X-ray imaging properties are reported for HgI₂ and PbI₂, as candidate materials for future direct detection x- ray image sensors, including the first results from screen- printed HgI₂ arrays. The leakage current of PbI₂ is reduced by using new deposition conditions, but is still larger than HgI₂. Both HgI₂ and PbI₂ have high spatial resolution but new data shows that the residual image spreading of PbI₂ is not due to k-edge fluorescence and its possible origin is discussed. HgI₂ has substantially higher sensitivity than PbI₂ at comparable bias voltages, and we discuss the various loss mechanisms. Unlike PbI₂, HgI₂ shows a substantial spatially non-uniform response that is believed to originate from the large grain size, which is comparable to the pixel size. We obtain zero spatial frequency DQE values of 0.7 - 0.8 with PbI(subscript 24/ under low energy exposure conditions. A model for signal generation in terms of the semiconducting properties of the materials is presented.

Initial results from an examination of the performance of small-area, high-spatial-resolution, active matrix, flat- panel imager prototypes, under conditions relevant to mammographic imaging, are reported. These prototypes are based on two 512 X 512 pixel designs: (1) a 75 micrometers pitch indirect-detection array incorporating a continuous photodiode surface operated with a 34 mg/cm² Gd₂O₂S:Tb phosphor screen and with a 150 micrometers thick structured CsI(Tl) scintillator on a fiber-optic plate (which was designed to provide high spatial resolution at the expense of light output); and (2) two 100 micrometers pitch direct-detection arrays with approximately 47 micrometers and approximately 100 micrometers thick PbI₂ photoconductive coatings.

The image quality produced by two examination stands designed for thoracic imaging was evaluated. One of the two stands was using a CsI/a-Si flat panel detector (FD) as the imaging system. The other system was using an imaging plate with a transparent support which is read out from both-sides (tIP). 75 images of an anthropomorphic thorax phantom were produced with both examination stands and on a screen-film-system. Different structures - representing pathological alterations - have been superimposed to the phantom. An ROC analysis of these images has been done. The structures simulating different pathology have been evaluated with respect to the frequency components of their X-ray intensity pattern. This was done by using the method presented at the SPIE 'Medical Imaging' conferences 1999 and 2000 for investigating the normal anatomy. The simulated structures represent different pathological structures sufficiently. The frequency range of their X-ray intensity pattern is in the same order as that found for anatomical structures in the lung. The performed ROC-analysis shows very promising results for both detectors. For both detectors there will probably be a potential to reduce the X-ray dose per exposure without loosing relevant diagnostic quality. This potential is slightly higher for the FD corresponding to its slightly higher DQE at lower frequencies.

Amorphous silicon active matrix flat-panel imagers have gained considerable significant in digital diagnostic medical imaging applications in view of their large area readout capability. The pixel, forming the fundamental unit of the active matrix, consists of a detector and readout circuit. The most widely used architecture is a passive pixel sensor (PPS) where the pixel consists of a detector and an a-Si:H thin-film transistor readout switch. While the PPS has the advantage of being compact and amenable towards high-resolution imaging, reading the low PPS output signal require external circuitry such as column charge amplifiers. More importantly, these amplifiers add a large noise component to the PPS that reduces the minimum readable sensor input signal. This work presents an alternate pixel architecture that can perform on-pixel input signal amplification, i.e. an active pixel sensor (APS). Two operating modes of the APS, voltage output (V-APS) and current output (C-APS) are introduced but the focus is on C- APS. Theoretical calculations indicating the feasibility of the C-APS for low-noise, real time imaging applications (e.g. fluoroscopy) are presented. Specifically, signal gain, dynamic range, readout rate and noise of the C-APS are examined. Lastly, initial experimental results of C-APS linearity and gain are presented in addition to a discussion on APS threshold voltage stability.

Flat X-ray detectors based on CsI:Tl scintillators and amorphous silicon photodiodes are known to exhibit temporal artefacts (ghost images) which decay over time. Previously, these temporal artefacts have been attributed mainly to residual signals from the amorphous silicon photodiodes. More detailed experiments presented here show that a second class of effects, the so-called gain effects, also contributes significantly to the observed temporal artefacts. Both the residual signals and the photodiode gain effect have been characterized under various exposure conditions in the study presented here. The results of the experiments quantitatively show the decay of the temporal artefacts. Additionally, the influence of the detector's reset light on both effects in the photodiode has been studied in detail. The data from the measurements is interpreted based on a simple trapping model which suggests a strong link between the photodiode residual signals and the photodiode gain effect. For the residual signal effect a possible correction scheme is described. Furthermore, the relevance of the remaining temporal artefacts for the applications is briefly discussed for both the photodiode residual signals and the photodiode gain effect.

The image quality of digital radiography systems is influenced by the interplay of many sources of image degradation. These include technology-dependent sources of image blur and noise. This paper systematically investigates the relative influence of the most common sources of image degradation for the entire digital radiography image chain including the detector, image processing and display using well established methods based on linear systems theory. Image quality is quantified in terms of NEQ and DQE. Baseline data are taken from experimental blur and noise measurements from several digital radiography systems, state-of-the-art image processing algorithms and current displays. A generalized theoretical model is exercised to demonstrate the relative importance of the intrinsic noise sources for both direct- and indirect-conversion digital radiography technologies. Since clinical imaging requires a complete system, including a detector, image processing and display, the model is extended to predict the role of image processing and display. This demonstrates the importance of the choice of display parameters such as those that control grayscale rendering, equalization and edge restoration. In summary, the performance of digital radiography systems depends on the interplay of many sources of image degradation from capture to display. Their effects are shown to be interdependent.

A theory of the imaging behavior of structured phosphor layers is developed based on linear analysis and Carlo model modeling and tested by comparison with experiments. The experiments include: the evaluation of Swank factor; modulation transfer function (MTF) measured as a function of spatial frequency f on CsI phosphor layers of different thicknesses and optical properties (e.g. presence or absence of reflective backing layer). These measurements on structured screens were used to establish phosphor parameters (e.g. absorption and scattering lengths). In addition, MTF(f), Wiener noise power spectra [NPS(f)] and detective quantum efficiency [DQE(f)] measurements were obtained from the literature for x-ray image intensifiers and indirect conversion flat panel detectors constructed using CsI layers. These results were compared with the theory using the previously established phosphor parameters. The effects of K-fluorescence and depth dependence of the MTF on DQE (i.e. Lubberts' effect) were investigated.

In digital radiography, radial asymmetry may be present. The use of a one-dimensional representation of the resolution properties can therefore be questioned. Although measurements are often done in two orthogonal directions, there may be a need for a more detailed description. A method of measuring the two-dimensional presampling modulation transfer function (MTF) has therefore been developed. A finely sampled 'disk spread function' is obtained by imaging an aperture mask, consisting of N2 circular holes arranged in an NxN manner in an opaque material, in such a way that each hole is positioned at a different phase relative to the sampling coordinates of the detector system. This spread function is resampled, extrapolated, Fourier transformed, and finally corrected for the finite hole size in order to obtain the presampling MTF. The method was tested on a computed radiography (CR) system through measurements with a prototype mask, consisting of 100 holes of radius 0.2 mm drilled in a lead alloy. The results were compared with measurements using the slit method, and were found to be consistent. Problems associated with the method, e.g. errors due to incorrect alignment of the holes in the aperture mask with the beam, and limitations due to the finite hole size, are discussed.

In July of 2000, a new Working Group (WG33) in IEC/SC62B met for the first time to develop a new IEC standard for the measurement of the detective quantum efficiency of digital x-ray imaging detectors. Experts from meanwhile nine countries are involved. Because of the manifold systems on the market intended for different applications, it was decided to limit the scope of the standard to static systems to be used in general radiography. Dynamic systems (fluoroscopy), systems for mammography and dental systems are excluded but may be treated in additional standards later on.

To judge the potential benefit of a new x-ray detector technology and to be able to compare different technologies, some standard performance measurements must be defined. In addition to technology-related parameters which may influence weight, shape, image distortions and readout speed, there are fundamental performance parameters which directly influence the achievable image quality and dose efficiency of x-ray detectors. A standardization activity for detective quantum efficiency (DQE) for static detectors is already in progress. In this paper we present a methodology for noise power spectrum (NPS), low frequency drop (LFD) and signal to electronic noise ratio (SENR), and the influence of these parameters on DQE. The individual measurement methods are described in detail with their theoretical background and experimental procedure. Corresponding technical phantoms have been developed. The design of the measurement methods and technical phantoms is tuned so that only minimum requirements are placed on the detector properties. The measurement methods can therefore be applied to both static and dynamic x-ray systems. Measurement results from flat panel imagers and II/TV systems are presented.

This paper addresses issues in the calculation of a detectability measure for the ideal linear (Hotelling) observer performing a detection task on a digital radiograph. The main computational problem is that the inverse of a very large covariance matrix is required. The conventional approach is to assume some form of stationarity and argue that the matrix is diagonalized by discrete Fourier transformation, but there are many reasons why this assumption is unrealistic. After a brief review of the underlying mathematics, we present several practical algorithms for computing the detectability and some hints as to when each is applicable. The main conclusion is that large matrices should not be feared.

The detective quantum efficiency (DQE) of an imaging system describes how well the ideal observer performs for a specific imaging detection task. While it can be calculated from measured image data and used to quantify system performance, it is rarely used for assessing fluoroscopic systems. This is primarily due to the effects of system lag. Lag results in a temporal averaging of image signals that reduces noise. As a consequence, the measured DQE of fluoroscopic systems having lag will be erroneously high relative to systems having less lag. This effect can be substantial, resulting in measures of the DQE that can be 15-40% greater than the 'lag-free' DQE. The description of a spatial-temporal NPS and DQE is presented as a means of accommodating system lag. The spatial-temporal NPS has units mm2 s and the spatial-temporal DQE is unitless. Using this generalized interpretation, the (conventional) spatial NPS and DQE are described as sections of the spatial-temporal NPS and DQE along the zero temporal-frequency axis. Calculation of spatial-temporal metrics requires determining an effective temporal aperture related to the temporal MTF. It is shown, both theoretically and experimentally, that the spatial component of the spatial-temporal DQE of a system operating in a fluoroscopic mode is the same as the conventional DQE of the same system operating in a radiographic mode under quantum-noise limited conditions.

The low x-ray exposures used in fluoroscopic applications (0.1 - 10mR at the sensor surface) mean that the requirements for sensor gain and noise are particularly strict. The achievable DQE is determined by a number of factors, including the sensor quantum efficiency, x-ray absorption Swank factor, secondary quanta conversion efficiency, internal gain (e.g. the number of electrons collected per visible photon produced in the phosphor), and additive noise. The influence of these factors is examined for three direct detection x-ray sensors (PbI2, a-Se and GaAs), and one indirect detector sensor (CsI). Although the characteristics of these sensors are very different, it is demonstrated that all are appropriate for use in fluoroscopic applications as a replacement for current image intensifier based systems.

As amorphous selenium based flat panel detectors gain more interest for direct, real-time x-ray imaging, we report in this paper the performance of such a detector by ANRAD Corporation. This new detector is based on a 1536 X 1536 array of amorphous silicon TFT pixels coupled with a 1000 micrometers selenium converter biased at 10 V/micrometers . Each 150 micrometers X 150 micrometers pixel is made of a thin film transistor, a storage capacitor and a collecting electrode having a geometrical fill factor of 77% and an effective fill factor of nearby 100%.

We have developed new X-ray image intensifier (Flat II) system using a large-area electron multiplier for real-time fluoroscopy. This paper describes the development of prototype Flat II that has been carried out recently. The heart of new prototype is the multiplier with a smaller channel pitch (0.2 mm) located adjacent to a photo-cathode. We performed reduction of the channel pitch of multiplier, and the optimization of the geometrical arrangement- a photo-cathode, a multiplier and an output phosphor screen in order to improve the performance of Flat II. Moreover, we discuss the spatial aliasing associated with a pitch pattern of multiplier and CCD camera. Finally, physical characteristics such as modulation transfer function (MTF), etc. are discussed.

In order to study the advantage and remaining problems of FPD (flat panel detector) for clinical use by the real-time DR (digital radiography) system, we developed a prototype system using a scintillator type FPD and which was compared with previous I.I.-CCD type real-time DR. We replaced the X- ray detector of DR-2000X from I.I.-4M (4 million pixels)-CCD camera to the scintillator type dynamic FPD(7' X 9', 127 micrometers ), which can take both radiographic and fluoroscopic images. We obtained the images of head and stomach phantoms, and discussed about the image quality with medical doctors.

Our work is to investigate and understand the factors affecting the imaging performance of amorphous selenium (a-Se) flat-panel detectors for digital mammography. Both theoretical and experimental methods were developed to investigate the spatial frequency dependent detective quantum efficiency [DQE(f)] of a-Se flat-panel detectors for digital mammography. Since the k-edge of a-Se is 12.66 keV and within the energy range of a mammographic spectrum, a cascaded linear system model was developed which takes into account the effect of k-fluorescence on the modulation transfer function (MTF), noise power spectrum (NPS) and DQE(f) of the detector. This model was used to understand the performance of a prototype detector with 85 mm pixel size. The presampling MTF, NPS and DQE(f) of the prototype were measured, and compared to the theoretical calculation by the model. The calculation showed that k-fluorescence reduces the MTF by 15% at the Nyquist frequency (f_NY) of the prototype detector, and the NPS at f_NY was reduced to 82% of that at zero spatial frequency. Because of the decrease in both MTF and NPS at high spatial frequencies, k-fluorescence only has a small degradation effect on DQE(f) for mammography. The measurement of presampling MTF of the prototype detector revealed an additional source of blurring, which was attributed to the blocking layer at the interface between a-Se and the active matrix. This introduced high frequency drop in both presampling MTF and NPS, and reduced aliasing in the NPS. As a result, the DQE(f) of the prototype detector at f_NY approaches 50% of that at zero spatial frequency.

We have performed small angle X-ray scattering measurements at the Synchrotron Radiation Source at Daresbury to make an initial assessment of the diagnostic information obtainable from various breast tissues. These experiments have indicated that high quality interpretable diffraction data can be rapidly produced from breast core cut biopsy specimens. We have demonstrated a remarkable and systematic difference between the X-ray scattering from normal, benign and malignant breast tissue collagen. Our findings indicate that it may be possible to use molecular structure characteristics of breast tissue as novel markers of disease progression.

Detection of mass lesions in mammograms is essentially limited by image variation due to normal patient structure, which has an average power-spectrum of the form `1/f³'. Image noise plays little role in limiting mass detection. The contrast-detail (CD) diagram for lesion detection in mammographic structure is novel, for both human and model observers. Contrast thresholds increase with increasing signal size for signals larger than about 1 mm, with CD slopes of about 0.3 for humans and 0.4 for model observers. Similar results were obtained in search experiments. The work was done using hybrid images, with of tumor masses (extracted from specimen radiographs) added to digitized mammographic backgrounds. We have been able to explain the results using a number of observer models. These results demonstrate that CD diagrams based on image noise-limited detection with simple phantoms are not useful for evaluation of mass detection in mammograms--so more realistic approaches are necessary in order to model mammographic imaging systems for optimization.

Currently, there is significant interest in quantifying the performance of digital radiography systems. Digital radiography systems can be thought of as continuous linear shift-invariant systems followed by sampling. This view, along with the large number of pixels used for flat-panel systems, has motivated much of the work which attempts to extend figures of merit developed for analog systems, in particular, the detective quantum efficiency (DQE) and the noise equivalent quanta (NEQ). A more general approach looks at the system as a continuous-to-discrete mapping and evaluates the signal-to-noise ratio (SNR) completely from the discrete data. In this paper, we study the effect of presampling blur on these figures of merit. We find that even in this idealized model the DQE/NEQ formulations do not accurately track the behavior of the fully digital SNR. Therefore, DQE/NEQ cannot be viewed as indicators of the signal-known-exactly/background-known-exactly (SKE/BKE) detectability. In order to make design decisions by optimizing detectability one must work with the fully digital definition of detectability.

The modulation transfer function and detective quantum efficiency are modeled for a Full Field Digital Mammography detector constructed with a CsI scintillator deposited on an amorphous silicon active matrix array. The model is evaluated against experimental measurements using different exposure levels, x-ray tube voltages, target composition and beam filtrations as well as varying thicknesses and compositions of filtration materials placed in the path between the tube and detector. Available x-ray tube emission spectrum models were evaluated by comparison against the measured transmission through aluminum. The observed variation of DQE at zero spatial frequency among different target/filter conditions, acrylic filtration thicknesses and kVp is well characterized by a x-ray model. This variation is largely accounted for by just two effects -- the attenuation of x-rays through the detector enclosure and the stopping power of x-rays in the CsI layer. Additional considerations such as the Lubberts effect were included in the analysis in order to match the measured DQE(k) as a function of spatial frequency, k. The pixel aperture and light channeling through the scintillator shape the MTF which acts favorably to avoid aliasing due to digital sampling.

A selenium-based flat-panel direct converter detector suitable for digital mammography was developed. The detector is based on a TFT-array with a resolution of 2816 X 2048 pixels, and a pixel pitch of 85 micrometers . Although the geometric fill factor for each pixel is around 70% , the effective fill factor for the detector is closer to 88% due to internal electric field shaping within the selenium layer. A selenium multilayer p-i-n structure of 200 micrometers was deposited onto the array by selectively doping the regions near each contact to produce unipolar conducting blocking layers. This structure absorbs more than 95% of a typical mammography beam.

The storage phosphor RbBr:Tl⁺ can be grown in needles via vacuum deposition. Thanks to reduced lateral light diffusion thick needle screens still offer acceptable resolution. Due to its low intrinsic X-ray absorption, however, a RbBr:Tl⁺ needle screen does not lead to a better absorption/resolution compromise than a BaFBr_1-xI_x:Eu²⁺ powder screen. CsBr:Eu²⁺ does combine high specific X-ray absorption and the possibility of needle growth. Its blue emission, peaking at 440 nm and near IR stimulation band, with maximum at 685 nm, make it well suited for use in CR systems. Sensitivity and sharpness of a 500 (mu) thick CsBr:Eu²⁺ needle screen were measured in a flying-spot scanner. The number of photostimulated light quanta per absorbed X-ray quantum is higher than for BaFBr_1-xI_x:Eu²⁺. At 70 kVp and 0.5 mm Cu filtration, equal sharpness is obtained for 85% vs. 46% X-ray absorption in BaFBr_1-xI_x:Eu²⁺ screens. DQE was measured at 2.5 (mu) Gy, 70 kVp, and 0.5 mm Cu filtration for a CsBr:Eu²⁺ needle screen in a flying-spot scanner. Up to 3 lp/mm, DQE was 2 times higher than for state-of-the-art CR systems and equal to the DQE claimed for flat panel DR systems, based on a-Si photodiodes combined with a CsI:Tl scintillator layer.

A 2.5 cm by 5 cm, 512 by 1024 pixel CMOS photodiode array detector designed specifically for digital radiography will be described. All necessary scanning and readout circuitry is integrated within the detector. The small pixel spacing of 48 micrometers allows the imager to easily achieve the 10 lp/mm resolution required for the targeted interventional mammography application. Direct coupling to the scintillator and a pixel fill factor of more than 80% leads to high DQE over a large range of exposure values. The detector exhibits very low dark current of about 30 pA/cm² at room temperature, which allows for low-noise operation and long integration times. Read noise of less than 200 electrons rms and a saturation level of 2.8 X 10⁶ electrons combine for a large dynamic range greater than 80 dB. The conversion gain of the detector is 0.5 (mu) V/electron. Combined with a Gd₂O²S scintillator, the imager achieves up to 25% MTF at 10 lp/mm and a DC detective quantum efficiency of 50%. The detector design is optimized for x-ray energies between 10 kV and 50 kV, but can be retrofitted with different scintillator screens to cover a large range of imaging applications up to 160 kV.

We report the results of performance measurements for an amorphous silicon flat panel detector used in a cardiovascular imaging system. The detector contains 1024 x 1024 elements on a 0.2 mm pitch for an active image area of about 20.5 x 20.5 cm². The system allows imaging at fluoroscopic and dynamic cardiac record exposure levels at rates of up to 30 Hz. We measured MTF, NPS, DQE, contrast ratio, response uniformity, resolution uniformity, and lag. Measurements were made on 28 production detectors. The MTF was greater than 0.2 at 2.5 cycles/mm. Contrast ratio was several hundred, indicating negligible long range scatter (veiling glare) within the detector. The DQE of the detector was measured at exposures typical of fluoroscopic imaging, dynamic cardiac record imaging, and digital subtraction angiography (DSA). The DQE was at least 0.65, 0.54, and 0.34 at 0, 1, and 2 cycles/mm, respectively, for all of these exposure levels. The response of the detector varied by less than 12% across its surface. The MTF, measured at nine positions over the surface of the detector, was found to have a maximum difference among positions of less than 0.05 at both 1 and 2 cycles/mm. First frame lag was less than 5%.

A production 41cm square digital flat panel detector constructed of amorphous silicon with a cesium iodide scintillator was characterized for its applicability to 'fast' radiographic imaging. Representative fast imaging clinical applications include tomosynthesis and dual energy utilizing two exposures. The detector supports a read time of 125msec at a radiographic dynamic range of 7mR, giving a frame rate of 7.5 acquisitions per second. Detector performance measures of lag, detector sensitivity, and DQE were evaluated as a function of acquisition rate and exposure. The first frame lag is approximately 2% of the stimulus and decays to less than 0.5% for all subsequent frames. In a sequence at constant exposure, the lag builds up as the signal level grows to 5%, reaching a steady state after approximately 15 frames. The DQE measurement is affected by lag and must be corrected. The GE radiographic detector has outstanding dynamic performance and gives all indications that it will meet the fast imaging application requirements.

A method is developed that enables comprehensive and automatic image quality testing of storage phosphor-based computed radiography (CR) systems. The measurements made with this method are both quantitative and objective, and are consistent with the guidelines set forth by the American Association of Physics in Medicine Task Group 10. The method is implemented as a toolkit that consists of a test phantom, a procedure for making controlled x-ray exposures, and software that automatically analyzes phantom and flat-field images then compares the measured results against a set of tolerances. By using this new method, acceptance testing and periodic image quality control testing of CR systems can be performed quickly and efficiently. A complete CR test that includes measurements for three cassette sizes can be finished in about one hour.

A technique called Variable-Resolution X-ray (VRX) detection greatly increases the spatial resolution in computed tomography (CT) and digital radiography (DR) as the field size decreases. The technique is based on a principle called `projective compression' that allows both the resolution element and the sampling distance of a CT detector to scale with the subject or field size. For very large (40 - 50 cm) field sizes, resolution exceeding 2 cy/mm is possible and for very small fields, microscopy is attainable with resolution exceeding 100 cy/mm. This paper compares the benefits obtainable with two different VRX detector geometries: the single-arm geometry and the dual-arm geometry. The analysis is based on Monte Carlo simulations and direct calculations. The results of this study indicate that the dual-arm system appears to have more advantages than the single-arm technique.

The CT image is a representation of the patient's anatomy as measured in terms of such physical characteristics as density, electron density, and atomic number. The process of sampling the patient's molecular composition with the x-ray beam is subject to varied physical effects that degrade the ability of the CT image data set to accurately represent the tissues of the body. To assess the impact of patient and scanner related characteristics on the final CT image a Monte Carlo based modeling has been developed that can simulate such effects as scatter and beam hardening on the reconstructed CT image. By selectively studying the effects of these variables, the model can be used as a design tool for improving the diagnostic capabilities of a CT scanner and /or correcting or eliminating unwanted sources of variation in the CT image. Initial simulations model the unique geometry of Electron Beam CT with a clinical goal of correcting for scanner and patient related physical factors that may cause variations in the assessment of Coronary Artery Calcium.

Cardiac imaging is still a challenge to CT reconstruction algorithms due to the dynamic nature of the heart. We have developed a new reconstruction technique, called the Flexible Algorithm, which achieves high temporal resolution while it is robust to heart-rate variations. The Flexible Algorithm, first, retrospectively tags helical CT views with corresponding cardiac phases obtained from associated EKG. Next, it determines a set of views for each slice, a stack of which covers the entire heart. Subsequently, the algorithm selects an optimum subset of views to achieve the highest temporal resolution for the desired cardiac phase. Finally, it spatiotemporally filters the views in the selected subsets to reconstruct slices. We tested the performance of our algorithm using both a dynamic analytical phantom and clinical data. Preliminary results indicate that the Flexible Algorithm obtains improved spatiotemporal resolution for a large range of heart rates and variations than standard algorithms do. By providing improved image quality at any desired cardiac phase, and robustness to heart rate variations, the Flexible Algorithm enables cardiac applications in CT, including those that benefit from multiphase information.

X-ray projection mammography, using a film/screen combination or digital techniques, has proven to be the most effective imaging modality for early detection of breast cancer currently available. However, the inherent superimposition of structures makes small carcinoma (a few millimeters in size) difficult to detect in the occultation case or in dense breasts, resulting in a high false positive biopsy rate. The cone-beam x-ray projection based volume imaging using flat panel detectors (FPDs) makes it possible to obtain three-dimensional breast images. This may benefit diagnosis of the structure and pattern of the lesion while eliminating hard compression of the breast. This paper presents a novel cone-beam volume CT mammographic imaging protocol based on the above techniques. Through computer simulation, the key issues of the system and imaging techniques, including the x-ray imaging geometry and corresponding reconstruction algorithms, x-ray characteristics of breast tissues, x-ray setting techniques, the absorbed dose estimation and the quantitative effect of x-ray scattering on image quality, are addressed. The preliminary simulation results support the proposed cone-beam volume CT mammographic imaging modality in respect to feasibility and practicability for mammography. The absorbed dose level is comparable to that of current two-view mammography and would not be a prominent problem for this imaging protocol. Compared to traditional mammography, the proposed imaging protocol with isotropic spatial resolution will potentially provide significantly better low contrast detectability of breast tumors and more accurate location of breast lesions.

With the introduction of helical, multi-detector computed tomography (CT) scanners having sub-second scanning speeds, clinicians are currently investigating the role of CT in cardiac imaging. In this paper, we describe a four-dimensional (4D) x-ray attenuation model of a human heart and the use of this model to assess the capabilities of both hardware and software algorithms for cardiac imaging. We developed a model of the human thorax, composed of several analytical structures, and a model of the human heart, constructed from several elliptical surfaces. A model for each coronary vessel consists of a torus placed at a suitable location on the heart's surface. The motion of the heart during the cardiac cycle was implemented by applying transformational operators to each surface composing the heart. We used the 4D model of the heart to generate forward projection data, which then became input into a model of a CT imaging system. The use of the model to predict image quality is demonstrated by varying both the reconstruction algorithm (sector-based, half-scan) and CT system parameters (gantry speed, spatial resolution). The mathematical model of the human heart, while having limitations, provides a means to rapidly evaluate new reconstruction algorithms and CT system designs for cardiac imaging.

Real-time 3D CT is a high-speed cone-beam CT (4D-CT) with good low-contrast detectability. In a single rotation, a voxel data set with higher spatial resolution over a wide z- axis range can be obtained. Using continuous rotation, temporarily continuous voxel data sets and 3D dynamic images can be acquired. We have developed a large area 2D detector for 4D-CT, and we have got the first 4D-CT test images.

Dual-energy subtraction imaging increases the sensitivity and specificity of pulmonary nodule detection in chest radiography by reducing the contrast of overlying bone structures. Recent development of a fast, high-efficiency detector enables dual-energy imaging to be integrated into the traditional workflow. We have modified a GE Revolution^TM XQ/i chest imaging system to construct a dual-energy imaging prototype system. Here we describe the operating characteristics of this prototype and evaluate image quality. Empirical results show that the dual-energy CNR is maximized if the dose is approximately equal for both high and low energy exposures. Given the high detector DQE, and allocation of dose between the two views, we can acquire dual-energy PA and conventional lateral images with total dose equivalent to a conventional two-view film chest exam. Calculations have shown that the dual-exposure technique has superior CNR and tissue cancellation than single-exposure CR systems. Clinical images obtained on a prototype dual-energy imaging system show excellent tissue contrast cancellation, low noise, and modest motion artefacts. In summary, a prototype dual-energy system has been constructed which enables rapid, dual-exposure imaging of the chest using a commercially available high-efficiency, flat-panel x-ray detector. The quality of the clinical images generated with this prototype exceeds that of CR techniques and demonstrates the potential for improved detection and characterization of lung disease through dual-energy imaging.

A new 5C1 beam line was recently built on Pohang Light Source (PLS) in Korea for possible medical applications. An imaging system on the beam line consisted of silicon wafer attenuators, tungsten slit, CdWO₄ scintillator, CCD camera, and personal computer. Both spatial resolution test pattern and glass capillary filled with air bubbles were imaged to evaluate resolution and phase-contrast effects for synchrotron imaging system, respectively. Mammographic accreditation phantom, cancerous human breast specimens, and fish and nude mouse were also imaged. Both spatial resolution and image quality of the synchrotron imaging system were compared with those of conventional mammography system. The images of resolution test pattern showed a few tens of micrometer spatial resolution on synchrotron imaging system. The images of capillary filled with air bubbles revealed phase contrast effect. Mammographic accreditation phantom, human breast specimen, and animal images with synchrotron imaging system showed much higher resolution with great details of object and higher enhanced contrast compared to those with conventional mammography system. We successfully implemented and tested a micrometer resolution imaging system on PLS 5C1 beam line. Using a simple and inexpensive synchrotron imaging system on PLS 5C1 beam line, micrometer resolution images for resolution test pattern, mammographic accreditation phantom, human breast specimens, and live animals were acquired and visually evaluated.

In recent years, the scan speed of computed tomography (CT) has increased significantly. Not long ago, the state-of-the- art CT was only capable of completing a single scan in 1.0 s per gantry rotation. Nowadays, 0.5 s per revolution is nearly an industry standard. Faster scan speeds demand faster sampling of the projections to combat aliasing artifacts, and higher x-ray tube output to ensure sufficient x-ray photon flux delivered to the scan. These demands often exceed the technological capability of these components. In this paper we performed a detailed analysis on the characteristics of the view aliasing artifact. Based on our analysis and clinical observations, we propose an adaptive view synthesis (AVS) scheme that effectively reduces the demands on the data acquisition system. Detailed performance comparison between the full view sampling and the adaptive view synthesis are performed through computer simulations and phantom experiments. Our analysis indicates that AVS is adequate for routine clinical applications.

The display of low-contrast structures and fine microcalcifications is essential for the early diagnosis of breast cancer. In order to achieve a high image quality level with a minimum amount of radiation delivered to the patient, the use of different spectra (Mo or Rh anode and filters) was introduced. The European Synchrotron Radiation Facility is able to produce a monochromatic beam with a high photon flux. It is thus a powerful tool to study the effect of beam energy on image quality and dose in mammography. Our image quality assessment is based on the calculation of the size of the smallest microcalcification detectable on a radiograph, derived from the statistical decision theory. The mean glandular dose is simultaneously measured. Compared with conventional mammography units, the monochromaticity of synchrotron beams improves contrast and the use of a slit instead of an anti-scatter grid leads to a higher primary beam transmission. The relative contribution of these two effects on image quality and dose is discussed.

The clinical goal of breast imaging is to detect tumor masses when they are as small as possible, preferably less than 10 mm in diameter. Conventional film-screen mammography is the most effective tool for the early detection of breast cancer currently available. However, conventional mammography has relatively low sensitivity to detect small breast cancers (under several millimeters). Specificity and the positive predictive value of mammography remain limited owing to an overlap in the appearance of benign and malignant lesions, and surrounding structure. The limitations accompanying conventional mammography is to be addressed by incorporating a cone beam volume CT reconstruction technique with a recently developed flat panel detector. A computer simulation study has been performed to prove the feasibility of developing a flat panel detector-based cone beam volume CT breast imaging (FPD-CBVCTBI) technique. In this study, a preliminary phantom experiment is conducted to verify the findings in the computer simulation using a prototype flat panel detector-based cone beam volume CT scanner. The results indicate that the FPD-CBVCTBI technique effectively removes structure overlap and significantly improves the detectability of small breast tumors. This suggests that FPD-CBVCTBI is a potentially powerful breast-imaging tool.

The usefulness of Fourier-based measures of imaging performance has come into question for the evaluation of digital imaging systems. Figures of merit such as detective quantum efficiency are relevant for linear, shift-invariant systems with stationary noise. However, no digital imaging system is shift invariant, and realistic images do not satisfy the stationarity condition. Our methods for task- based evaluation of imaging systems, based on lesion detectability, do not require such assumptions. We have computed the performance of Hotelling and nonprewhitening matched-filter observers for the task of lesion detection in digital radiography.

Gadovist, a 1.0-molar Gd contrast agent from Schering AG, Berlin, Germany, in use in clinical MRI in Europe, was evaluated as a radiography contrast agent. In a collaboration with Brookhaven National Laboratory (BNL), Schering AG is developing several such lanthanide-based contrast agents, while BNL evaluates them using different x-ray beam energy spectra. These energy spectra include a 'truly' monochromatic beam (0.2 keV energy bandwidth) from the National Synchrotron Light Source (NSLS), BNL, tuned above the Gd K-edge, and x-ray-tube beams from different kVp settings and beam filtrations. Radiographs of rabbits' kidneys were obtained with Gadovist at the NSLS. Furthermore, a clinical radiography system was used for imaging rabbits' kidneys comparing Gadovist and Conray, an iodinated contrast agent. The study, using 74 kVp and standard Al beam filter for Conray and 66 kVp and an additional 1.5 mm Cu beam filter for Gadovist, produced comparable images for Gadovist and Conray; the injection volumes were the same, while the radiation absorbed dose for Gadovist was slightly smaller. A bent-crystal silicon monochromator operating in the Laue diffraction mode was developed and tested with a conventional x-ray tube beam; it narrows the energy spectrum to about 4 keV around the anode tungsten's K' line. Preliminary beam-flux results indicate that the method could be implemented in clinical CT if x-ray tubes with approximately twice higher output become available.

This investigation tests the hypothesis that volume anisotropy intrinsic to tomosynthetic reconstructions can be minimized through integration of contiguously sampled orthogonal projections. The basic idea involves the acquisition of data from two orthogonal projection geometries using total disparity angles of 90 degree(s) for each projection series. This allows projections at the angular extremes of one series to correspond to projections produced at opposite angular extremes of the series orthogonal to the first.

Preliminary MTF and LCD results obtained on several volumetric computed tomography (VCT) systems, employing amorphous flat panel technology, are presented. Constructed around 20-cm x 20-cm, 200-mm pitch amorphous silicon x-ray detectors, the prototypes use standard vascular or CT x-ray sources. Data were obtained from closed-gantry, benchtop and C-arm-based topologies, over a full 360 degrees of rotation about the target object. The field of view of the devices is approximately 15 cm, with a magnification of 1.25-1.5, providing isotropic resolution at isocenter of 133-160 mm. Acquisitions have been reconstructed using the FDK algorithm, modified by motion corrections also developed by GE. Image quality data were obtained using both industry standard and custom resolution phantoms as targets. Scanner output is compared on a projection and reconstruction basis against analogous output from a dedicated simulation package, also developed at GE. Measured MTF performance is indicative of a significant advance in isotropic image resolution over commercially available systems. LCD results have been obtained, using industry standard phantoms, spanning a contrast range of 0.3-1%. Both MTF and LCD measurements agree with simulated data.

Arthritis is a painful condition with enormous societal impact. Arthritis damages the articular cartilage between adjacent bones in a joint, which is seen radiographically as narrowing of the joint space width (JSW). JSW is an important arthritis outcome measure however a single radiographic image is a 2D projection of a 3D structure and diseased areas can be obscured. To quantify the JSW in three dimensions we have applied digital tomosynthesis imaging to hand radiography. A tomosynthesis algorithm, developed for use in chest radiography, was modified to provide reconstructed slices through the bones that formed joints of the hand. The methodology was tested using simulated radiographs of dry-bone specimens from 3 hand skeletons. Estimates to the JSW in 3D were made from the reconstructed slices. The algorithm produced tomographic slices through the bones of the joint with minimal loss of spatial resolution. We discovered that hand radiography is ideally suited for tomosynthesis imaging due to the small amount of scatter and lack of truncation artifacts. We have demonstrated the utility of digital tomosynthesis for use in quantifying JSW for arthritis assessment. The method shows promise for improving the assessment of disease progression.

Digital tomosynthesis is a method for reconstructing arbitrary planes in an object from a series of projection radiographs, acquired with limited angle tube movement. Conventional 'shift and add' tomosynthesis suffers from the presence of blurring artifacts, created by objects located outside of each reconstructed plane. Matrix inversion tomosynthesis (MITS) uses known geometry, and a set of coupled linear algebra equations to solve for the blurring function in each reconstructed plane, enabling removal of the unwanted out-of-plane blur artifacts. For this paper, both MITS and conventional tomosynthesis reconstructions were generated for a simulated impulse located at varying distance from the detector, and also an anthropomorphic chest phantom. Exploration of the effects of total angular tube movement, number of projection radiographs acquired, and number of planes reconstructed via matrix inversion tomosynthesis, on residual out-of-plane blur ensued. We conclude that optimization of image acquisition and plane reconstruction parameters can improve slice image quality. In all examined scenarios, the MITS algorithm outperforms conventional tomosynthesis in removing out-of-plane blur.

Recently new technologies for detecting biomolecules have been developed and are opening a new era of medical imaging. Chemiluminescence and fluorescence are emerging as promising tools for these tasks. These molecules emit optical photons that can be observed outside the body. Unfortunately, they are heavily scattered and absorbed in biological tissue. This is an obstacle for determining a way of mapping an original source distribution. In order to overcome this obstacle, we suggest a new concept, Optical Emission Computed Tomography and test its feasibility with computer simulation.

Cone-beam computed tomography (CBCT) based upon large-area flat-panel imager (FPI) technology is a flexible and adaptable technology that offers large field-of-view (FOV), high spatial resolution, and soft-tissue imaging. The imaging performance of FPI-based cone-beam CT has been evaluated on a computer-controlled bench-top system using an early prototype FPI with a small FOV (20.5 X 20.5 cm²). These investigations demonstrate the potential of this exciting technology. In this report, imaging performance is evaluated using a production grade large-area FPI (41 X 41 cm²) for which the manufacturer has achieved a significant reduction in additive noise. This reduction in additive noise results in a substantial improvement in detective quantum efficiency (DQE) at low exposures. The spatial resolution over the increased FOV of the cone-beam CT system is evaluated by imaging a fine steel wire placed at various locations within the volume of reconstruction. The measured modulation transfer function (MTF) of the system demonstrates spatial frequency pass beyond 1 mm-1 (10% modulation) with a slight degradation at points off the source plane. In addition to investigations of imaging performance, progress has also been made in the integration of this technology with a medical linear accelerator for on-line image-guided radiation therapy. Unlike the bench-top system, this implementation must contend with significant geometric non-idealities caused by gravity-induced flex of the x-ray tube and FPI support assemblies. A method of characterizing and correcting these non-idealities has been developed. Images of an anthropomorphic head phantom qualitatively demonstrate the excellent spatial resolution and large FOV achievable with the cone-beam approach in the clinical implementation.

Near-infrared spectroscopy is a relatively new imaging method, which can provide important information on concentrations of oxy-and deoxy-hemoglobin in cortical areas of the brain. We discuss the advantages of the integration of magnetic resonance and optical imaging techniques and present the results of our experimental study on the comparison of optical and fMRI signals obtained simultaneously on humans during functional activity and at rest. In all subjects we found a good collocation of the brain activity centers revealed by both methods. We also found a high temporal correlation between the BOLD signal (fMRI) and the deoxy-hemoglobin concentration (near-infrared spectroscopy) in the subjects who exhibited low fluctuations in superficial head tissues. The contamination of optical signals by superficial tissue layers urges applying algorithms of three-dimensional optical tomography.

Monochromatic parallel beam imaging produces high subject contrast, high resolution, and low patient dose. Polycapillary collimating optics can be used to create a beam of sufficient intensity for monochromatization from a conventional source. Monochromatization is achieved by diffraction from a single crystal. Contrast, resolution, and intensity measurements were performed with both high and low angular acceptance crystals. Testing was first done at 8 keV with an intense copper rotating anode, then preliminary 17.5 keV measurements were made with a low power molybdenum source. At 8 keV, contrast enhancement was a factor of 5 relative to the polychromatic case, in good agreement with theoretical values. At 17.5 keV, monochromatic subject contrast is a factor of 2 times greater than the conventional polychromatic contrast. An additional factor of two increase in contrast is expected from the removal of scatter obtained from using the air gap which is allowable from the parallel beam. The measured angular resolution after the crystal was 0.6 mrad for a silicon crystal. The use of polycapillary collimating optics allowed monochromatic imaging measurements using a conventional rotating anode source and computed radiography plate in 300 mAs.

A copper crystal lens designed to focus gamma ray energies of 100 to 200 keV has been assembled at Argonne National Laboratory. In particular, the lens has been optimized to focus the 140.6 keV gamma rays from technetium-99 m typically used in radioactive tracers. This new approach to medical imaging relies on crystal diffraction to focus incoming gamma rays in a manner similar to a simple convex lens focusing visible light. The lens is envisioned to be part of an array of lenses that can be used as a complementary technique to gamma cameras for localized scans of suspected tumor regions in the body. In addition, a 2- lens array can be used to scan a woman's breast in search of tumors with no discomfort to the patient. The incoming gamma rays are diffracted by a set of 828 copper crystal cubes arranged in 13 concentric rings, which focus the gamma rays into a very small area on a well-shielded NaI detector. Experiments performance with technetium-99 m and cobalt 57 radioactive sources indicate that a 6-lens array should be capable of detecting sources with (mu) Ci strength.

A new method for evaluating Magnetic Resonance Elastography (MRE) wave images is introduced, which consists of both local frequency estimation (LFE) and simulation of wave patterns by a coupled harmonic oscillator (CHO) approach. It is shown that i) LFE performs improved reconstruction by use of Gauss filters and ii) CHO calculations can help to refine the resulting wave speed or elasticity map by taking local attenuation, reflection, and diffraction into account. The performance of new LFE and CHO calculations is demonstrated by MRE experiments on a gel phantom as well as by simulations of shear waves in a brain phantom, such as for potential in-vivo MRE-experiments.

A new type of medical imaging system based on crystal diffraction is being developed at the Advanced Photon Source at Argonne National Laboratory. It is designed to image very small amounts of radioactivity in the human body. The system has very fine spatial resolution (1-3 mm) and very high sensitivity. By combining two or more lenses, one can generate a 3-D image of the cancer. Micro-Curie sources can be detected with relative ease. These features make the system very useful to confirm or reject possible sites for a cancer in the human body following a full body scan. It could also have considerable use as a method of checking for breast cancer.

Preliminary results are presented from the PaxScan 4030A; a 40x30cm, 2048 x 1536 landscape, flat panel imager, with 194um pixel pitch. This imager builds on our experience with the PaxScan 2520, a 127um real-time flat panel detector capable of both high-resolution radiography and low dose fluoroscopy. While the PS2520 has been applied in C-arms, neuroangiography, cardiac imaging and small area radiographic units, the larger active area of the PaxScan 4030A addresses the broader applications of angiography, general R&F and cone-beam CT. The PaxScan 4030A has the same electrical and software interfaces as the PS2520; however, a number of innovations have been incorporated into the 4030A to increase its versatility. The most obvious change is that the data interface between the receptor and command processor has been reduced to one very flexible and thin fiber-optic cable. A second new feature for the 4030A is the use of split datalines. Split datalines facilitate scanning the two halves of the array in parallel, cutting the readout time in half and increasing the time window for pulsed x-ray delivery to 15ms at 30fps. In addition, split datalines result in lower noise, which, coupled with the larger signal of the 194um pixels, enables high quality imaging at lower fluoroscopy doses rates.

The design, development and evaluation of a portable x-ray detector are described. The completed detector has a pixel pitch of 100 micrometers , an active imaging area of 22.5 x 27.5 cm² (9 x 11 inch²), package outer dimensions of 32.5 x 32.5 cm², a thickness of only 20 mm, and a weight of around 2.8 kg. A number of significant advances in the design and production processes were needed to produce such a compact detector with such a small pixel pitch, while maintaining the image quality achieved a current detector (CXDI-22) which has a 160 mm pixel pitch. These include the development of a low power readout IC, advances in detector packaging design, concentrating on lightweight and strong components, and redesign of the pixel structure to improve the fill-factor. A comparison is made of the imaging characteristics of this new detector with the CXDI-22 detector, and it is shown that the new detector demonstrates improved CTF, and NEQ. The new dete