurysm model was placed over an anthropomorphic head phantom and a coil was guided into the aneurysm under fluoroscopic guidance using the detector. Image sequences from the procedure are presented demonstrating the high resolution of this SSXII.

The new Solid State X-ray Image Intensifier (SSXII) is a high-resolution, high-sensitivity, real-time region-ofinterest (ROI) x-ray imaging detector. Evaluations were made of both standard linear systems metrics (MTF, DQE) and total system performance with generalized linear systems metrics (GMTF, GDQE) including scatter and geometric un-sharpness for simulated clinical conditions. The SSXII is based on a 1k x 1k EMCCD sensor coupled to a 300 μm thick CsI(Tl) phosphor through a 2.88:1 fiber optic taper resulting in a 37 μm effective pixel size and an effective 3.7 cm x 3.7 cm square field-of-view (FOV). Standard methods were used to calculate MTF, NNPS and DQE. Generalized metrics were calculated and compared for three different magnifications (1.03, 1.11 and 1.2) and three different focal spots (0.3 mm, 0.5 mm and 0.8 mm) for a scatter fraction of 0.28. For an RQA5 spectrum, at 5 cycles/mm the MTF was found to be 0.06 and DQE was 0.04, while the DQE(0) was 0.60. Focal spot un-sharpness and scatter significantly degrades the GMTF and GDQE performance of the detector. A low frequency drop is caused by scatter and is almost independent of focal spot size and magnification. The degradation for middle range frequencies is caused by geometric un-sharpness and increases with focal spot size and magnification. This degradation was least in the case of the small focal spot and almost independent of magnification. In spite of this degradation, the high resolution SSXII with a small FOV may have a significant impact on ROI image-guided neuro-interventions since it demonstrates far better performance than standard current detectors.

Knowledge of the uncertainty associated with Monte Carlo estimates is useful for determining when to stop a simulation run when statistical uctuations fall below a desired tolerance level, and for designing and analyzing variance reduction techniques. In this work, we discuss how to analytically calculate the uncertainty of Monte Carlo variance estimates from higher order moments of the distribution of events. In addition, we show how these expressions can be incorporated in the study of x-ray imaging detectors to manage runtime and precision of the simulation. Our analysis can be used to design variance reduction techniques for Monte Carlo simulations when, as in many cases in imaging, the variance and not the mean, is the quantity of interest.

Software breast phantoms have been developed for use in evaluation of novel breast imaging systems. Software phantoms are flexible allowing the simulation of wide variations in breast anatomy, and provide ground truth for the simulated tissue structures. Different levels of phantom realism are required depending on the intended application. Realistic simulation of dense (fibroglandular) tissue is of particular importance; the properties of dense tissue – breast percent density and the spatial distribution – have been related to the risk of breast cancer. In this work, we have compared two methods for simulation of dense tissue distribution in a software breast phantom previously developed at the University of Pennsylvania. The methods compared are: (1) the previously used Gaussian distribution centered at the phantom nipple point, and (2) the proposed combination of two Beta functions, one modeling the dense tissue distribution along the chest wall-to-nipple direction, and the other modeling the radial distribution in each coronal section of the phantom. Dense tissue distributions obtained using these methods have been compared with distributions reported in the literature estimated from the analysis of breast CT images. Qualitatively, the two methods produced rather similar dense tissue distributions. The simulation based upon the use of Beta functions provides more control over the simulated distributions through the selection of the various Beta function parameters. Both methods showed good agreement to the clinical data, suggesting both provide a high level of realism.

While great advances are made toward making highly realistic, surface models of the human anatomy, very little has been done to fill these bounded surfaces with models of anatomical texture. We propose a method whereby realistic anatomically-based computed tomography (CT) texture can be incorporated into voxelized versions of the 4D extended cardiac-torso (XCAT) phantom. Our source of texture comes from patient CT scans from the Duke CT imaging database. These image-sets were de-noised using anisotropic diffusion. Two organs were selected from which texture was obtained, liver and lungs. From each organ, multiple regions of interest (ROIs) were taken and tiled side-by-side to create a larger image. Textures for the liver and lungs were extrapolated using ImageQuilting, based on the tiled images. Next, a NURBSbased XCAT phantom was voxelized at the same resolution as the textures. The texture was then placed in the voxelized phantoms. Finally, CT simulations of the phantoms with and without the textures were compared against each other, using the power spectral density. This work shows that there is a way whereby the XCAT phantoms can be textured to give more realistic appearance in CT simulations. It is anticipated that this method would find great use in making projections of the XCAT phantom look more realistic and allow for the phantoms to not only be utilized in dosimetrical evaluations, but in image quality studies as well.

We have investigated the multi-energy performance of our most recent prototype CT scanner with CdTe-based counting detector. With its small pixel pitch of 225 μm this device is prepared for the high X-ray fluxes occurring in clinical CT. Each of these pixels is equipped with two adjustable counters. The ASIC architecture of the detector allows configuration of the counter thresholds in chess patterns, enabling data acquisition in up to four energy bins. We have studied the material separation capability of counting CT with respect to potential clinical applications. Therefore we have analyzed contrast and noise properties in material decomposed CT images using up to four base materials. We have studied contrast agents containing iodine, gadolinium, or gold, and the body-like materials calcium, fat, and water. We describe the mathematical framework used in this work and demonstrate the general multi-energy capability of counting CT with simulations and experimental data from our prototype scanner. To prove the clinical relevance of our studies we compare the results to those obtained with well-established dual-kVp techniques recorded at same patient dose and with identical image sharpness.

Photon counting detectors offer some unique features for x-ray imaging. If designed correctly, photon counting detectors have no readout noise and no dark counts. This is an important feature in for example low dose CT imaging where the total dose is distributed over a large number of projections from different angles. In addition to this it is also possible to incorporate pulse height discrimination of each photon event, thus enabling the recording of images from multiple energy intervals in a single exposure. We demonstrate the performance of a newly developed dual-energy fast photon counting detector with 100μm pixel size that can be read out up to 1000fps. It consists of a three side buttable ASIC that is bump bonded to a CdTe converter. The very high conversion efficiency of CdTe makes the detector suitable for a wide range of applications requiring high spatial resolution at low doses. The efficiency of the detector is maintained all the way out to the edge of the chip which opens up the possibility to build larger detectors still fulfilling medical requirements. The novel detector incorporates a charge sharing correction feature and the effect of this function is demonstrated using the DQE measurements for different spectra as well as with spectrum reconstruction from Cd¹⁰⁹ and Am²⁴¹ radioactive sources. We show that this charge sharing correction feature affects the properties of NPS and MTF, and the energy resolution is greatly enhanced. Measurements are also compared to a simulation model for the detector system.

Photon counting detectors are expected to bring along various clinical benefits in CT imaging. Among the benefits of these detectors is their intrinsic spectral sensitivity that allows to resolve the incident X-ray spectrum. Their capability for multi-energy imaging enables material segmentation, but it is also possible to use the spectral information to create fused gray-scale CT images with improved imaging properties. We have developed and investigated an optimization method that maximizes the image contrast-to-noise ratio, making use of the spectral information in data recorded with a counting detector with up to six energy thresholds. The resulting merged gray-scale CT images exhibit significantly improved CNR² for a number of clinically established, potentially novel and hypothetical contrast agents in the thin absorber approximation. In this work we motivate and describe the optimization method, provide the deduced optimal sets of threshold energies and mixing weights, and summarize the maximally achievable gain in CNR² for each contrast agent under study.

Spectral computed tomography (CT) with photon-counting detectors (PCDs) has the potential to substantially advance diagnostic CT imaging by reducing image noise and dose to the patient, by improving contrast and tissue specificity, and by enabling molecular and functional imaging. The current PCD technology, however, is limited by two main factors: imperfect energy measurement (spectral response effects, SRE) and limited count rate (pulse-pileup effects, PPE, due to detector dead-times). The overall goal of our research is to develop compensation algorithms for these sources of data distortions, to demonstrate that PCDs are suitable for clinical CT, and to identify key clinical applications for spectral CT with PCDs. We have already developed an iterative compensation scheme that includes a forward projection model of the imaging chain and that can compensate for either SRE or PPE distortions separately by maximizing a penalized log-likelihood function. In this paper we describe the evaluation of a combined, cascaded SRE and PPE model for PCDs and compare the models to data acquired with an experimental table-top PCD-CT system. The separation into count-rate independent effects (SRE) and count-rate dependent effects (PPE) allows cascading the forward model. First, the SRE model is evaluated using low count rates. Then the PPE model is cascaded and the combined SRE+PPE model is evaluated. Several different attenuators were used, including K-edge materials and the models and data were compared for various count-rate conditions.

Single-photon counting (SPC) x-ray imaging has the potential to improve image quality and enable new advanced energy-dependent methods. Recently, cascaded systems analysis (CSA) has been extended to the description of the detective quantum efficiency (DQE) of SPC detectors. In this article we apply the new CSA approach to the description of the DQE of hypothetical direct-conversion selenium (Sc) and cadmium zinc telluride (CdZnTc) detectors including the effects of poly-energetic x-ray spectra, stochastic conversion of x-ray energy to electron­ hole (c-h) pairs, depth-dependent collection of e-h pairs using the Hecht relation, additive electronic noise, and thresholding. Comparisons arc made to an energy-integrating model. For this simple model, with the exception of thick (1- 10 mm) Sc-bascd convertors, we found that the SPC DQE was 5-20 %greater than that of the energy­ integrating model. This trend was tnw even when additive noise was included in the SPC model and excluded from the energy-integrating model. However, the DQE of SPC detectors with poor collection efficiency (such as thick (<1 mm) Sc detectors) and high levels of additive noise can be degraded by 40-90 % for all energies and x-ray spectra considered. vVhile photon-counting approaches arc not yet ready for routine diagnostic imaging, the available DQE is equal to or higher than that of conventional energy-integrating detectors under a wide range of x-ray energies and convertor thickness. However, like energy-integrating detectors, the DQE of SPC detectors will be degraded by the combination of poor collection efficiency and high levels of additive noise.

Semiconductor based photon-counting detectors for x-ray CT have a number of advantages over energy integrat­ ing detectors, including reduced electronic and Swank noise, increased dynamic range, capability of spectral CT for material decomposition, and improved SNR characteristics through energy weighting. Quite a few clinical applications could benefit from high-resolution spectral CT. For example, in breast CT the visualization of mi­ crocalcifications and assessment of tumor microvasculature after contrast enhancement require spatial resolution on the order of 100 μm or better. A straightforward approach to increasing spatial resolution by decreasing the detector pixel size, leads to two major problems: 1) fabricating circuitry with small pixels becomes very costly, and 2) inter-pixel charge spreading can obviate any improvement in spatial resolution. In this study, we have used computer simulations to investigate position estimation algorithms that utilize charge sharing to achieve sub-pixel position resolution. To study these algorithms, we model a simple detector geometry with a pixellated 5 x 5 anode array, and use conditional probability functions modeling electron-hole charge transport in CZT. We used COMSOL Multiphysics software to map the distribution of charge pulses in the detector. Performance of two x-ray interaction position estimation algorithms were evaluated: 1) method of maximum likelihood, and 2) a fast, practical algorithm that can be realistically implemented in a readout ASIC, providing identification of the quadrant of the pixel in which interaction occurred. Both methods exhibit good sub-pixel resolution performance, however their actual efficiency is limited by electronic noise.

We have developed dual energy (DE) iodine contrast imaging functions with a commercial mammography and tomosynthesis system. Our system uses a tungsten target x-ray tube and selenium direct conversion detector. Conventional low energy (LE) images were acquired with existing Rh, Ag and Al filters at the screening doses while the high energy images (HE) were acquired with new Cu filters at half of the screening doses. In DE 2D mode, a pair of LE and HE images was taken with one second delay time between and with anti-scatter grid. In DE 3D mode, 22 views of alternating LE and HE were taken over 15 degrees angle in seven seconds without grid while tube was scanned continuously. We used log-subtraction algorithm to obtain clean DE images with the subtraction factor K derived empirically. In 3D mode, the subtraction was applied to each pair of LE and HE slices after reconstruction. The x-ray technique optimization was done with simulation and phantom study. We performed both phantom and patient studies to demonstrate the advantage of iodine contrast imaging. Among several new things in our work, a selenium detector optimized for DE imaging was tested and a large dose advantage was demonstrated; 2D and 3D DE images of a breast under same compression were acquired with a unique DE combo mode of the system, allowing direct image quality comparison between 2D and 3D modes. Our study showed that new DE system achieved good image quality. DE imaging is be a promising modality to detect breast cancer.

Stationary Digital Breast Tomosynthesis (s-DBT) is a carbon nanotube based breast imaging device with fast image acquisition and decent resolution. In this paper, we investigate several representative reconstruction methods with the recently improved s-DBT system and also introduce a two-step reconstruction strategy with Pre-computed Backprojection based penalized-likelihood (PPL). This strategy reconstructs three dimensional (3-D) images with a desired resolution properties by choosing the corresponding smoothing parameter, which is evaluated in advance by studying simulated data. Our experiments show that the current s-DBT system has been greatly improved with respect to the performance of image reconstructions. PPL method exhibits controllable pixel precision, high image contrast and low noise on reconstructed images. Therefore, the enhanced Contrast Noise Ratio (CNR) from PPL method benefits both micro-calcifications and mass of the breast-equivalent phantom.

A new algorithm is suggested to compute one or several virtual projection images directly from cone-beam data acquired in a tomosynthesis geometry. One main feature of this algorithm is that it does not involve the explicit reconstruction of a 3D volume, and a subsequent forward-projection operation, but rather operates using solely 2D image processing steps. The required 2D processing is furthermore based on the use of pre-computed entities, so that a significant speed-up in the computations can be obtained. The presented algorithm can be applied to a variety of CT geometries, and is here investigated for a mammography application, to simulate virtual mammograms from a set of low-dose tomosynthesis projection images. A first evaluation from real measured data is given.

We are investigating human-observer models that perform clinically realistic detection and localization tasks as a means of making reliable assessments of digital breast tomosynthesis images. The channelized non-prewhitening (CNPW) observer uses the background known exactly task for localization and detection. Visual-search observer models attempt to replicate the search patterns of trained radiologists. The visual-search observer described in this paper utilizes a two-phase approach, with an initial holistic search followed by directed analysis and decision making. Gradient template matching is used for the holistic search, and the CNPW observer is used for analysis and decision making. Spherical masses were embedded into anthropomorphic breast phantoms, and simulated projections were made using ray-tracing and a serial cascade model. A localization ROC study was performed on these images using the visual-search model observer and the CNPW observer. Observer performance from the two computer observers was compared to human observer performance. The visual-search observer was able to produce area under the LROC curve values similar to those from human observers; however, more research is needed to increase the robustness of the algorithm.

Digital breast tomosynthesis (DBT) is suggested to have superior performance compared to 2D mammography in terms of cancer visibility, especially in the case of dense breasts. However, the overall performance of tomosynthesis for screening applications, and the manner in which tomosynthesis should be optimally used for screening remains unclear. This motivates the development of software tools that can insert user-defined synthetic pathology of realistic appearance into clinical tomosynthesis images for subsequent use in virtual clinical trials. We present a method for inserting lesions grown using Diffusion Limited Aggregation, previously validated in 2D mammograms, into clinical DBT images. A preliminary pilot study was used to validate the realism of the masses, wherein three readers each viewed 19 cases and rated the realism of the inserted masses. Each case included a simulated mass inserted in the tomosynthesis projections and the counterpart digital 2D mammogram. These results show that masses can be successfully embedded in the tomosynthesis projections and can produce visually authentic DBT images containing synthetic pathology. These results will be used to further optimize the appearance of these masses in DBT for an upcoming validation.

Dark-field imaging has the potential to overcome limitations in computed tomography (CT) investigating relatively weakly absorbing material. However, an object-position dependence of the visibility loss in dark-field imaging is observed. This effect might be negligible for small objects, but, for acquisition geometries using fanangle apertures and field of views as those in human CT scanners, the object-position dependence of visibility loss has to be taken into consideration if the scattering structure within the object is in the range of the grating periods, i.e. micrometer. This work examines the effect of object-position dependent visibility loss in dark-field imaging experimentally, investigates its consequences and presents an algorithm which solves the corresponding reconstruction problem.

Grating-based X-ray phase-contrast imaging (XPCI) is a promising modality to increase soft-tissue contrast in medical imaging and especially in the case of mammography. Several groups worldwide have performed investigations on grating-based Talbot-Lau X-ray imaging of breast tissue, but in most cases focussed on the soft tissue contrast enhancement of the differential phase image. In this contribution, we present promising measurements with a Talbot-Lau interferometer of several mas­tectomy breast tissue samples especially focussed on the sensitivity of the dark-field signal of microcalcifications and with a comparable dose value to conventional mammography. We can present a contrast improvement for calcifications in surrounding breast tissue for the dark-field image by a factor of 10 related to the attenuation image.

Although the relative ease of implementation and compact nature of grating-based differential phase contrast CT (DPC-CT) has sparked tremendous enthusiasm for potential medical applications, the pros and cons of this imaging method remains to be addressed before an actual clinical system can be constructed. To address these unknowns, either numerical simulations or direct hardware implementations can be used. However, both approaches have their limitations. It is highly desirable to develop a research method to enable imaging performance prediction for a future DPC-CT system from the performance of an available absorption CT (ACT) system. In this paper, a theoretical framework was developed to accurately predict the noise properties and detection performance of DPC-CT from that of conventional ACT. The framework was derived based on a fundamental noise relationship between DPC-CT and ACT and was experimentally validated. An example has been given in the paper on how the framework can be utilized to predict model observer detectability index of a DPC breast CT constructed based on an existing absorption breast CT. This framework is expected to become a valuable tool in addressing the following questions: (i) With a fixed radiation dose in a particular clinical application, how well can a specific detection/discrimination imaging task can be performed provided that an existing ACT scanner is modified into a DPC-CT by inserting a grating interferometer, which is characterized by a few design parameters (e.g., pitches and duty cycles of the gratings, relative distance between the gratings, etc.) into the ACT system? (ii) If a DPC-CT system can outperform an ACT for certain detection/discrimination tasks under the constraint of identical radiation dose to the image object, how would one optimize design parameters of the gratings in order to maximize its potential clinical benefits?

The edge illumination principle was first proposed at Elettra (Italy) in the late nineties, as an alternative method for achieving high phase sensitivity with a very simple and flexible set-up, and has since been under continuous development in the radiation physics group at UCL. Edge illumination allows overcoming most of the limitations of other phase-contrast techniques, enabling their translation into a laboratory environment. It is relatively insensitive to mechanical and thermal instabilities and it can be adapted to the divergent and polychromatic beams provided by X-ray tubes. This method has been demonstrated to work efficiently with source sizes up to 100m, compatible with state-of-the-art mammography sources. Two full prototypes have been built and are operational at UCL. Recent activity focused on applications such as breast and cartilage imaging, homeland security and detection of defects in composite materials. New methods such as phase retrieval, tomosynthesis and computed tomography algorithms are currently being theoretically and experimentally investigated. These results strongly indicate the technique as an extremely powerful and versatile tool for X-ray imaging in a wide range of applications.

Today’s commercial X-ray micro computed tomography (CT) specimen systems are based on microfocus sources, 2D pixel array cameras and short source-to-detector distances (i.e. cone-beam configurations). High resolution is achieved by means of geometric magnification. The further development of such devices to acquire phase and scattering contrast images can dramatically enhance their range of applications. Due to the compact geometries, which imply a highly diverging beam, the gratings must be curved to maintain highest imaging performance over a large field of view. We report about the implementation of extremely compact Talbot and Talbot- Lau type grating interferometers which are compatible to the geometry of typical micro CT systems. For the analytical description of the imaging system, formulas are presented describing the dependency of the sensitivity on geometric parameters, camera and source parameters. Further, the imaging pipeline consisting of the data acquisition protocol, radiographic phase retrieval and tomographic image reconstruction is illustrated. The reported methods open the way for an immediate integration of phase and scattering contrast imaging on table top X-ray micro CT scanners.

The Talbot-Lau grating interferometer enables refraction based imaging with conventional X-ray tubes, offering the promise of a new medical imaging modality. The fringe contrast of the normal incidence interferometer is however insufficient at the >40 keV photon energies needed to penetrate thick body parts, because the thin absorption gratings used in the interferometer become transparent. To solve this problem we developed a new interferometer design using gratings at glancing incidence. For instance, using 120 μm thick Au gratings at 10° incidence we increased several fold the interferometer contrast for a spectrum with ~58 keV mean energy. Tests of DPC-CT at 60-80kVp using glancing angle interferometers and medically relevant samples indicate high potential for clinical applications. A practical design for a slot-scan DPC-CT system for the knee is proposed, using glancing angle gratings tiled on a single substrate.

The effects of beam hardening have previously been extended from absorption imaging to phase contrast imaging, showing a similar, albeit reduced, effect in the phase images. The effect of beam hardening on the interferometer performance, however, has not been demonstrated. In this work, the visibility reduction on a differential phase contrast imaging system due to spectral changes as a result of beam hardening is demonstrated. The implication of this reduction is an artificial increase in noise for the phase contrast image through highly-attenuating regions of the object. In addition, false signal will be recorded in the dark-field image, which normally shows only highly-scattering objects and interfaces. The results show that with added beam filtration, the effect is reduced, just as with more traditional beam hardening artifacts. However, the effect also means that one must also take into account the desired imaging task when determining the system’s design energy.

This study investigates the possibility of using the model observer detectability to represent human detection performance in differential phase contrast CT (DPC-CT). Five model observers were investigated, including the prewhitening (PW) observer, the prewhitening observer with eye filter and internal noise (PWEi), the nonprewhitening (NPW) observer, the non-prewhitening observer with eye filter (NPWE), and the channelized Hotelling observer (CHO). Human 2AFC experiments involving four physicist observers were also performed to provide reference to the model observer results. The contrast thresholds required for 92% correct decision rate in 2AFC tests were evaluated for each observer. While all five model observers show good correlations with human observers, the CHO generated the best quantitative agreement with human observer results. Compared with the time-consuming human observer method, the model observer method is of much higher efficiency for the evaluation and optimization of DPC-CT.

Current advances in CT dose reduction and image quality improvement mechanisms rely on moving towards a more patient / image specific approach. For example, statistical reconstruction algorithms tailor reconstruction weights to patient attenuation values and CT protocols make use of different kVp and mAs settings for different body regions and sizes. In this paper, for the first time to our knowledge, experimental results are presented in which a dynamic bowtie filter was used to tailor the imaging dose during a CT acquisition. The device used to implement a dynamic bowtie is referred to as a digitial beam attenuator (DBA). A non-DBA (non-modulated) CT scan was also performed for comparison. At half of the imaging dose, the noise uniformity of the DBA CT images was 37% better than the non-DBA scan. The use of a dynamic bowtie filter may also prove useful in photon counting CT (PCCT) where high fluence rates and large differences in fluence across projections and from view to view make implementing PCCT difficult. Results are presented in this paper showing dynamic range requirements of 7.9 and 13.8 for DBA and non-DBA scans respectively. The results also show that when using the DBA, the dose delivered can be reduced by 1.9 times for the small phantom used in this study. The differences in dynamic range and dose are somewhat small compared to what would be seen clinically due to the small size of the phantom used in the study. Using a more clinically relevant phantom, dynamic range differences of 22 times and dose reductions on the order of 4 times can be observed in which the effects of the DBA will be more dramatic.

Purpose: Dual-energy cone-beam CT (DE-CBCT) is an emerging technology with potential application in diagnostic imaging and image-guided interventions. This paper reports DE-CBCT feasibility and investigates decomposition algorithms for maximizing low-dose performance for reconstruction-based DE decomposition. A framework of binary decision theory is used to examine the accuracy of DE decompositions obtained from analytical reconstructions of differentially filtered low-energy (LE) and high-energy (HE) data and from penalized likelihood (PL) reconstructions with differential regularization using quadratic and total variation penalties. Methods: Accurate DE-CBCT decomposition benefits from consideration of all system noise components. Filtered backprojection (FBP) reconstruction-based decomposition was investigated with differential filtering of LE and HE data. Penalized likelihood reconstruction-based decomposition with differential regularization was hypothesized to further improve low-dose performance, especially when coupled with regularization through a total variation edge preserving penalty that encourages piecewise smooth images. Performance of decomposition was assessed in terms of a binary hypothesis framework of sensitivity, specificity, and accuracy. Studies involved experiments on a DE-CBCT testbench, phantoms of variable material type and concentration, and cadavers (knee arthrography). Results: Studies support the overall feasibility of accurate, low-dose DE-CBCT at concentration down to 5 mg/ml (iodine), dose ~3-6 mGy, and accuracy of material classification ~90%. Reconstruction-based decomposition with quadratic PL performed comparably to FBP. PL with a total variation penalty provided edge preservation and piecewise smooth images that aided DE classification and achieved improved performance over FBP and quadratic PL, reaching accuracy of ~0.98 for 2 mg/mL iodine at 3.2 mGy, compared to approx. 0.9 for FBP and quadratic PL. Conclusions: Accurate material decomposition with DE-CBCT is feasible at low dose and benefits from a rigorous assessment of noise mechanisms among various reconstruction-based techniques. The work points to the potential for non-linear iterative reconstruction methods for high-quality decomposition at low material concentration and dose.

In the current workflow of ischemic stroke management, it is highly desirable to obtain perfusion information with the C-arm CBCT system in the interventional room. Due to hardware limitations, the data acquisition speed of the current Carm CBCT systems is relatively slow and only 7 time frames are available for a 45 s perfusion study. In this study, a novel temporal recovery method was proposed to recover contrast enhancement curves in C-arm CBCT perfusion studies. The proposed temporal recovery problem is a constrained optimization problem. Two numerical methods were used to solve the proposed problem. The feasibility of proposed temporal recovery method was validated with numerical experiments. Both solvers can achieve a satisfactory solution for the temporal recovery problem, while the result of the Bregman algorithm is more accurate than that from the CG. In vivo animal studies were used to demonstrated the improvement of the proposed method in C-arm CBCT perfusion. A stoked canine model was scanned with both C-arm CBCT and diagnostic CT. Perfusion defects can be clearly indentified from the cerebral blood flow (CBF) map of diagnostic CT perfusion. Without the temporal recovery technique, these defects can hardly be identified from the CBCT CBF map. After applying the proposed temporal recovery method, the CBCT CBF map well correlates with the CBF map from diagnostic CT.

We propose a motion compensation approach based on algebraic reconstruction in case of coarse motion grid. The method is tested with Shepp-Logan phantom.

In X-ray imaging, a reduction of the field of view (FOV) is proportional to a reduction in radiation dose. The resulting truncation, however, is incompatible with conventional tomographic reconstruction algorithms. This problem has been studied extensively. Very recently, a novel method for region of interest (ROI) reconstruction from truncated projections with neither the use of prior knowledge nor explicit extrapolation has been published, named Approximated Truncation Robust Algorithm for Computed Tomography (ATRACT). It is based on a decomposition of the standard ramp filter into a 2D Laplace filtering (local operation) and a 2D Radon-based filtering step (non-local operation). The 2D Radon-based filtering that involves many interpolations complicates the filtering procedure in ATRACT, which essentially limits its practicality. In this paper, an optimization for this shortcoming is presented. That is to apply ATRACT in one dimension, which implies that we decompose the standard ramp filter into the 1D Laplace filter and a 1D convolutionbased filter. The convolution kernel was determined numerically by computing the 1D impulse response of the standard ramp filtering coupled with the second order anti-derivative operation. The proposed algorithm was evaluated by using a reconstruction benchmark test, a real phantom and a clinical data set in terms of spatial resolution, computational efficiency as well as robustness of correction quality. The evaluation outcomes were encouraging. The proposed algorithm showed improvement in computational performance with respect to the 2D ATRACT algorithm and furthermore maintained reconstructions of high accuracy in presence of data truncation.

Dedicated breast CT (bCT) technology may be useful for patients with high risk of developing breast cancer. Previous studies have shown that bCT outperforms mammography in the visualization of mass lesions, however mammography is superior in identifying microcalcifications. The Breast Tomography Project at UC Davis has led to development of three dedicated breast CT scanners that produce high resolution, fully tomographic images, overcoming tissue superposition effects found in mammography while maintaining an equivalent radiation dose. Over 600 patients have been imaged in an ongoing clinical trial. The first patient scan was performed on the latest bCT scanner developed at UC Davis, called Cambria, on April 12, 2012. The main differences between Cambria and the previous scanners are in using a pulsed xray source (generator and tube) instead of continuous x-ray sources, and also in using the non-binning mode of the flatpanel fluoroscopic detector. The spatial resolution characteristics of the new scanner were investigated and the results show significant improvement in the overall MTF properties. Based on these results, it was concluded that using the pulsed x-ray tube, we were able to restore the MTF degradation caused by motion blurring effect that exists in previous generations of bCT. Moreover, MTF analysis shows that using the detector in the native acquisition mode (1 x 1) results in superior spatial resolution which will likely bring considerable improvement to the delineation of microcalcifications.

The quantification of lung nodule volume from CT images provides valuable information for cancer diagnosis and staging. However, the usefulness of quantification depends on its precision. Direct assessment of the volume quantification precision involves multiple steps and can become intractable for a multiplicity of protocols. To assess quantification precision efficiently, we developed a prediction model, named the estimability index (e’). e’ provides a prediction of precision based on the characteristics of image noise and resolution, the nodule being quantified, and the segmentation software. It was further calibrated against empirical precision for 45 protocols of various reconstruction algorithms, slice thickness, and dose level. Results showed a strong correlation established between e’ and the empirical precision across all 45 protocols, demonstrating e’ as an effective surrogate of quantification precision. This study provides a useful framework for the optimization of CT protocols in terms of quantification precision. It also enables fast assessment of protocol compliance in terms of precision for biomarker quantification.

C-arm cone-beam CT is an emerging tool for intraoperative imaging, but currently exhibits modest soft-tissue imaging capability. This work adapts a spectrum of statistical iterative reconstruction approaches to C-arm CBCT and investigates performance in imaging of low-contrast tasks pertinent to soft-tissue surgical guidance. Experiments involved a mobile C-arm and phantoms and cadavers presenting soft-tissue structures imaged using 3D FBP and penalized likelihood reconstruction. Statistical reconstruction - especially non-quadratic PL - boosted soft-tissue image quality through reduction of noise and artifacts, therefore presenting promise for interventional imaging. Further investigation of task-specific performance may overcome conventional tradeoffs in noise, resolution, and dose.

Purpose: Nonstationarity of CT noise presents a major challenge to the assessment of image quality. This work presents models for imaging performance in both filtered backprojection (FBP) and penalized likelihood (PL) reconstruction that describe not only the dependence on the imaging chain but also the dependence on the object as well as the nonstationary characteristics of the signal and noise. The work furthermore demonstrates the ability to impart control over the imaging process by adjusting reconstruction parameters to exploit nonstationarity in a manner advantageous to a particular imaging task. Methods: A cascaded systems analysis model was used to model the local noise-power spectrum (NPS) and modulation transfer function (MTF) for FBP reconstruction, with locality achieved by separate calculation of fluence and system gain for each view as a function of detector location. The covariance and impulse response function for PL reconstruction (quadratic penalty) were computed using the implicit function theorem and Taylor expansion. Detectability index was calculated under the assumption of local stationarity to show the variation in task-dependent image quality throughout the image for simple and complex, heterogeneous objects. Control of noise magnitude and correlation was achieved by applying a spatially varying roughness penalty in PL reconstruction in a manner that improved overall detectability. Results: The models provide a foundation for task-based imaging performance assessment in FBP and PL image reconstruction. For both FBP and PL, noise is anisotropic and varies in a manner dependent on the path length of each view traversing the object. The anisotropy in turn affects task performance, where detectability is enhanced or diminished depending on the frequency content of the task relative to that of the NPS. Spatial variation of the roughness penalty can be exploited to control noise magnitude and correlation (and hence detectability). Conclusions: Nonstationarity of image noise is a significant effect that can be modeled in both FBP and PL image reconstruction. Prevalent spatial-frequency-dependent metrics of spatial resolution and noise can be analyzed under assumptions of local stationarity, providing a means to analyze imaging performance as a function of location throughout the image. Knowledgeable selection of a spatially-varying roughness penalty in PL can potentially improve local noise and spatial resolution in a manner tuned to a particular imaging task. Keywords: cascaded systems analysis, nonstationarity, filtered backprojection, penalized-likelihood reconstruction, noise-power spectrum, covariance matrix, imaging task, detectability index

Assessment of the resolution properties of nonlinear imaging systems is a useful but challenging task. While the modulation transfer function (MTF) fully describes contrast resolution as a function of spatial frequency for linear systems, an equivalent metric does not exist for systems with significant nonlinearity. Therefore, this preliminary investigation attempts to classify and quantify the amount of scaling and distortion imposed on a given image signal as the result of a nonlinear process (nonlinear image processing algorithm). As a proof-of-concept, a median filter is assessed in terms of its principle frequency response (PFR) and distortion response (DR) functions. These metrics are derived in frequency space using a sinusoidal basis function, and it is shown that, for a narrow-band sinusoidal input signal, the scaling and distortion properties of the nonlinear filter are described exactly by PFR and DR, respectively. The use of matched sinusoidal basis and input functions accurately reveals the frequency response to long linear structures of different scale. However, when more complex (multi-band) input signals are considered, PFR and DR fail to adequately characterize the frequency response due to nonlinear interaction effects between different frequency components during processing. Overall, the results reveal the context-dependent nature of nonlinear image processing algorithm performance, and they emphasize the importance of the basis function choice in algorithm assessment. In the future, more complex forms of nonlinear systems analysis may be necessary to fully characterize the frequency response properties of nonlinear algorithms in a context-dependent manner.

X-ray scatter degrades image contrast, uniformity and CT number accuracy in cone-beam computed tomography (CBCT). Correction methods based on the scatter kernel superposition (SKS) technique are efficient and suitable for many clinical applications but still produce residual errors due to limitations in the scatter kernel models. To reduce these errors, we propose to generate a first-pass reconstruction using a set of default SKS parameters followed by limited Monte Carlo simulations that are then used to perturb and refine key kernel parameters in order to obtain an improved second-pass correction. To test the approach, we used the fast adaptive scatter kernel model (fASKS) employing asymmetric kernels for the first-pass scatter correction and then used GEANT4 to simulate scatter-to-primary ratios in selected projections allowing for refined scatter estimates. The results show that a minimal number of projections require simulation in order to adequately perturb scatter kernel parameters for all projections. Compared to the default asymmetric kernels, the refined kernels reduced CT number errors from 24 HU to 15 HU in a large pelvis phantom resulting in a more uniform and accurate image.

Variational methods are useful for solving ill-posed inverse imaging problems by minimizing a cost function with a data fidelity term and a regularization term. For statistical X-ray computed tomography (CT) image reconstruction, penalized weighted least-squares (PWLS) criteria with edge-preserving regularization can improve quality of the reconstructed image compared to traditional filtered back-projection (FBP) reconstruction. Nevertheless, the huge dynamic range of the statistical weights used in PWLS image reconstruction leads to a highly shift-variant local impulse response, making effective preconditioning difficult. To overcome this problem, iterative algorithms based on variable splitting were proposed recently. However, existing splitting-based iterative algorithms do not consider the (unknown) gain fluctuations that can occur between views. This paper proposes a new variational formulation for splitting-based iterative algorithms where the unknown gain parameter vector and the image are estimated jointly with just simple changes to the original algorithms. Simulations show that the proposed algorithm greatly reduces the shading artifacts caused by gain fluctuations yet with almost unchanged computational complexity per iteration.

The increase in the use of CT scanning in the clinical setting is raising concerns from the medical community. In order to reduce the dose of ionizing radiation imparted to patients during CT scans, statistical image recon­ struction was proposed. This family of algorithm aims at improving image noise characteristics by modeling the stochastic x-ray detection process in the reconstruction algorithm. It was shown however that statistical recon­ struction may lead to an anisotropic spatial resolution. In this abstract, we study this tradeoff in the context of a statistical formulation of the dose reduction using prior image constrained compressed sensing framework (DR-PICCS). Two numerically-simulated phantoms and a dataset acquired in vivo were used for this evaluation. It is demonstrated that the inclusion of a statistical model in DR-PICCS may whiten the NPS and uniformize the noise spatial distribution in the image. However, the images may suffer from an anisotropic spatial resolution while the images reconstructed using DR-PICCS without statistical model have more isotropic spatial resolution. Due to the flexibility offered in PICCS, a specially-designed prior image processing method has been used in statistical DR-PICCS to palliate for the anisotropy in spatial resolution.