This PDF file contains the front matter associated with SPIE Proceedings Volume 7258, including the Title Page, Copyright information, Table of Contents, In Memoriam: Robert F. Wagner, and the Conference Committee listing.

In order to discuss the cost-benefit ratio of CT examinations in children, one must be familiar with the reasons why CT can provide a high collective or individual dose. The reasons include increasing CT use as well as lack of attention to dose reduction strategies. While those have been substantial efforts for dose reduction, additional work is necessary to prevent unnecessary radiation exposure. This responsibility is shared between science and medicine, industry, regulatory agencies, and patients as well.

C-arm based cone-beam CT (CBCT) has evolved into a routine clinical imaging modality to provide threedimensional tomographic image guidance before, during, and after an interventional procedure. It is often used to update the clinician to the state of the patient anatomy and interventional tool placement. Due to the repeatedly use of CBCT, the accumulated radiation dose in an interventional procedure has become a concern. There is a strong desire from both patients and health care providers to reduce the radiation exposure required for these exams. The overall objective of this work is to propose and validate a method to significantly reduce the total radiation dose used during a CBCT image guided intervention. The basic concept is that the first cone-beam CT scan acquired at the full dose will be used to constrain the reconstruction of the later CBCT scans acquired at a much lower radiation dose. A recently developed new image reconstruction algorithm, Prior Image Constrained Compressed Sensing (PICCS), was used to reconstruct subsequent CBCT images with lower dose. This application differs from other applications of the PICCS algorithm, such as time-resolved CT or fourdimensional CBCT (4DCBCT), because the patient position may be frequently changed from one CBCT scan to another during the procedure. Thus, an image registration step to account for the change in patient position is indispensable for use of the PICCS image reconstruction algorithm. In this paper, the image registration step is combined with the PICCS algorithm to enable radiation dose reduction in CBCT image guided interventions. Experimental results acquired from a clinical C-arm system using a human cadaver were used to validate the PICCS algorithm based radiation dose reduction scheme. Using the proposed method in this paper, it has been demonstrated that, instead of 300 view angles, this technique requires about 20 cone-beam view angles to reconstruct CBCT angiograms. This signals a radiation dose reduction by a factor of approximately fifteen for subsequent acquisitions.

The purpose of this study is to develop a method for estimating patient-specific dose from abdomen-pelvis CT examinations and to investigate dose variation across patients in the same weight group. Our study consisted of seven pediatric patients in the same weight/protocol group, for whom full-body computer models were previously created based on the patients' CT data obtained for clinical indications. Organ and effective dose of these patients from an abdomen-pelvis scan protocol (LightSpeed VCT scanner, 120-kVp, 85-90 mA, 0.4-s gantry rotation period, 1.375-pitch, 40-mm beam collimation, and small body scan field-of-view) was calculated using a Monte Carlo program previously developed and validated for the same CT system. The seven patients had effective dose of 2.4-2.8 mSv, corresponding to normalized effective dose of 6.6-8.3 mSv/100mAs (coefficient of variation: 7.6%). Dose variations across the patients were small for large organs in the scan coverage (mean: 6.6%; range: 4.9%-9.2%), larger for small organs in the scan coverage (mean: 10.3%; range: 1.4%-15.6%), and the largest for organs partially or completely outside the scan coverage (mean: 14.8%; range: 5.7%-27.7%). Normalized effective dose correlated strongly with body weight (correlation coefficient: r = -0.94). Normalized dose to the kidney and the adrenal gland correlated strongly with mid-liver equivalent diameter (kidney: r = -0.97; adrenal glands: r = -0.98). Normalized dose to the small intestine correlated strongly with mid-intestine equivalent diameter (r = -0.97). These strong correlations suggest that patient-specific dose may be estimated for any other child in the same size group who undergoes the abdomen-pelvis scan.

Several experimental studies of the noise power spectrum (NPS) of cone-beam CT have been performed, but less attention has been paid to the analytical derivation of the 3D NPS. It is well known that noise in cone beam CT is nonstationary. When the image is reconstructed by using the FDK algorithm, different back-projection and cosine weightings are used for different reconstruction regions. In addition, the ray density and cone angle are also spatially dependent. As a result, a cone beam system has non-stationary noise. In order to characterize the noise behavior, we first construct the 3D NPS in a local reconstruction volume. Because cone beam rays passing through a small volume can be approximated as parallel rays, the 3D NPS of a small volume can be constructed by considering all effects. The 3D NPS of a large volume (which describes the average noise behavior but may not be valid throughout the volume) was generated by summing 3D NPS of small sub-volumes. The method was validated with computer simulations with uniform noise in the projection. 3D noise spectra of 3cm, 6cm, and 10 cm thick volumes were generated, and the radial NPS at different kz planes were compared with those of the analytically constructed 3D NPS. The results showed excellent matching for all cases. With the proposed method, non-stationary noise behavior of any local volume can be analyzed. The non-stationary noise behavior also causes a high-frequency roll-off in the NPS of a large volume, and as the volume size increases, the roll-off frequency decreases because the larger volume has more heterogeneous noise behavior.

Cone-beam systems designed for breast cancer detection bear a unique radiation dose limitation and are vulnerable to the additive noise from the detector. Additive noise is the signal fluctuation from detector elements and is independent of the incident exposure level. In this study, two different approaches (single pixel based and region of interest based) to measure the additive noise were explored using continuously acquired air images at different exposure levels, with both raw images and flat-field corrected images. The influence from two major factors, inter-pixel variance and image lag, were studied. The pixel variance measured from dark images was used as the gold standard (for the entire detector 15.12±1.3 ADU²) for comparison. Image noise propagation through reconstruction procedures was also investigated and a mathematically derived quadratic relationship between the image noise and the inverse of the radiation dose was confirmed with experiment data. The additive noise level was proved to affect the CT image noise as the second order coefficient and thus determines the lower limit of the scan radiation dose, above which the scanner operates at quantum limited region and utilizes the x-ray photon most efficiently.

We have previously introduced an intensity-weighted region-of-interest (IWROI) technique for image-guided radiation therapy (IGRT), where the outer ROI receives lower exposure than the inner ROI by use of intensity weighting (IW-) filters during the scan. Image noise level, as a result, in the outer ROI will be higher than that in the inner ROI after reconstruction. CBCT images are to be registered to planning CT images and there exist a number of imageregistration algorithms, some of which are relatively more robust against image noise. The IW-filters should be designed so that the outer ROI images have noise levels slightly less than the tolerance of the registiration algorithm resulting in maximum reduction to patient exposure. Therefore, a proper study can guide the filter design based on noise characteristics of the image reconstruction. As a part of the study, we investigated the noise propagation from the projections to the reconstructed image in this work. Also, we obtained the experimental noise characteristics of projections for varying exposure, simulated a pelvis imaging using the NCAT phantom, and performed an image registration of the reconstructed image to the phantom image.

We have investigated the relationship between scan parameters and image quality in fourdimensional cone-beam computed tomography (4D-CBCT) performed with a flat panel imager in image-guided radiotherapy. We have determined upper bounds on scan time while achieving objective thresholds of image quality, namely in noise performance and minimization of view aliasing artifacts. A slow-gantry design for 4D-CBCT was used, in which we slow down clinical linear accelerator gantry speed from the typical 1.0 rpm speed to 0.1 - 0.125 rpm, to ensure the projection angle spacing between two consecutive respiratory cycles is less than 3 degrees. A respiratory monitoring device was used to record the respiratory signal for temporal correlation of the projection data for 4D-CBCT image reconstruction. Four patient data sets were acquired. Reference images were reconstructed with all projection data and were compared with images reconstructed with 50%, 33% and 20% of the projection data. These three partial data reconstructions are simulations of scans with shorter acquisition times. The main image degradations in the short scan simulation image sets are streaking artifacts and poor signal to noise ratio, both caused by sparse projection sampling. The amount of streaking artifacts and SNR in each image set is quantified. By allowing some streaking artifacts and not compromising the assessment of tumor motion, we produce images that suggest that a reduction in scan time from 3 to 6 min to approximately 2 min may be possible, making 4D-CBCT feasible in a clinical setting.

The latest generation of nano computed tomography (nano-CT) systems with sub-micrometer focus X-ray source is expected to yield non-invasive imaging of internal microstructure of objects with isotropic spatial resolution in the range of hundreds nanometers. Most recently commercial systems have become available for purchase. The quantitative characterization of the performance of nano-CT systems is important for evaluating the accuracy of size and density measurements of fine details in nano-CT images. The point spread function (PSF) and modulation transfer function (MTF) are calculated most commonly from the measurement of thin wire phantom for measuring the spatial resolution of clinical CT systems. However, a consistent method for describing the spatial resolution of nano-CT has not been utilized due to the requirement of a nanowire which is a wire of diameter of the order of tens of nanometers. This paper presents a method to characterize the spatial resolution in x/y-scan plane (transversal orientation) of nano-CT systems using a relatively large microwire in the PSF measurement. In this method, the MTF computed from the PSF is estimated on the basis of a two-Gaussian PSF model. Experimenting with microwire images with three different diameter sizes (3μm, 10μm, 30μm) obtained by the synchrotron radiation CT, we demonstrate the potential usefulness of the method for describing the spatial resolutions of nano-CT systems.

Future generations of CT systems would need a mean to cover an entire organ in a single rotation. A way to accomplish this is to physically increase detector size to provide, e.g., 120~160mm z (head-foot) coverage at iso. The x-ray cone angle of such a system is usually 3~4 times of that of a 64-slice (40mm) system, which leads to more severe cone beam artifacts in cardiac scans. In addition, the extreme x-ray take-off angles for such a system cause severe heel effect, which would require an increase in anode target angle to compensate for it. One shortcoming of larger target angle is that tube output likely decreases because of shorter thermal length. This would result in an increase of image noise. Our goal is to understand from a physics and math point of view, what is the clinical acceptable level of artifacts, resolution, and noise impact. The image artifacts are assessed through computer simulation of a helical body phantom and visual comparison of reconstructed images between a 140mm system and a 64-slice system. The IQ impact from target angle increase is studied analytically and experimentally by first finding the proper range of target angles that give the acceptable heel effect, then estimating the impact on peak power (flux) and z resolution using an empirical model of heel effect for given target angle and analytical models of z resolution and tube current loading factor for given target thermal length. The results show that, for a 140mm system, 24.5% of imaging volume exhibits more severe cone beam artifacts than a 64-slice system, which also brings up a patient dose concern. In addition, this system may suffer from a 36% peak power (flux) loss, which is equivalent to about 20% image noise increase. Therefore, a wide coverage CT system using a single x-ray source is likely to face some severe challenges in IQ and clinical accuracy.

The tomographic reconstruction of the beating heart requires dedicated methods. One possibility is gated reconstruction, where only data corresponding to a certain motion state are incorporated. Another one is motioncompensated reconstruction with a pre-computed motion vector field, which requires a preceding estimation of the motion. Here, results of a new approach are presented: simultaneous reconstruction of a three-dimensional object and its motion over time, yielding a fully four-dimensional representation. The object motion is modeled by a time-dependent elastic transformation. The reconstruction is carried out with an iterative gradient-descent algorithm which simultaneously optimizes the three-dimensional image and the motion parameters. The method was tested on a simulated rotational X-ray acquisition of a dynamic coronary artery phantom, acquired on a C-arm system with a slowly rotating C-arm. Accurate reconstruction of both absorption coefficient and motion could be achieved. First results from experiments on clinical rotational X-ray coronary angiography data are shown. The resulting reconstructions enable the analysis of both static properties, such as vessel geometry and cross-sectional areas, and dynamic properties, like magnitude, speed, and synchrony of motion during the cardiac cycle.

Purpose: To achieve three dimensional isotropic dynamic cardiac CT imaging with high temporal resolution for evaluation of cardiac function with a slowly rotating C-arm system. Method and Materials: A recently introduced extension to compressed sensing, viz. Prior Image Constrained Compressed Sensing (PICCS), in which a prior image is used as a constraint in the reconstruction has enabled this application. An in-vivo animal experiment (e.g. a beagle model) was conducted using an interventional C-arm system. The imaging protocol was as follows: contrast was injected, the contrast equilibrated, breathing was suspended for ~14 seconds during which time 420 equally spaced projections were acquired. This data set was used to reconstruct a fully sampled blurred image volume using the conventional FDK algorithm (e.g. the prior image). Then the data set was retrospectively gated into 19 phases according to the recorded ECG signal (heart rate ~ 95bpm) and images were reconstructed with the PICCS algorithm. Results: Cardiac MR was used as the gold standard due to its high temporal resolution. The same short-axis slice was selected from the PICCS-CT data set and the MR data set. Manual contouring on the peak systolic and peak diastolic frames was performed to assess the ejection fraction contribution from this single plane. The calculated ejection fractions with PICCS-CT agreed well with the MR results. Conclusion: We have demonstrated the ability to use a slowly rotating interventional C-arm system in order to make measurements of cardiac function. The new technique provides high isotropic spatial resolution (~0.5 mm) along with high temporal resolution (~ 33 ms). The evaluation of cardiac function demonstrated a great agreement with single slice cardiac MR.

We present new acquisition modes of a recently introduced dual-source computed tomography (DSCT) system equipped with two X-ray tubes and two corresponding detectors, mounted onto the rotating gantry with an angular offset of typically 90°. Due to the simultaneous acquisition of complementary data, the minimum exposure time is reduced by a factor of two compared to a single-source CT system (SSCT). The correspondingly improved temporal resolution is beneficial for cardiac CT. Also, maximum table feed per rotation in a spiral mode can be increased by a factor of 2 compared to SSCT, which provides benefits both for cardiac CT and non-cardiac CT. In an ECG-triggered mode the entire cardiac volume can be scanned within a fraction of one cardiac RR-cycle. At a rotation time of 0.28s using a detector with 64×0.6 mm beam collimation, the scan time of the entire heart is less than 0.3s at a temporal resolution of 75 ms. It will be shown, that the extremely fast cardiac scan reduces the patient dose to a theoretical lowest limit: for a 120 kV scan the dose level for a typical cardiac CT scan is well below 2 mSv. Using further protocol optimization (scan range adaptation, 100kV), the radiation dose can be reduced below 1mSv.

This study evaluates new applications using a novel navigation system with electromagnetic (EM) tracking in clinical routine. The navigation system (iGuide CAPPA, CAS innovations, Erlangen, Germany) consists of a PC with dedicated navigation software, the AURORA tracking system (NDI, Waterloo Ontario, Canada) and needles equipped with small coils in their tips for EM navigation. After patient positioning a 3D C-arm data set of the spine region of interest is acquired. The images are reconstructed and the 3D data set is directly transferred to the navigation system. Image loading and image to patient registration are performed automatically by the navigation system. For image acquisition a C-arm system with DynaCT option (AXIOM Artis, Siemens Healthcare, Forchheim, Germany) was used. As new clinical applications we performed kyphoplasty for reconstruction of collapsed vertebrae. All interventions were carried out without any complication. After a single planning scan the radiologists were able to place the needle in the designated vertebra. During needle driving 2D imaging was performed just in a few cases for control reasons. The time between planning and final needle positioning was reduced in all cases compared to conventional methods. Moreover, the number of control scans could be markedly reduced. The deviation of the needle to the planned target was less than 2 mm. The use of DynaCT images in combination with electromagnetic tracking-based navigation systems allows a precise needle positioning for kyphoplasty.

C-arm systems may be used as front ends for cone-beam CT. The resulting image quality is affected by several factors, including the source trajectory, the reconstruction algorithm, and the accuracy of the data. The standard source trajectory is a circular arc spanning a little more than 180 degrees. However, since a planar source trajectory satisfies Tuy's completeness condition only within a subset of the source plane, the resulting images are bound to exhibit "cone-beam artifacts" off the source plane. The cure consists in using a source trajectory that satisfies Tuy's completeness condition everywhere within the volume of interest. Such a source trajectory must be non-planar. To keep the scan time short, the source trajectory should also consist of a single, smooth segment. A favorable source trajectory of this kind is a curve known as spherical spiral. We implemented a spherical spiral on a laboratory C-arm system, along with a standard circular arc. An anthropomorphic head phantom was scanned using both source trajectories and otherwise identical scan parameters. Images were reconstructed using a short scan version of the FDK algorithm (circular arc) and the cone-beam Fourierfiltered backprojection (CBFFBP) algorithm presented earlier. Images obtained with the circular arc showed cone-beam artifacts. Images obtained with the spherical spiral did not. The results also demonstrate the good performance of the CBFFBP algorithm.

Linear discriminate analysis (LDA) is applied to dual kVp CT and used for tissue characterization. The potential to quantitatively model both malignant and benign, hypo-intense liver lesions is evaluated by analysis of portal-phase, intravenous CT scan data obtained on human patients. Masses with an a priori classification are mapped to a distribution of points in basis material space. The degree of localization of tissue types in the material basis space is related to both quantum noise and real compositional differences. The density maps are analyzed with LDA and studied with system simulations to differentiate these factors. The discriminant analysis is formulated so as to incorporate the known statistical properties of the data. Effective kVp separation and mAs relates to precision of tissue localization. Bias in the material position is related to the degree of X-ray scatter and partial-volume effect. Experimental data and simulations demonstrate that for single energy (HU) imaging or image-based decomposition pixel values of water-like tissues depend on proximity to other iodine-filled bodies. Beam-hardening errors cause a shift in image value on the scale of that difference sought between in cancerous and cystic lessons. In contrast, projection-based decomposition or its equivalent when implemented on a carefully calibrated system can provide accurate data. On such a system, LDA may provide novel quantitative capabilities for tissue characterization in dual energy CT.

We create a series of detailed computerized phantoms to estimate patient organ and effective dose in pediatric CT and investigate techniques for efficiently creating patient-specific phantoms based on imaging data. The initial anatomy of each phantom was previously developed based on manual segmentation of pediatric CT data. Each phantom was extended to include a more detailed anatomy based on morphing an existing adult phantom in our laboratory to match the framework (based on segmentation) defined for the target pediatric model. By morphing a template anatomy to match the patient data in the LDDMM framework, it was possible to create a patient specific phantom with many anatomical structures, some not visible in the CT data. The adult models contain thousands of defined structures that were transformed to define them in each pediatric anatomy. The accuracy of this method, under different conditions, was tested using a known voxelized phantom as the target. Errors were measured in terms of a distance map between the predicted organ surfaces and the known ones. We also compared calculated dose measurements to see the effect of different magnitudes of errors in morphing. Despite some variations in organ geometry, dose measurements from morphing predictions were found to agree with those calculated from the voxelized phantom thus demonstrating the feasibility of our methods.

We perform simulation studies of proposed square and hexagonal geometries of a multi-source X-ray micro-computed tomography (CT) system. The system uses linear arrays of the carbon nano-tube (CNT)-based X-ray sources which are individually addressable. In the square geometry, two linear source arrays and two area detectors form a square; whereas in the hexagonal geometry, three linear source arrays and three area detectors form a hexagon. The tomographic angular sampling for both geometries requires no motion of the sources or subject. Based on the sinogram maps, the hexagonal geometry has improved angular coverage than the square geometry. The ordered-subset convex iterative algorithm is implemented in both geometries for reconstructions from cone-beam projection data. The reconstructed images from both geometries are generally consistent with the phantom, although some streaking artifacts due to the limited-angle nature of the geometries are observed. The two geometries show similar performance in resolution-noise tradeoff. The gap-free hexagonal geometry produces lower mean squared error in the reconstructed images; when gaps between the source arrays and detectors are introduced, the angular coverage of the hexagonal geometry degrades faster and becomes worse than the square geometry. The impact of gaps on the imaging properties must be studied further.

A number of groups are currently investigating tomographic imaging of the breast, but the optimal design and acquisition parameters for such systems remains uncertain. One useful tool for investigating optimal parameters is computer simulation software. A computer program that simulates xray transport through a breast object model followed by signal and noise propagation through a CsI flatpanel detector has been modified, restructured and enhanced in order to provide a fast yet sufficiently accurate research tool. The main focus of this work was to validate the simulated response of a CsI flatpanel detector with a real detector namely, the Paxscan 2520 (Varian Medical Systems, Salt Lake City, UT). Preliminary results indicate that the program provides comparable quantitative accuracy, that can be used to obtain accurate and meaningful results to assist in research in tomosynthesis and CT breast imaging system design.

Images of mastectomy breast specimens have been acquired with a bench top experimental Cone beam CT (CBCT) system. The resulting images have been segmented to model an uncompressed breast for simulation of various CBCT techniques. To further simulate conventional or tomosynthesis mammographic imaging for comparison with the CBCT technique, a deformation technique was developed to convert the CT data for an uncompressed breast to a compressed breast without altering the breast volume or regional breast density. With this technique, 3D breast deformation is separated into two 2D deformations in coronal and axial views. To preserve the total breast volume and regional tissue composition, each 2D deformation step was achieved by altering the square pixels into rectangular ones with the pixel areas unchanged and resampling with the original square pixels using bilinear interpolation. The compression was modeled by first stretching the breast in the superior-inferior direction in the coronal view. The image data were first deformed by distorting the voxels with a uniform distortion ratio. These deformed data were then deformed again using distortion ratios varying with the breast thickness and re-sampled. The deformation procedures were applied in the axial view to stretch the breast in the chest wall to nipple direction while shrinking it in the mediolateral to lateral direction re-sampled and converted into data for uniform cubic voxels. Threshold segmentation was applied to the final deformed image data to obtain the 3D compressed breast model. Our results show that the original segmented CBCT image data were successfully converted into those for a compressed breast with the same volume and regional density preserved. Using this compressed breast model, conventional and tomosynthesis mammograms were simulated for comparison with CBCT.

Breast density has been recognized as one of the major risk factors for breast cancer. However, breast density is currently estimated using mammograms which are intrinsically 2D in nature and cannot accurately represent the real breast anatomy. In this study, a novel technique for measuring breast density based on the segmentation of 3D cone beam CT (CBCT) images was developed and the results were compared to those obtained from 2D digital mammograms. 16 mastectomy breast specimens were imaged with a bench top flat-panel based CBCT system. The reconstructed 3D CT images were corrected for the cupping artifacts and then filtered to reduce the noise level, followed by using threshold-based segmentation to separate the dense tissue from the adipose tissue. For each breast specimen, volumes of the dense tissue structures and the entire breast were computed and used to calculate the volumetric breast density. BI-RADS categories were derived from the measured breast densities and compared with those estimated from conventional digital mammograms. The results show that in 10 of 16 cases the BI-RADS categories derived from the CBCT images were lower than those derived from the mammograms by one category. Thus, breasts considered as dense in mammographic examinations may not be considered as dense with the CBCT images. This result indicates that the relation between breast cancer risk and true (volumetric) breast density needs to be further investigated.

In this work, we investigated the visibility of microcalcifications in CCD-based cone beam CT (CBCT) breast imaging. A paraffin cylinder with a diameter of 135 mm and a thickness of 40 mm was used to simulate a 100% adipose breast. Calcium carbonate grains, ranging from 140-150 to 200-212 μm in size, were used to simulate the microcalcifications. Groups of 25 same size microcalcifications were arranged into 5 × 5 clusters. Each cluster was embedded at the center of a smaller (15 mm diameter) cylindrical paraffin phantom, which were inserted into a hole at the center of the breast phantom. The breast phantom with the simulated microcalcifications was scanned on a bench top experimental CCDbased cone beam CT system at various exposure levels with two CCD cameras: Hamamatsu's C4742-56-12ER and Dalsa 99-66-0000-00. 300 projection images were acquired over 360° and reconstructed with Feldkamp's backprojection algorithm using a ramp filter. The images were reviewed by 6 readers independently. The ratios of visible microcalcifications were recorded and averaged over all readers. These ratios were plotted as the function of measured image signal-to-noise ratio (SNR) for various scans. It was found that 94% visibility was achieved for 200-212 μm calcifications at an SNR of 48.2 while 50% visibility was achieved for 200-212, 180-200, 160-180, 150-160 and 140-150 μm calcifications at an SNR of 25.0, 35.3, 38.2, 42.2 and 64.4, respectively.

Digital breast tomosynthesis (DBT) shows potential for improving breast cancer detection. However, this technique has not yet been fully characterized with consideration of the various uncertainties in the imaging chain and optimized with respect to system acquisition parameters. To obtain maximum diagnostic information in DBT, system optimization needs to be performed across a range of patients and acquisition parameters to quantify their impact on tumor detection performance. In addition, a balance must be achieved between x-ray dose and image quality to minimize risk to the patient while maximizing the system's detection performance. To date, researchers have applied a task-based approach to the optimization of DBT with use of mathematical observers for tasks in the signal-known-exactly background-known-exactly (SKE/BKE) and signal-known-exactly background-known statistically (SKE/BKS) paradigms1-3. However, previous observer models provided insufficient treatment of the spatial correlations between multi-angle DBT projections, so we incorporated this correlation information into the modeling methodology. We developed a computational approach that includes three-dimensional variable background phantoms for incorporating background variability, accurate ray-tracing and Poisson distributions for generating noise-free and noisy projections of the phantoms, and a channelized-Hotelling observer4 (CHO) for estimating performance in DBT. We demonstrated our method for a DBT acquisition geometry and calculated the performance of the CHO with Laguerre-Gauss channels as a function of the angular span of the system. Preliminary results indicate that the implementation of a CHO model that incorporates correlations between multi-angle projections gives different performance predictions than a CHO model that ignores multi-angle correlations. With improvement of the observer design, we anticipate more accurate investigations into the impact of multi-angle correlations and background variability on the performance of DBT.

While diagnostic improvement via breast tomosynthesis has been notable, the full potential of tomosynthesis has not yet been realized. This is because of the complex task of optimizing multiple parameters that constitute image acquisition and thus affect tomosynthesis performance. Those parameters include dose, number of angular projections, and the total angular span of those projections. In this study, we investigated the effects of acquisition parameters, independent of each other, on the overall diagnostic image quality of tomosynthesis. Five mastectomy specimens were imaged using a prototype tomosynthesis system. 25 angular projections of each specimen were acquired at 6.2 times typical single-view mammographic dose level. Images at lower dose levels were then simulated using a noise modification routine. Each projection image was supplemented with 84 simulated 3 mm 3D lesions embedded at the center of 84 non-overlapping ROIs. The projection images were then reconstructed using a filtered-back projection (FBP) algorithm at 224 different combinations of acquisition parameters to investigate which one of the many possible combinations maximized performance. Performance was evaluated in terms of a Laguerre-Gauss channelized Hotelling observer model-based measure of lesion detectability. Results showed that performance improved with an increase in the total acquisition dose level and the angular span. At a constant dose level and angular span, the performance rolled-off beyond a certain number of projections, indicating that simply increasing the number of projections in tomosynthesis may not necessarily improve its performance. The best performance was obtained with 15-17 projections spanning an angular arc of ~45° - the maximum tested in our study, and for an acquisition dose equal to single-view mammography. The optimization framework developed in this framework is applicable to other reconstruction techniques and other multi-projection systems.

Digital breast tomosynthesis (DBT) has been shown to decrease breast structural noise thus improving the detection of masses. However decreased detectability of microcalcifications was observed, and several studies have been performed to investigate the benefit of taking an additional central projection view after a DBT scan. Our study investigates the effect of variable angular dose distribution within a single DBT scan. Using a prototype DBT system with uniform angular dose distribution, several DBT scans (25 projection views over 40 degree angular range) were performed using different glandular doses. A subset of projection images was selected from each scan to form composite DBT scans (25 views each) with different angular dose distribution schemes (ADS). Two examples of ADS were: 1) seven central views with four times the dose of periphery; and 2) five central projections with six times the dose of periphery. The total dose for each ADS was the same as a reference scan with uniform dose distribution (1.5 mGy). They also had the same number of views and angular range, and were reconstructed using identical reconstruction filter settings. The detectability of calcifications, the 3D MTF and NPS of the system, and the ideal observer object detectability index for all cases were compared. The results showed that higher dose for the central views improve the detectability of calcifications. However magnitude of improvement depends on the reconstruction method and the size of the object.

Digital breast tomosynthesis (DBT) is a limited-view, limited-angle computed tomography (CT) technique that has the potential to yield improved lesion conspicuity over that of standard digital mammography. To maintain short acquisition time, the detector must have a rapid temporal response. Transient effects like lag and ghosting have been noted previously in digital mammography systems, but for the times between successive views (approx. 1 minute), their impact on image quality is generally negligible. However, tomosynthesis imaging requires much shorter times between projection images (< 1 s). Under these conditions, detectors that may have been acceptable for digital mammography may not be suitable for tomosynthesis. Transient effects will generally cause both a loss of signal and an increase in image noise. A cascaded systems analysis is used to determine the effect of lag on image quality in a DBT system. It is shown that in the projection images, lag results in artifacts appearing as a "trail" of prior exposures. The effect of lag on image quality is also evaluated with a simple Monte Carlo simulation of a cone-beam tomosynthesis image formation incorporating a filtered back-projection algorithm.

We previously proposed a three-dimensional computerized breast phantom that combines empirical data with the flexibility of mathematical models¹. The goal of this project is to enhance the breast phantom to include a more detailed anatomy than currently visible and create additional phantoms from different breast CT data. To improve the level of detail in our existing segmentations, the breast CT data is reconstructed at a higher resolution and additional image processing techniques are used to correct for noise and scatter in the image data. A refined segmentation algorithm is used that incorporates more detail than previously defined. To further enhance high-resolution detail, mathematical models, implementing branching algorithms to extend the glandular tissue throughout the breast and to define Cooper's ligaments, are under investigation. We perform the simulation of mammography and tomosynthesis using an analytical projection algorithm that can be applied directly to the mathematical model of the breast without voxelization². This method speeds up image acquisition, reduces voxelization artifacts, and produces higher resolution images than the previously used method. The realistic 3D computerized breast phantom will ultimately be incorporated into the 4DXCAT phantom^3-5 to be used for breast imaging research.

A major problem of high-resolution positron-emission-tomography (PET) are subject movements during acquisition. We propose a new motion compensation algorithm called "Blind Motion-Compensated Reconstruction" (BMCR) that is able to deal with frames of extremely low statistics in the case of smooth motion. Our algorithm reconstructs both image and rigid motion just from the recorded data and does not need any external motion tracking. This is achieved by combining image reconstruction and motion compensation into one mathematical framework which consists of a cost functional and an optimization method. The cost functional basically consists of a difference term which ensures consistency of the estimated parameters to the model and some regularization terms which render the problem mathematically well-posed. The optimization method aims at finding a pair of image and transformation/motion such that the cost functional is minimal. Such a combined framework can overcome problems of existing algorithms which separate reconstruction and motion compensation. These algorithms usually try to get the motion information by registering reconstructed frames one to each other (in image space). Their main drawback is that the registration step is likely to be of low accuracy or even fail completely for low-statistics frames. A quantitative and visual comparison suggests that BMCR is superior to state-of-the-art intrinsic methods.

Motion artifacts are a significant issue in medical image reconstruction. There are many methods for incorporating motion information into image reconstruction. However, there are fewer studies that focus on deformation regularization in motioncompensated image reconstruction. The usual choice for deformation regularization has been penalty functions based on the assumption that tissues are elastic. In the image registration field, there have been some methods proposed that impose deformation invertibility using constraints or regularization, assuming that organ motions are invertible transformations. However, most of these methods require very high memory or computation complexity, making them poorly suited for dealing with multiple images simultaneously in motion-compensated image reconstruction. Recently we proposed an image registration method that uses a simple penalty function based on a sufficient condition for the local invertibility of deformations.¹ That approach encourages local invertibility in a fast and memory-efficient way. This paper investigates the use of that regularization method for the more challenging problem of joint image reconstruction and nonrigid motion estimation. A 2D PET simulation (based on realistic motion from real patient CT data) demonstrates the benefits of such motion regularization for joint image reconstruction/registration.

Molecular imaging using dynamic positron emission tomography (PET) can provide in vivo images of physiologically or biochemically important parameters. Direct reconstruction of parametric images from dynamic PET sinograms is statistically more efficient than the conventional indirect methods, which perform image reconstruction and kinetic modeling in two separate steps. Most existing direct reconstruction methods are derived based on a known blood input function. This paper presents a direct reconstruction algorithm using a simplified reference tissue model, which does not require a blood input function. We have derived a minorization-maximization algorithm to find the penalized maximum likelihood solution. Computer simulations show that the proposed method has better bias-variance tradeoff than the conventional indirect method for estimating parametric images of receptor binding potential using dynamic PET.

Today, attenuation corrected SPECT, typically performed using CT or Gadolinium line source based transmission scans, is more and more becoming standard in many medical applications. Moreover, the information about the material density distribution provided by these scans is key for other artifact compensation approaches in advanced SPECT reconstruction. Major drawbacks of these approaches are the additional patient radiation and hardware/maintenance costs as well as the additional workflow effort, e.g. if the CT scans are not performed on a hybrid scanner. It has been investigated in the past, whether it is possible to recover this structural information solely from the SPECT scan data. However, the investigated methods often result in noticeable image artifacts due to cross-dependences between attenuation and activity distribution estimation. With the simultaneous reconstruction method presented in this paper, we aim to effectively prevent these typical cross-talk artifacts using a-priori known atlas information of a human body. At first, an initial 3D shape model is coarsely registered to the SPECT data using anatomical landmarks and each organ structure within the model is identified with its typical attenuation coefficient. During the iterative reconstruction based on a modified ML-EM scheme, the algorithm simultaneously adapts both, the local activity estimation and the 3D shape model in order to improve the overall consistency between measured and estimated sinogram data. By explicitly avoiding topology modifications resulting in a non-anatomical state, we ensure that the estimated attenuation map remains realistic. Several tests with simulated as well as real patient SPECT data were performed to test the proposed algorithm, which demonstrated reliable convergence behaviour in both cases. Comparing the achieved results with available reference data, an overall good agreement for both cold as well as hot activity regions could be observed (mean deviation: -5.98%).

Imaging in general is becoming increasingly important in the medical science. At the cell level it is possible to label and trace almost individual molecules in vivo to study biochemical reactions using microscopy. In vivo imaging of living organisms is today mainly accomplished by PET, SPECT and fMRI. The problem is that the spatial resolution for realistic image acquisition times is of the order 1-3 mm, which is a serious limitation. We propose a new imaging modality, based on the same principles as SPECT but with drastically improved efficiency and spatial resolution. This is achieved by incorporating a large number of x-ray lenses between the detectors and the object. In current SPECT a pin-hole geometry is standard, involving an unfortunate trade-off between efficiency and spatial resolution, our solution would change this. The agent for radiolabelling is assumed to be ¹²⁵I, with an emission peak at 27 keV, since it is widely used and easy to handle. The large area, photon counting detectors will consist of a columnar CsI scintillator coupled to a CMOS integrated circuit for electronic read-out. Our simulations of the entire system and of the detector indicate that a resolution of 50 μm for the system is possible.

The behaviour of spins at a classical level in the presence of an external magnetic field is completely described by the Bloch equation. This is the main equation governing the magnetic resonance imaging (MRI) system, and as such the Bloch equation is extensively used for design purposes. Here we have transferred the Bloch equation to the spherical coordinate system that, to the best of our knowledge, has not previously been applied in this field. The mathematical framework and the simulation results show that this fresh view of the Bloch equation provides better insight into the magnetic resonance (MR) phenomenon. In this mathematical framework, without using spinor space, the order of the Bloch equation is reduced in a much simpler way and can therefore provides a novel insight to the slice selection problem. Simulation results are presented for a variety of slice selective pulses, with and without post excitation rephasing gradients. In this paper nonlinear gradient is tried as well which shows an improvement in uniformity of the selected slice. It is feasible to find an analytic approximate solution to the Bloch equation in spherical coordinate system by adopting averaging techniques over spatial variables available available in nonlinear dynamical systems. We anticipate that our new description of the MR phenomenon will allow researchers to revisit the excitation pattern design question to achieve better slice selectivity or to find an optimal excitation pattern from a theoretical basis. This has the potential to result in practical improvements affecting all forms of MR imaging.

Parallel MRI can achieve increased spatiotemporal resolution in MRI by simultaneously sampling reduced k-space data with multiple receiver coils. One requirement that different parallel MRI techniques have in common is the need to determine spatial sensitivity information for the coil array. This is often done by smoothing the raw sensitivities obtained from low-resolution calibration images, for example via polynomial fitting. However, this sensitivity post-processing can be both time-consuming and error-prone. Another important factor in Parallel MRI is noise amplification in the reconstruction, which is due to non-unity transformations in the image reconstruction associated with spatially correlated coil sensitivity profiles. Generally, regularization approaches, such as Tikhonov and SVD-based methods, are applied to reduce SNR loss, at the price of introducing residual aliasing. In this work, we present a regularization approach using in vivo coil sensitivities in parallel MRI to overcome these potential errors into the reconstruction. The mathematical background of the proposed method is explained, and the technique is demonstrated with phantom images. The effectiveness of the proposed method is then illustrated clinically in a whole-heart 3D cardiac MR acquisition within a single breath-hold. The proposed method can not only overcome the sensitivity calibration problem, but also suppress a substantial portion of reconstruction-related noise without noticeable introduction of residual aliasing artifacts.

To understand the role of the tongue in speech production, it is desirable to directly image the motion and strain of the muscles within the tongue. Magnetic resonance tagging-which was originally developed for cardiac imaging-has previously been applied to image both two-dimensional and three-dimensional tongue motion during speech. However, to quantify three-dimensional motion and strain, multiple images yielding two-dimensional motion must be acquired at different orientations and then interpolated - a time-consuming task both in image acquisition and processing. Recently, a new MR imaging and image processing method called zHARP was developed to encode and track 3D motion from a single slice without increasing acquisition time. zHARP was originally developed and applied to cardiac imaging. The application of zHARP to the tongue is not straightforward because the tongue in repetitive speech does not move as consistently as the heart in its beating cycle. Therefore tongue images are more susceptible to motion artifacts. Moreover, these artifacts are greatly exaggerated as compared to conventional tagging because of the nature of zHARP acquisition. In this work, we re-implemented the zHARP imaging sequence and optimized it for the tongue motion analysis. We also optimized image acquisition by designing and developing a specialized MRI scanner triggering method and vocal repetition to better synchronize speech repetitions. Our method was validated using a moving phantom. Results of 3D motion tracking and strain analysis on the tongue experiments demonstrate the effectiveness of this method.

Neutron stimulated emission computed tomography (NSECT) is being developed as a non-invasive technique to diagnose iron overload in the liver. It uses inelastic scatter interactions between fast neutrons and iron nuclei to quantify localized distributions of iron within the liver. Preliminary studies have demonstrated the feasibility of iron overload detection through NSECT using a Monte-Carlo simulation model in GEANT4. The work described here uses the GEANT4 simulation model to analyze iron-overload detection sensitivity in NSECT. A simulation of a clinical NSECT system was designed in GEANT4. Simulated models were created for human liver phantoms with concentrations of iron varying from 0.5 mg/g to 20 mg/g (wet). Each liver phantom was scanned with 100 million neutron events to generate gamma spectra showing gamma-lines corresponding to iron in the liver. A background spectrum was obtained using a water phantom of equal mass as the liver phantom and was subtracted from each liver spectrum. The height of the gamma line at 847 keV (corresponding to ⁵⁶Fe) was used as a measure of the detected iron concentration in each background-corrected spectrum. The variation in detected gamma counts was analyzed and plotted as a function of the liver iron concentration to quantify measurement error. Analysis of the differences between the measured and expected value of iron concentration indicate that NSECT sensitivity for detection of iron in liver tissue may lie in the range of 0.5 mg/g - 1 mg/g, which represents a clinically significant range for iron overload detection in humans.

We describe an improved optoacoustic tomography method, that utilizes a diffusion-based photon propagation model in order to obtain quantified reconstruction of targets embedded deep in heterogeneous scattering and absorbing tissue. For the correction we utilize an iterative finite-element solution of the light diffusion equation to build a photon propagation model. We demonstrate image improvements achieved by this method by using tissue-mimicking phantom measurements. The particular strength of the method is its ability to achieve quantified reconstructions in non-uniform illumination configurations resembling whole-body small animal imaging scenarios.

Various medical imaging techniques exist to detect the early development of tissue damage. However, a widely commercialized device that can be easily used and is cost effective is still needed. Through a literature review, we examined ultrasound, microwave tomography, and ultra-wideband (UWB) technology. Out of these techniques, UWB is the most promising since it has the capability to detect small adjustments in dielectric properties, which can change with minor alterations in perfusion and internal pressure. These minor alterations are vital in detecting the onset of ischemia, which precedes many serious conditions affecting tissue health. In addition to its ability in detection, UWB also has the potential to become a widely accessible technology to hospitals. Using software called XFdtd, we simulated ultrawideband pulses propagating through planes designed to resemble tissue in its dielectric properties. After testing several sizes of the horn antenna and configurations for the wire and port, the antenna's near field was finally able to reach the distance necessary to penetrate the tissue model. The resulting graph of voltage versus time was generated from the received antenna signal and it will be compared to the graphs that result after the dielectric properties of the model have been changed to simulate tissue injury. Through this manipulation of the tissue model, the sensitivity and selectivity of UWB in measuring small fluctuations in perfusion can be determined. In this future work with XFdtd, we want to show that UWB is a novel and viable technique in detecting early tissue injury.

The near-ubiquity of large area, active matrix, flat-panel imagers (AMFPIs) in medical x-ray imaging applications is a testament to the usefulness and adaptability of the relatively simple concept of array pixels based on a single amorphous silicon (a-Si:H) TFT coupled to a pixel storage capacitor. Interestingly, the fundamental advantages of a-Si:H thin film electronics (including compatibility with very large area processing, high radiation damage resistance, and continued development driven by interest in mainstream consumer products) are shared by the rapidly advancing technology of polycrystalline silicon (poly-Si) TFTs. Moreover, the far higher mobilities of poly-Si TFTs, compared to those of a- Si:H, facilitate the creation of faster and more complex circuits than are possible with a-Si:H TFTs, leading to the possibility of new classes of large area, flat panel imagers. Given recent progress in the development of initial poly-Si imager prototypes, the creation of increasingly sophisticated active pixel arrays offering pixel-level amplification, variable gain, very high frame rates, and excellent signal-to-noise performance under all fluoroscopic and radiographic conditions (including very low exposures and high spatial frequencies), appears within reach. In addition, it is conceivable that the properties of poly-Si TFTs could allow the development of large area imagers providing single xray photon counting capabilities. In this article, the factors driving the possible realization of clinically practical active pixel and photon counting imagers based on poly-Si TFTs are described and simple calculational estimates related to photon counting imagers are presented. Finally, the prospect for future development of such imagers is discussed.

A dual mode current-programmed, current-output active pixel sensor (DCAPS) in amorphous silicon (a-Si:H) technology is introduced for digital X-ray imaging, and in particular, for hybrid fluoroscopic and radiographic imagers. Here, each pixel includes an extra capacitor that selectively is coupled to the pixel capacitance to realize the dual mode behavior. Pixel structure, operation and characteristics are presented. The proposed DCAPS circuit was fabricated and assembled using an in-house bottom gate inverted staggered a-Si:H thin film transistor (TFT) process. Gain, lifetime, transient performance as well as noise analysis were carried out. The results are promising and demonstrate that the DCAPS enables dual mode X-ray imaging while compensating for the long term electrical and thermal stress related a-Si TFT threshold voltage (V_t) shift.

Incomplete charge collection due to poor electron mobility in amorphous selenium (a-Se) results in depth-dependent signal variations. The slow signal rise-time for the portion of the induced charge due to electron-movement towards the anode and significant electron trapping cause ballistic deficit. In this paper, we investigate Frisch-grid detector design to reduce the depth dependent noise, increase the photon count-rate, and improve the spectral performance of positively biased amorphous selenium radiation detectors. In addition, we analyze the impact of using the Frisch grid detector design on x-ray sensitivity, detective quantum efficiency (DQE), modulation transfer function (MTF), and image lag of integrating-mode a-Se radiation detectors. Preliminary results based on theory are presented for emerging digital medical imaging modalities such as mammography tomosynthesis and fluoroscopy.

The SSXII is a novel x-ray imager designed to improve upon the performance limitations of conventional dynamic radiographic/fluoroscopic imagers related to resolution, charge-trapping, frame-rate, and instrumentation-noise. The SSXII consists of a CsI:Tl phosphor coupled via a fiber-optic taper (FOT) to an electron-multiplying CCD (EMCCD). To facilitate investigational studies, initial designs enable interchangeability of such imaging components. Measurements of various component and configuration characteristics enable an optimization analysis with respect to overall detector performance. Photon transfer was used to characterize the EMCCD performance including ADC sensitivity, read-noise, full-well capacity and quantum efficiency. X-ray sensitivity was measured using RQA x-ray spectra. Imaging components were analyzed in terms of their MTF and transmission efficiency. The EMCCD was measured to have a very low effective read-noise of less than 1 electron rms at modest EMCCD gains, which is more than two orders-ofmagnitude less than flat panel (FPD) and CMOS-based detectors. The variable signal amplification from 1 to 2000 times enables selectable sensitivities ranging from 8.5 (168) to over 15k (300k) electrons per incident x-ray photon with (without) a 4:1 FOT; these sensitivities could be readily increased with further component optimization. MTF and DQE measurements indicate the SSXII performance is comparable to current state-of-the-art detectors at low spatial frequencies and far exceeds them at higher spatial frequencies. The instrumentation noise equivalent exposure (INEE) was measured to be less than 0.3 μR out to 10 cycles/mm, which is substantially better than FPDs. Component analysis suggests that these improvements can be substantially increased with further detector optimization.

In addition to causing loss of contrast and blurring in an image, scatter also makes quantitative measurements of xray attenuation impossible. Many devices, methods, and models have been developed to eliminate, estimate, and correct for the effects of scatter. Although these techniques can reduce the impact of scatter in a large-area image, no methods have proven to be practical and sufficient to enable quantitative analysis of image data in a routine clinical setting. This paper describes a method of scatter correction which uses moderate x-ray collimation in combination with a correction algorithm operating on data obtained from large-area flat panel detector images. The method involves acquiring slot collimated images of the object, and utilizing information from outside of the collimated region, in addition to a priori data, to estimate the scatter within the collimated region. This method requires no increase dose to the patient while providing high image quality and accurate estimates of the primary x-ray data. This scatter correction technique was validated through beam stop experiments and comparison of theoretically calculated and measured contrast of thin aluminum and polymethylmethacrelate objects. Measurements taken with various background material thicknesses, both with and without a grid, showed that the slot-scatter corrected contrast and the theoretical contrast were not significantly different given a 99% confidence interval. However, the uncorrected contrast was found to be significantly different from the corrected and theoretical contrasts. These findings indicate that this method of scatter correction can eliminate the effect of scatter on contrast and potentially enable quantitative x-ray imaging.

For task specific evaluation of imaging systems it is necessary to obtain detailed descriptions of their noise and deterministic properties. In the past we have developed an experimental and theoretical methodology to estimate the deterministic detector response of a digital x-ray imaging system, also known as the H matrix. In this paper we have developed the experimental methodology for the evaluation of the quantum and electronic noise of digital radiographic detectors using the covariance matrix K. Using the H matrix we calculated the transfer of a simulated coronary artery constriction through an imaging system's detector, and with the covariance matrix we calculated the detectability (or Signal-to-Noise Ratio) and the detection probability. The eigenvalues and eigenvectors of the covariance matrix were presented and the electronic and quantum noise were analyzed. We found that the exposure at which the electronic noise equals the quantum noise at 90 kVp was 0.2 μR. We compared the ideal Hotelling observer with the Fourier definition of the SNR for a toroidal stenosis on a cylindrical vessel. Because of the shift-invariance and cyclo-stationarity assumptions, the Fourier SNR overestimates the performance of imaging systems. This methodology can be used for task specific evaluation and optimization of a digital x-ray imaging system.

In digital radiography, conventional DQE evaluations are performed under idealized conditions that do not reflect typical clinical operating conditions. For this reason, we have developed and evaluated an experimental methodology for measuring theeffective detective quantum efficiency (eDQE) of digital radiographic systems and its utility in chest imaging applications.To emulate the attenuation and scatter properties of the human thorax across a range of sizes, the study employed pediatric and adult geometric chest imaging phantoms designed for use in the FDA/CDRH Nationwide Evaluation of X-Ray Trends (NEXT) program and a third phantom configuration designed to represent the bariatric population. The MTF for each phantom configuration was measured using images of an opaque edge device placed at the nominal surface of each phantom and at a common reference point. For each phantom, the NNPS was measured in a uniform region within the phantom image acquired at an exposure level determined from a prior phototimed acquisition. Scatter measurements were made using a beam-stop technique. These quantities were used along with measures of phantom attenuation and estimates of x-ray flux, to compute the eDQE at the beam-entrance surface of the phantoms, reflecting the presence of scatter, grid, magnification, and focal spot blur. The MTF results showed notable degradation due to focal spot blurring enhanced by geometric magnification, with increasing phantom size. Measured scatter fractions were 33%, 34% and 46% for the pediatric, adult, and bariatric phantoms, respectively. Correspondingly, the measured narrow beam transmission fractions were 16%, 9%, and 3%. The eDQE results for the pediatric and adult phantoms correlate well at low spatial frequencies but show degradation in the eDQE at increasing spatial frequencies for the adult phantom in comparison to the pediatric phantom. The results for the bariatric configuration showed a marked decrease in eDQE in comparison to the adult phantom results, across all spatial frequencies, attributable to the combined differences in geometric magnification, and scatter. The eDQE metric has been demonstrated to be sensitive to body habitus suggesting its usefulness in assessing system response across a range of chest sizes and potentially making it a useful factor in protocol assessment and optimization.

In digital mammography noise characteristics are measured in quality control procedures. In the European Guidelines a method of measurement to investigate noise in digital mammography systems was proposed to evaluate the presence of additional noise beside quantum noise. However this method of noise analysis does not discriminate sufficiently between systems with and without additional noise. Therefore a different noise analysis is proposed. In this analysis the noise of a digital system is subdivided into three components: electronic, quantum and structured noise and the noise dose dependency of these components is studied. The usefulness of this analysis in both the frequency and spatial domain is investigated on a number of DR and CR systems. The results show that large differences between digital mammography systems exists. Some systems do have a large range in detector dose for which quantum noise is the largest noise component. For one system however, electronic and structured noise are more dominant. In addition to the differences between systems smaller differences in noise characteristics exist between different target-filter combinations on a particular system. These differences might be attributed to the limited flatfield calibration, the heel effect and difference in sensitivity. The noise analysis in both the frequency and spatial domain give useful information about the noise characteristics of systems. The analysis in the spatial domain is relatively easy to perform and to interpret. This analysis might be suitable for QC purposes. The analysis in the frequency domain does give additional information and might be used for thorough investigations.

European Guidelines for quality control in digital mammography specify minimum and achievable standards of image quality in terms of threshold contrast, based on readings of images of the CDMAM test object by human observers. However the methodology is time-consuming and has large inter- and intra-observer error. To overcome these problems a software program is available to automatically read CDMAM images. An alternative approach would be to predict threshold contrast from measurements of DQE and MTF using a model of the imaging process. A simple signal-matched noise-integration model has been used to predict the contrast detail response of five different types of commercial digital mammography system (Siemens Inspiration, GE Senographe DS, and three types of Konica Minolta computerised radiography system). Measurements were made of the MTF and DQE of each detector and the noise equivalent apertures calculated. For each system sets of 16 images of the CDMAM test object were acquired at a range of dose levels and contrast-detail plots obtained using human observers and automated reading. The theoretically and experimentally determined threshold contrasts were compared. An encouragingly good level of agreement was found between the experimental data and theoretical predictions.

A common method for evaluating projection mammography is Contrast-Detail (CD) curves derived from the CD phantom for Mammography (CDMAM). The CD curves are derived either by human observers, or by automated readings. Both methods have drawbacks which limit their reliability. The human based reading is significantly affected by reader variability, reduced precision and bias. On the other hand, the automated methods suffer from limited statistics. The purpose of this paper is to develop a simple and reliable methodology for the evaluation of mammographic imaging systems using the Signal Known Exactly/Background Known Exactly (SKE/BKE) detection task for signals relevant to mammography. In this paper, we used the spatial definition of the ideal, linear (Hotelling) observer to calculate the task-specific SNR for mammography and discussed the results. The noise covariance matrix as well as the detector response H matrix of the imaging system were estimated and used to calculate the SNR_SKEBKE for the simulated discs of the CDMAM. The SNR as a function of exposure, disc diameter and thickness were calculated.

Amorphous selenium (a-Se) has been incorporated successfully in direct conversion flat panel x-ray detectors, and has demonstrated superior image quality in screening mammography and digital breast tomosynthesis (DBT) under energy integration mode. The present work explores the potential of a-Se for photon counting detectors in DBT. We investigated major factors contributing to the variation in the charge collected by a pixel upon absorption of each x-ray photon. These factors included x-ray photon interaction, detector geometry, charge transport, and the pulse shaping and noise properties of the photon counting readout circuit. Experimental measurements were performed on a linear array test structure constructed by evaporating an a-Se layer onto an array of 100 μm pitch strip electrodes, which are connected to a 32 channel low noise photon counting integrated circuit. The measured pulse height spectrum (PHS) under polychromatic xray exposure was interpreted quantitatively using the factors identified. Based on the understanding of a-Se photon counting performance, design parameters were proposed for a 2D detector with high quantum efficiency and count rate that could meet the requirements of photon counting detector for DBT.

We detail the integration of amorphous silicon (a-Si) active pixel sensor (APS) test arrays with an overlying amorphous selenium (a-Se) x-ray photoconductor, and report on results of their x-ray response and imaging properties. The a-Se/a-Si APS arrays incorporate a two-transistor (2T) gate-switched pixel amplifier architecture designed to provide high detector array resolution, as well as a controllable on-pixel gain. The direct x-ray detectors consist of in-house fabricated, dual mode active and passive sensor arrays with detector element (del) pitches of 100 μm and 200 μm, coated with 80 μm thick stabilized amorphous selenium. These selenium layers were selected for preliminary work and represent a quantum efficiency (QE) of 69% for x-ray spectra (tungsten target, 2 mm Al filtration) of 30 kVp. Detector response was evaluated for a-Se biasing electric fields of both 5 V/μm and 10 V/μm. A detector dark current of 110 pA/cm² (0.01 pA/100 μm del) at 10V/μm electric field, a controllable detector conversion gain up to 15.3 nA/mR at 30 kVp were measured. Active pixel gains of 6.7 and 9.6 were measured for 100μm and 200μm pitch detectors respectively. The amplified readout exhibits a better detection limit (by one order of magnitude) compared to the passive readout implemented on the same pixel. Capabilities of amplified pixels such as nondestructive readout, as well as programmable pixel conversion gain, and dynamic range control are demonstrated. In light of their adaptable gain and dynamic range, these detectors represent a promising technology for high-resolution high gain x-ray digital imaging, particularly in mammography tomosynthesis.

A novel mammography detector with dual amorphous-Selenium (α-Se) layer has been developed that employs photoinduced discharge in its readout procedure. The detector consists of a bias electrode, a thick α-Se layer for x-ray-to-electron conversion, an electron-trapping layer (ETL), a thin α-Se layer for the photoinduced discharge, alternately arranged transparent and opaque stripe electrodes, and a linear optical source for readout. The detector directly converts x-rays into electrons to accumulate the electrons in ETL. When photoinduced discharge arises on the transparent electrodes through readout light irradiation, an accumulated electron image is transferred to the transparent electrodes and detected as signals at charge amplifiers, which we call 'Photoconductive Switching' readout. Readout efficiency and image readout speed have been improved enough to be used as a practical level by the adoption of the alternatelyarranged stripe electrodes. Furthermore, such a simple stripe-electrode structure accomplishes low electrical-noise readout and easy fabrication of fine-pitch pixels. A prototype 50 ìm pixel-pitch detector with the 18×24 cm imaging area has been prepared, which shows high DQE performances more than 64, 48, and 28% at 1, 4, and 7 lp/mm, respectively, not only in the usual dose range of around 100 μGy but at the low dose of 32 μGy. The 'Photoconductive Switching' readout mechanism makes it possible to realize the high DQE and the finest resolution of 50 μm in the direct-conversion detectors for full-field digital mammography.

We present the first evaluation of a recently developed silicon-strip detector for photon-counting dual-energy breast tomosynthesis. The detector is well suited for tomosynthesis with high dose efficiency and intrinsic scatter rejection. A method was developed for measuring the spatial resolution of a system based on the detector in terms of the three-dimensional modulation transfer function (MTF). The measurements agreed well with theoretical expectations, and it was seen that depth resolution was won at the cost of a slightly decreased lateral resolution. This may be a justifiable trade-off as clinical images acquired with the system indicate improved conspicuity of breast lesions. The photon-counting detector enables dual-energy subtraction imaging with electronic spectrumsplitting. This improved the detectability of iodine in phantom measurements, and the detector was found to be stable over typical clinical acquisition times. A model of the energy resolution showed that further improvements are within reach by optimization of the detector.

Recently, Tomosynthesis (TS) has been evaluated as a useful diagnostic imaging examination for the breast, the lung and orthopedics. However the size of the reconstructed region is limited by the mechanical acquisition motion of the X-ray tube and image detector so it is not possible to generate long view images for the spinal columns or the lower limbs examinations. Long-View Tomosynthesis (LVTS) method uses a different acquisition motion and post processing algorithm but results in a similar high resolution image slice free of anatomy above and below slice of interest. This method consists of three steps. First, acquire multi images while X-ray tube and Flat Panel Detector (FPD) are moving continuously in same linear direction. Then each image is divided into strips and strips from different images having similar X-ray beam trajectory are stitched together. Then multi slice coronal images are reconstructed from the long stitched images using filtered backprojection technique (FBP) which is similar to reconstruction algorithms used with Computed Tomography (CT) and TS. As a result, LVTS has 1.6 cycle/mm spatial resolution and 432[mm] × 800[mm] image size at a maximum. We conclude that LVTS improves depiction of long view tomograms, which can not be acquired by TS. Like TS, LVTS can produce images for weight bearing or partial weight bearing anatomy that is not possible with CT since LVTS has been integrated onto a tilting table.

For the last few years, development and optimization of three-dimensional (3D) x-ray breast imaging systems, such as breast tomosynthesis and computed tomography, has drawn much attention from the medical imaging community, either academia or industry. However, the trade offs between patient safety and the efficacy of the devices have yet to be investigated with use of objective performance metrics. Moreover, as the 3D imaging systems give depth information that was not available in planar mammography, standard mammography quality assurance and control (QA/QC) phantoms used for measuring system performance are not appropriate since they do not account for background variability and clinically relevant tasks. Therefore, it is critical to develop QA/QC methods that incorporate background variability with use of a task-based statistical assessment methodology.¹ In this work, we develop a physical phantom that simulates variable backgrounds using spheres of different sizes and densities, and present an evaluation method based on statistical decision theory,² in particular, with use of the ideal linear observer, for evaluating planar and 3D x-ray breast imaging systems. We demonstrate our method for a mammography system and compare the variable phantom case to that of a phantom of the same dimensions filled with water. Preliminary results show that measuring the system's detection performance without consideration of background variability may lead to misrepresentation of system performance.

Tomosynthesis is an imaging technique that has gained renewed interest with recent advancements of flat-panel digital detectors. Because of the wide range of potential applications, a systematic analysis of 3D tomosynthesis imaging systems would contribute to the understanding and development. This paper extends a systematic evaluation of thoracic tomosynthetic imaging performance as a function of imaging parameters, such as the number of projections, tomosynthesis orbital extent, and reconstruction filters. We evaluate lung nodule detectability and anatomical clutter as a function of tomosynthesis orbital extent using anthropomorphic phantoms and a table-top acquisition system. Tomosynthesis coronal slices were reconstructed using the FDK algorithm for cone-beam geometry from 91 projections uniformly distributed over acquisition orbital extents (θ) ranging from 10° to 180°. Visual comparisons of different tomosynthesis reconstructions of a lung nodule show the progressive decrease of anatomical clutter as θ increases. Additionally, three quantitative figures of merit were computed and compared: signal-difference-to-noise ratio (SDNR), anatomical clutter power spectrum (PS), and theoretical detectability index (DI). Lung nodule SDNR increases as θ increases from 0° to 120°. Anatomical clutter PS shows that the clutter magnitude and correlation decrease as θ increases, increasing detectability. Similarly, 2D and 3D DI increase as θ increases in the anatomical dominated exposure ranges. On the other hand, 2D slice DI is lower than the 3D DI for larger θ (e.g. 120°), because of the information loss in the depth direction for 2D slices. In other words, inspecting 3D is better for larger acquisition orbital extents, because the extra information acquired at larger angles cannot be fully recovered from 2D tomosynthesis reconstruction slices. In summary, detectability in tomosynthesis reconstructions for thoracic imaging increases as fixed dose is distributed over a larger acquisition orbital extent (up to 120°).