The development of new detectors for diagnostic x-ray imaging is a complex and expensive endeavour. An understanding of fundamental performance potential and limitations is therefore critical to the wise allocation of research resources. We present a Monte Carlo study in which the fundamental spatial resolution limitations imposed by x-ray interactions were determined for both direct conversion amorphous selenium (a-Se) and indirect conversion cesium iodide (CsI) detectors. Using a simulated infinitesimal x-ray beam, the absorbed energy point spread function (PSF) in each detector material was scored within rectilinear bin sizes of 5 mm for incident x-ray energies between 10 and 100 keV. The modulation transfer function (MTF) was determined from each simulated PSF and characterized in terms of the 50% MTF frequency, f₅₀, and the equivalent passband, N_e. Both materials demonstrated: (i) a drop in f₅₀ (a-Se: 25%, CsI: 85%) and N_e (a-Se: 45%, CsI: 75%) immediately above the K-edge energy due to re-absorption of characteristic radiation, and (ii) a moderate recovery of f₅₀ and N_e levels with further increase in energy. In addition, within the diagnostic energy range and spatial frequency range of 0 -- 20 cycles/mm, the values of the fundamental MTF due to x-ray interactions remain above 50%. In general, we conclude that existing amorphous selenium and cesium iodide detectors operate far from fundamental spatial resolution limits in both mammography and radiography applications. Further reduction in detector element size will potentially improve spatial resolution in these detectors.

The current system performance metrics for Digital Radiographic detectors describe physical parameters, such as resolution (Modulation Transfer Function), noise (Noise Power Spectrum) and efficiency (Detective Quantum Efficiency). However, little has been done to substantiate the impact of these quantitative image quality metrics on a detector's utility for specific clinical tasks. In order to simulate the effects of these physical parameters, image modification routines were developed capable of modifying a perfect input image to the resolution and noise characteristics specified by an input MTF and input NPS and included sampling effects such as aliasing. Experimental verification of these routines showed excellent correspondence between the resolution and noise properties of the output images and the input NPS and MTF curves. In order to investigate the effect of noise and resolution on signal detection tasks, high-quality images containing simulated lesions are altered by the image modification routines to the resolution and noise properties of two commercial digital radiographic detectors, one direct and one indirect. The sets of modified images had noise properties consistent with acquisitions at comparable, clinically relevant exposures for the two detectors. An observer study is performed with the resultant images followed by a Receiver Operating Characteristic (ROC) analysis. The results revealed the direct detector had a higher area under the ROC curve with a statistically significant difference for a 2.75 mm nodule (A_z = 0.90 vs. 0.76, p<0.01). The findings illustrated the connection between the physical performance metrics and utility for the signal detection tasks necessary for clinical use.

Charge transport and trapping-limited sensitivity and signal spreading over neighboring pixels of a direct conversion pixellated x-ray image detector are calculated by using the final trapped charge distributions across the photoconductor and the weighting potential of the individual pixel. The analytical expressions for the final trapped charge distributions across the photoconductor are derived by analytically solving the continuity equation for both types of carriers (electrons and holes). We calculate collected charges at different pixels by considering square pixels arranged in a two dimensional array. We calculate the amount of collected charge per unit incident radiation, the x-ray sensitivity, in terms of normalized parameters; (a) the normalized absorption depth (= absorption depth/photoconductor thickness), (b) normalized electron schubweg (schubweg/thickness), (c) normalized hole schubweg, and (d) normalized pixel pitch (pixel size/thickness). The composite (finely sampled) line spread function (LSF) is calculated by calculating collected charges at different pixels and by considering diagnostic x-ray irradiation along a line. The modulation transfer function (MTF) due to distributed carrier trapping is calculated by taking Fourier transform of composite LSF and correcting for the square sampling aperture. The charge transport and trapping-limited sensitivity and resolution of pixellated x-ray detectors mostly depend on the mobility and lifetime product of charges that move towards the pixel electrodes and the extent of dependence increases with decreasing normalized pixel pitch. The polarity (negative or positive signal) and the quantity of induced signals in the surrounding pixels depend on the bias polarity and the rate of trapping of both types of carriers. Optimal sensitivity and resolution can be attained by ensuring that the carriers which drift towards the pixel electrodes have a schubweg much longer than the sample thickness.

The most widely used pixel 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, the data line capacitance, resistance, and the column charge amplifiers add a large noise component to the PPS that reduces the minimum readable sensor input signal. Building upon previous research into active pixel sensor (APS) based amplified pixel readout circuits, this work investiates a current-mediated APS (C-APS) x-ray detection array for diagnostic medical imaging applications. Preliminary tests indicate linear performance, and a programmable circuits gain via choice of supply voltage and sampling time. In addition, the performance of C-APS amplified pixels is measured from both, a-Si TFT metastability and noise performance perspectives. Theory and measurements indicate that the C-APS pixel architecture is promising for diagnostic medical imaging modalities including low noise, real-time fluoroscopy.

Although physical measurements such as detective quantum efficiency (DQE) and modulation transfer function (MTF) provide insights, quantitative optimization of x-ray flat panel detectors requires consideration of image quality as perceived by humans. Using experiments and human observer models, we quantified image quality as the ability to detect targets such as stents and guidewires used in interventional angiography. We realistically simulated direct and indirect flat panel detectors over a range of exposures to create realistic fluoroscopy sequences. We performed objective, m-alternative adaptive forced choice experiments and applied models of human detection to fit almost all experiments and predict results for similar tasks and processing. With regard to pixel size, the best size at low fluoroscopic exposures for detecting a 400 μm guide wire with a realistic, indirect detector was at about 200 μm and depended upon such device parameters as electronic noise. For both indirect and direct detectors at higher exposures, noise was not limiting, and a small pixel was desirable. With regard to binning, we determined that binning is desirable at low exposures even for detection of small objects such as a guide wire. A new alternate binning method was found to have superior image quality by utilizing the ability of humans to temporally fuse alternating images binned at different orientations. Comparing to magnification of analog image intensifiers, we determined that flat panel images can be digitally magnified without loss of image quality at a decreased average exposure.

An innovative hydrogenated amorphous silicon (a-Si:H) p-i-n photodiode based x-ray detector for medical imaging applications has been developed in this work. Basically, the detector is a p-i-n photodiode, with a very simple modification by depositing a stacked silicon nitride (SiN_x) layer on the p-layer (n-i-p-SiN_x) or n-layer (p-i-n-SiN_x) of this diode. The dielectric layer functioned as the major charge storage element of the pixel, and p-i-n as the photon-charge converter, separately. The charge storage capacity is larger as the nitride layer is thinner. Consequently, dynamic range, linearity, and data retention of the image array were significantly improved. The novel detector also offers a scheme to independently optimize the photo sensitivity and charge storage capacity of a p-i-n photodiode based pixel. Instead of the conventional p-i-n photodiodes, the novel detectors are proposed to employ in the active matrix, flat-panel imager, with the favor that the signal readout electronics and the TFT driving circuitry are unchanged. The changes include only the bias voltage, whch as a bi-level waveform, as well as the timing for turning on/off the switching thin film transistors (TFTs). The fundamentals of the n-i-p-SiN_x and p-i-n-SiN_x detectors are addressed, and the performances of these two novel detectors and the conventional p-i-n photodiode are compared. Additionally, the different performances, such as the speed, between n-i-p-SiN_x and p-i-n-SiN_x will be particularly discussed.

Clinical evaluation results are presented using a large area, real time, amorphous selenium (a:Se), flat panel detector (FPD). The detector comprises of 1 mm thick amorphous selenium layer deposited onto a TFT panel that has a pixel pitch of .15 mm. The field of view of the detector is about 14” x 14” that is large enough to be used in R/F as well as general angiography application including digital subtraction angiography (DSA). Due to its high spatial resolution and low noise performance, it is shown that the detector is well suited to replace conventional image intensifier systems as well as film-screen systems.

Mercuric iodide (HgI₂) and lead iodide (PbI₂) have been under development for several years as direct converter layers in digital x-ray imaging. Previous reports have covered the basic electrical and physical characteristics of these and several other materials. We earlier reported on 5cm x 5cm and 10cm x 10cm size imagers, direct digital radiography X-ray detectors, based on photoconductive polycrystalline mercuric iodide deposited on a flat panel thin film transistor (TFT) array, as having great potential for use in medical imaging, NDT, and security applications. This paper, presents results and comparison of both lead iodide and mercuric iodide imagers scaled up to 20cm x 25cm sizes. Both the mercuric iodide and lead iodide direct conversion layers are vacuum deposited onto TFT array by Physical Vapor Deposition (PVD). This process has been successfully scaled up to 20cm x 25cm -- the size required in common medical imaging applications. A TFT array with a pixel pitch of 127 microns was used for this imager. In addition to increasing detector size, more sophisticated, non-TFT based small area detectors were developed in order to improve analysis methods of the mercuric and lead iodide photoconductors. These small area detectors were evaluated in radiographic mode, continuous fluoroscopic mode and pulsed fluoroscopic mode. Mercuric iodide coating thickness ranging between 140 microns and 300 microns and lead iodide coating thickness ranging between 100 microns and 180 microns were tested using beams with energies between 40 kVp and 100 kVp, utilizing exposure ranges typical for both fluoroscopic and radiographic imaging. Diagnostic quality radiographic and fluoroscopic images have been generated at up to 15 frames per second. Mercuric iodide image lag appears adequate for fluoroscopic imaging. The longer image lag characteristics of lead iodide make it only suitable for radiographic imaging. For both material the MTF is determined primarily by the aperture and pitch of the TFT array (Nyquist frequency of ~3.93 mm^-1 (127 micron pixel pitch).

Foremost among the promising imaging performance characteristics of cone-beam CT using flat-panel imagers is the ability to form volumetric images with soft-tissue contrast visibility in combination with sub-millimeter 3-D spatial resolution. Each of these two essential characteristics is intimately related to the spatial-frequency-dependent signal and noise transfer characteristics of the imaging system. Therefore a thorough, quantitative analysis of the 3-D noise-equivalent quanta (NEQ) and detective quantum efficiency (DQE) is essential to understanding the volumetric imaging performance of such systems, identifying the factors that limit performance, and revealing their full potential. This paper presents investigation of the 3-D NEQ and DQE for volume CT systems based on direct and indirect-detection flat-panel imagers (FPIs). Classical descriptions of image noise in transaxial CT are extended to the case of non-ideal 2-D detectors and 3-D image reconstruction. Definitions of NEQ and DQE are extended to provide figures of merit for 3-D imaging performance. A complex interplay between the system transfer functions, 3-D noise aliasing, and the 3-D DQE is uncovered, revealing several important phenomena: (1) 3-D NPS aliasing is a significant factor in the reconstruction process affecting DQE; (2) the degree of 3-D NPS aliasing is different for direct and indirect-detection FPIs and is related in non-trivially to the detector MTF and reconstruction filter; (3) the 3-D NEQ depends significantly on the choice of reconstruction filter -- in contrast to the classical notion that NEQ is independent of such -- and the effect is wholly attributable to 3-D NPS aliasing; and (4) the 3-D DQE for volume reconstructions is asymmetric between transverse and sagittal/coronal planes. Results for 3-D NEQ and DQE are integrated with 3-D spatial-frequency-dependent descriptions of imaging task (e.g., ideal observer detection and/or discrimination tasks) to yield the 3-D detectability index, helping to bridge the gap between NEQ and the performance of model observers.

We have proposed a CT system design to rapidly produce volumetric images with negligible cone beam artifacts. The investigated system uses a large array scanned source with a smaller array of fast detectors. The x-ray source is electronically steered across a 2D target every few milliseconds as the system rotates. The proposed reconstruction algorithm for this system is a modified 3D filtered backprojection method. The data are rebinned into 2D parallel ray projections, most of which are tilted with respect to the axis of rotation. Each projection is filtered with a 2D kernel and backprojected onto the desired image matrix. To ensure adequate spatial resolution and low artifact level, we rebin the data onto an array that has sufficiently fine spatial and angular sampling. Due to finite sampling in the real system, some of the rebinned projections will be sparse, but we hypothesize that the large number of views will compensate for the data missing in a particular view. Preliminary results using simulated data with the expected discrete sampling of the source and detector arrays suggest that high resolution (<0.5 mm in all directions) images can be obtained in a single rotation with the proposed system and reconstruction algorithm.

Multi-slice computed tomography (MSCT) is one of the most recent technological advancements in CT. Several experimental studies have been conducted in the last few years on the peformance of MSCT. However, there is a lack of theoretical analysis on the slice sensitivity profile (SSP) and noise performance. In this paper, we derive several closed-form expressions to characterize these performance parameters under different detector configurations and acquisition modes. Following the common practice, the expressions are explicitly described for regions near the iso-center, although the same approach can be used to describe system performances away from the iso-center. Our models are validated against phantom experiments.

Phase-contrast tomography is a non-interferometric imaging technique for reconstructing the refractive index distribution of a weakly absorbing object from a set of tomographic projection measurements. In many practical situations, it is desirable to minimize the field of view (FOV) of the imaging system in order to increase the spatial resolution of the reconstructed image. When the object of interest is larger than the FOV, the measured projections are truncated and one is faced with a local tomography reconstruction problem. In this work, we investigate the local tomography problem for phase-contrast tomography and propose a simple algorithm for reconstructing image boundaries from untruncated and truncated phase-contrast projection data. Simulation studies are presented to corroborate our theoretical findings.

A high-resolution video (HRV) based EPID that is capable of matching the high spatial resolution and SNR (signal to noise ratio) of a-Si flat panel devices was developed at Siemens OCS (Oncology Care Systems) for cone beam CT. This system is using a high resolution CCD camera (1300 x 1030). The optical components and scintillator screen were modified to generate high-resolution images. A Pips-Pro QC3 phantom was used to compare the spatial resolution and the contrast to noise ratio (CNR) of a-Si flat panel (1024x1024) and HRV system. The QC phantom was placed at the linear accelerator iso-center, and the detectors were placed 40cm below iso-center. The measured f50 for the Siemens a-Si flat panel and HRV are 0.49 lp/mm, and 0.43 lp/mm; respectively. The image of the flat panel had already been corrected for Offset, Gain and Defective pixels; however, no correction was performed on the HRV system. Due to the fast readout of HRV, small pixel size, and adjustable camera lens aperture, a thicker scintillator screen would have resulted in an increase in SNR without sensor saturation during radiation treatment imaging. This is one of the main advantages of this system compared to the flat panels, and it makes the system ideal for cone beam reconstructions as well as regular therapy imaging. A new electronic readout is implemented in HRV control circuit that will synchronize the image acquisition with the linear accelerator, and thereby increases the SNR.

The purpose of the study was to find out whether the image quality in full-field digital mammography can be improved while lowering the patient dose by removing the anti-scatter grid. Moreover, a fast approximate computational algorithm was developed for determining the scattered field in a real mammogram. The method is non-iterative, robust against noise, and works without modification for any scatter-to-primary ratio. Furthermore, it is computationally effective since it is based on fast Fourier transform (FFT). It was found out that the wide dynamic range of digital detectors leads to decrease in patient dose from 10.9% up to 46.6% at breast thickness of 2cm and from 0.8% up to 40.8% at breast thickness of 4cm depending on the efficiency of the removed grid. At constant patient dose the increase in contrast-to-noise ratio is 5.8% - 36.9% and 0.4%-30.0% accordingly at those two breast thickness. The convolution-based X-ray scatter model was considered. The developed scatter removal method was demonstrated with simulated mammograms and applied to clinical full-field digital mammograms acquired with a high-end digital flat panel detector based on amorphous selenium. Errors in reconstructed scattered fields were 0.3% in case of an ideal simulated mammogram and 7.4% in case of a real simulated mammogram (3cm breast). Applications where the scattered field needs to be determined include 3-D mammography and dual-energy breast imaging. In screening mammography gray-scale optimization eliminates the effect of scattering.

Traditional film/screen mammograms are obtained using Molybdenum or Rhodium target x-ray tubes. The energy spectrum from these sources matches the limited latitude of film/screen systems. For digital imaging systems, the latitude is linear over a wide range of exposures and arbitrary H&D curves can be obtained with image processing. This allows the recorded contrast to noise ratio (CNR) to be optimized by considering a wide range of radiographic techniques. For this work, we modeled the radiographic process for a digital (amorphous selenium) mammography system. The optimal CNR relative to dose was determined for several target/filter combinations, for a wide range of kVp values, and for varying breast thickness. The target/filter combinations included: Mo/Mo, Mo/Rh, Rh/Rh, W/Al, W/Mo, W/Ag, and W/Sn. As breast thickness increased, the use of a tungsten target with a tin filter resulted in a 34% improvement in CNR for the same dose to the breast when compared to the use of a Molybdenum target with a Molybdenum filter. Notably, the W/Sn target/filter combination resulted in a significantly lower mA-s for the same breast dose (2/3 to 1/5 lower for a breast thickness from 4 to 8cm). In mammography applications, use of a Tungsten tube rather than the traditional Molybdenum tube should lead to significant reductions in exposure time and tube heat while maintaining similar image quality and dose.

Our work is to investigate the imaging performance of a direct, full-field prototype digital mammography detector as a function of x-ray exposure and detector operational conditions for digital mammography and advanced applications such as tomosynthesis. Theoretical and experimental methods previously developed for the study of small-area prototype detectors have been applied to the investigation of spatial frequency dependent detective quantum efficiency [DQE(f)] of the full-field prototype detector, which has 2816 x 2048 pixels with 85 μm pixel size. The focus of our study is the impact of scaling up the detector design on imaging performance, e.g. electronic noise, readout rate and image artifacts. The results showed that DQE(f) of the full-field detectors is in the same range as that measured from the small-area prototype detector, both of which are superior to existing technologies based on indirect detection. However DQE(f) drops more rapidly than the small-area prototype as exposure decreases, which is to be expected from the higher electronic noise of the full-field detector. Lag and ghosting, both of which can introduce image artifacts, were studied at typical screening mammography image intervals. The effect of lag can be eliminated with frequent update of the offset images. Ghosting at x-ray dose equivalent to a single view mammogram is negligible.

A performance evaluation of small-area, high-spatial-resolution, active matrix flat-panel imager (AMFPI) prototypes, operated under mammographic conditions, is reported. These prototypes are based on two 512 x 512 pixel imagers employing novel designs to enhance signal performance for direct and indirect detection. The indirect detection prototype is based on a 75 μm pixel pitch array incorporating a continuous photodiode design, as opposed to the discrete photodiode design used in conventional flat-panel imagers. This array was coupled to a pair of commercially-available x-ray converters: (1) a 34 mg/cm² Gd₂O₂S:Tb phosphor screen (Min-R, Kodak) and (2) a 150 μm thick structured CsI:Tl scintillator on a fiber-optic plate (FOS-HL, Hamamatsu). The direct detection prototype is based on a 100 μm pixel pitch array and uses a 240 μm thick, high gain mercuric iodide (HgI₂) photoconductor. Measurements of sensitivity, MTF, NPS and DQE were performed with a 26 kVp mammography beam attenuated by a 4 cm BR-12 breast phantom at various radiation exposures. Results from empirical studies of sensitivity indicate that these imagers offer a substantial enhancement in signal over conventional flat-panel imagers. Measurements of DQE for the imagers show values greater than those obtained from high performance mammographic film-screen systems, under some conditions. These studies also show that the FOS-HL imager configuration despite its lower MTF, exhibits DQE performance (up to approximately 0.77) superior or equivalent to that of the Min-R configuration due to better optical properties of the converter. In addition, despite a smaller pixel pitch, both of these indirect detection configurations exhibit improved DQE in comparison to similar configurations employing a 97 μm pitch discrete photodiode design, especially at low exposures. Results of DQE measurements from the HgI₂ photoconductor prototype are promising (DQE values up to approximately 0.6). Finally, calculations of potential DQE performance for hypothetical 50 μm pitch imagers, employing similar novel designs, were performed. These calculations were based on the cascaded systems formalism and used realistic inputs derived from empirical measurements. The results predict that the HgI₂ configuration would provide high DQE performance (up to approximately 0.9), which would be largely unaffected by the magnitude of exposure, due to the high gain of the photoconductor. These calculations also indicate that the continuous photodiode configuration would provide high DQE (up to approximately 0.8), degraded only at low exposure by the effect of additive noise.

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) owing to an overlap in the appearances of benign and malignant lesions, and surrounding structure. The limitations accompanying conventional mammography is to be addressed by incorporating a cone beam volume CT imaging technique with a recently developed flat panel detector. Computer simulation and preliminary studies have 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 specimen experiment is performed to confirm the findings in the computer simulation and previous phantom studies using the current prototype cone beam volume CT scanner. The results indicate that the CBVCTBI technique effectively removes structure overlap and significantly improves the detectability of small breast tumors. More importantly, the results also demonstrate CBVCTBI offers good correlation with pathology images with the radiation dose level less than or equal to that of conventional mammography. The results from this study suggest that FPD-CBVCTBI is a potentially powerful breast-imaging tool.

As the use of digital radiographic equipment in the morphological imaging field is becoming largely diffuse, the research of new and more performing devices from public institutions and industrial companies is in constant progress. Many of these devices are based on solid-state detectors as X-ray sensors. Semiconductor pixel detectors, originally developed in the high energy physics environment, have been then proposed as digital detector for medical imaging applications. In this paper a digital single photon counting device, based on silicon and GaAs pixel detectors, is presented. The detector is a thin slab of semiconductor crystal equipped with an array of 64 by 64 square contacts, 170-μm side. The data read-out is performed by a VLSI integrated circuit named Photon Counting Chip (PCC), developed within the MEDIPIX collaboration. Each chip cell geometrically matches the sensor pixel. It contains a charge preamplifier, a threshold comparator and a 15 bits pseudo-random counter and it is coupled to the detector by means of bump-bonding. Most important advantages of such a system, with respect to a traditional X-rays film/screen device, are the wider linear dynamic range (3x10⁴) and the higher performance in terms of MTF and DQE. Electronics read-out performance as well as imaging capabilities of the digital device will be presented. Images of mammographic phantoms acquired with a standard mammographic tube will be compared with radiographs obtained with traditional film/screen systems.

One of the most demanding applications in dynamic X-Ray imaging is Digital Subtraction Angiography (DSA). As opposed to other applications such as Radiography or Fluoroscopy, there has been so far limited attempts to introduce DSA with flat detector (FD) technology: Up to now, only part of the very demanding requirements could be taken into account. In order to enable an introduction of FD technology also in this area, a complete understanding of all physical phenomena related to the use of this technology in DSA is necessary. This knowledge can be used for detector design and performance optimization. Areas of research include fast switching between several detector operating modes (e.g. switching between fluoroscopy and high dose exposure modes and vice versa) and non stability during the DSA run e.g. due to differences in gain between subsequent images. Furthermore, effects of local and global X-Ray overexposure (due to direct radiation), which can cause temporal artifacts such as ghosting, may have a negative impact on the image quality. Pixel shift operations and image subtraction enhance the visibility of any artifact. The use of a refresh light plays an important role in the optimization process. Both an 18x18 cm² as well as a large area 30x40 cm² flat panel detector are used for studying the various phenomena. Technical measurements were obtained using complex imaging sequences representing the most demanding application conditions. Studies on subtraction test objects were performed and vascular applications have been carried out in order to confirm earlier findings. The basis for comparison of DSA is, still, the existing and mature IITV technology. The results of this investigation show that the latest generation of dynamic flat detectors is capable of handling this kind of demanding application. Not only the risk areas and their solutions and points of attention will be addressed, but also the benefits of present FD technology with respect to state-of-the-art IITV technology regarding DSA will be discussed.

We developed prototype Digital Subtraction Angiography (DSA) System with a new large area FPD. Dynamic range, MTF, Contrast ratio and line noise were much improved. The improved FPD is a scintillator-type detector, and has a 40 x 30 cm active area, 2048 x 1536 matrix with 194um pixel pitch. The Prototype DSA system has two x-ray detectors, the FPD and the I.I.-CCD camera, and we can choose them on demand. All images captured from both detectors at 3 frames/sec in DSA mode and 30 frames/sec in Fluoroscopy mode are forwarded to our image-processing unit. We applied the new DSA system to more than 150 studies and compared the results with images from the I.I.-CCD. In DSA mode, FPD System, which has a wide dynamic range, large detecting area, and good contrast ratio yielded superior angiogram images compared with the I.I-CCD system. In Fluoroscopy mode, we improved line noise and increased the contrast of catheters and guide wires with a new image processing technique. With these improvements, the image quality of the FPD System is superior to the I.I.-CCD system at the exposure range of over 2uR/frame (17.4 nGy/frame).

The imaging performance of a 34.5 x 34.5 cm² direct conversion flat-panel detector with a 1 mm thick amorphous selenium layer was measured over the fluoroscopic exposure range (0.56 - 10.8 μR/frame). The pixels measured 300 x 300 μm. Measurements of the modulation transfer function (MTF), the noise power spectrum (NPS), and the detective quantum efficiency (DQE) were made. By comparing the MTF to the sinc function the measured effective fill factor of the active matrix was determined to be almost 100%. The electronic noise of the active matrix was measured and found to be 3800 electrons. The DQE(f) was found to be better than the expected sinc² function. This was due to the presence of a pre-sampling blur identified as charge trapping at an interface in the a-Se layers. At the highest exposure investigated, the DQE(0) was found to be less than the quantum efficiency and the difference was ascribed to a combination of the electronic noise, a small drop in sensitivity due to the charge trapping blur, and incomplete charge collection.

Conventional 2D photodiodes used in multislice CT detectors impose a limitation on the number of slices that can be achieved. The limitations encountered are in the number of metal lines that can be run between photodiode elements without significantly reducing the photodiode active area and in the density of wirebonds that can be accomodated at the edge of the chip. In addition, the photodiodes cannot easily be tiled in two directions due to the need for wirebonding on at least one edge. We have developed a 2D back illuminated photodiode array for multislice CT that overcomes the above limitations. The light sensitivity, sensitivity profiles, risetime, crosstalk, linearity and shunt resistance of the back illuminated photodiodes have been evaluated. The results show that these parameters either meet or exceed the specifications required for CT. The back illuminated photodiode represents an enabling technology for truly large area CT detection systems.

Integrated dose sensing in Flat Detectors allows a during pulse control of the X-ray illumination without the need for external dose sensing devices. Standard designs of Flat Detectors do not allow during pulse dose sensing since the information is collected from the pixels only in the read-out phase after the X-ray illumination. This paper introduces a special detector plate design for obtaining dose sensing information directly from the X-ray detector while the X-ray pulse is being applied. This dose sensing information is read at a lower spatial resolution than the actual X-ray image but with a sub-millisecond temporal resolution. The dose sensing operates without any additional radiation burden on the patient and without attenuation of the image information. Experimental results from a small area (4x4 cm²) detector are presented, including an analysis of noise, linearity and cross-talk.

Fan-beam coherent scatter computed tomography (CSCT) is a novel X-ray based imaging method revealing structural information of tissue under investigation. The source of contrast is the angular-dependent coherent scatter cross-section, which is determined by the molecular structure. In this work a phantom consisting of water, tricalcium phosphate, collagen and fat was used to investigate the contrast resolution of these four tissue constituents. Scatter projections were measured in fan-beam 3rd generation CT-geometry using an experimental demonstrator set-up equipped with a 4.5 kW DC power X-ray tube and photon-counting detectors. Reconstruction was performed using two algorithms, one based on algebraic reconstruction technique (ART) and the other based on filtered back-projection (FBP). The reconstruction results of the two techniques are compared. Furthermore, scatter functions of the four components were extracted from the 3D data sets and compared to previous measurements. The applicability of this technique for medical diagnosis is discussed.

Two X-ray phase-contrast imaging techniques are compared in a quantitative way for future mammographic applications: Diffraction Enhanced Imaging (DEI) and Propagation. The first uses an analyzer crystal after the sample acting as an angular filter for X-rays refracted by the sample. The latter simply uses the propagation (Fresnel diffraction) of the monochromatic and partially coherent X-ray beam over large distances. Experiments to compare both modalities have been performed at the Topography Beamline of the European Synchrotron Radiation Facility. The respective set-ups and experimental parameters are described in detail. Depending on the object properties, the two techniques present a difference in area contrast and edge visibility. DEI shows an enhancement of area contrast for positions of the crystal corresponding to the tails of its rocking curve (RC) and a similar increase but inverted is also visible at the peak of its RC. At the tails, the contrast is mainly produced by ultra small angle scattering, at the peak, it is due to absorption and scatter rejection by the analyzer. At the flanks, it may disappear when attenuation and scattering effects compensate each other. However, an enhancement of the object edges is clearly noticeable, which mainly corresponds to the refracted part. Propagation reveals an improvement of the edge visibility with the distance and shows negligible area contrast for non-absorbing, large structures.

Preliminary experiments have been carried out in order to evaluate the potential of the Diffraction Enhanced Imaging (DEI) technique in combination with contrast agents not based on X-ray absorption properties, but that provide strong scattering signals. The contrast agents tested in this study are microbubble echo-enhancing agents, usually used in ultrasound examinations, which are completely invisible with conventional X-ray absorption techniques. A DEI set-up has been implemented at the Medical Beamline at the synchrotron radiation facility ELETTRA (Trieste, Italy). The analyzer crystal is a single flat silicon crystal utilized in the [111] reflection. By shifting the analyzer crystal to different positions of the rocking curve it is possible to detect the scattered photons; in particular, if the sample consists of a large number of particles with size smaller than the pixel size of the detector, an overall effect can be visualized. Phantoms containing ultrasound contrast agents have been built and imaged at different angular positions of the analyzer crystal at 17 keV and 25 keV. For all the phantoms a much stronger contrast has been measured in comparison to the contrast evaluated from the images produced with normal absorption methods.

Bi-plane correlation imaging (BCI) is a new imaging approach that utilizes angular information from a bi-plane digital acquisition in conjunction with computer assisted detection (CAD) to reduce the degrading influence of anatomical noise in the detection of subtle lesions in planar images. An anthropomorphic chest phantom, supplemented with added nodule phantoms (5-13 mm at the image plane), was imaged from different posterior projections within a ±12° range by moving the x-ray tube vertically and horizontally with respect to the detector. Each image was analyzed using a basic front-end single-view CAD algorithm. The correlation of the suspect lesions from the PA view with those from each of the oblique views was examined using a priori knowledge of the acquisition geometry. The correlated suspect lesions were registered as positive. Using an optimum --3° vertical geometry and processing parameters, BCI resulted in 62.5% sensitivity, 1.5 FP/image, and 0.885 PPV. The corresponding values from the observer experiment were 56% sensitivity, 10.8 FP/image, and 0.45 PPV, respectively. Compared to single-view CAD results, the BCI reduced sensitivity by 20%. However, the corresponding reduction in FPs was notably higher (94%) leading to 140% improvement in the PPV. Changes in processing parameters could result in higher PPV and lower FP/image at the expense of lower sensitivity. Similar findings were indicated for small (5-9 mm) and large (9-13 mm) nodules, but the relative improvement was significantly higher for smaller nodules. (The research was supported by a grant from the NIH, R21CA91806.)

In a previous paper (SPIE Medical Imaging 2001), a dual energy method for bone densitometry using a 2D digital radiographic detector has been presented. In this paper, calcium content quantification performance of the approach is precised. The main challenge is to achieve quantification using scatter-corrected dual energy acquisitions. Therefore a scatter estimation approach, based on an expression of scatter as a functional of the primary flux, has been developed. This expression is derived from the Klein and Nishina equation and includes tabulated scatter level values. The calcium quantification performances are validated on two configurations. A first one is issued from criteria developed by the French “Groupe de Recherche et d'Information sur les Osteoporoses.” It is based on the use of a phantom made of five 3mm thick PVC sheets in the form of five steps, representing five different bone mineral density values, included in a lucite container filled with water. Additional lucite plates can be put over the phantom. This phantom has been used for evaluation of quantification robustness versus patient thickness and composition variations, and for accuracy evaluation. The second configuration, composed of small calcified objects (representative of lung nodules), is used for evaluating capacities to differentiate calcified from non calcified nodules and to test calcium content quantification performance.

Osteoporosis is a disease characterized by a decrease of bone mineral density as well as by architecture modification leading to an increase of fracture risk. This paper is part of study investigating the possibility to extract some structural parameters quantifying trabecular bone architecture from radiographies realized in vivo. The first step of the study is the definition of optimum radiographic conditions (X-ray spectrum, detector) as well as the development of adapted image processing tools to extract relevant indexes characterizing architecture. Therefore a simulation process computing synthetic radiographies from trabecular bone samples has been developed. This process is done in three distinct steps: (1) Computation of a very high spatial resolution 3D μCT volume of a human trabecular bone sample from a series of acquisitions with a microtomography system using synchrotron radiation. (2) Transformation of the μCT volume in a materials voxel volume. (3) Simulation of the radiography projection by using the X-ray radiographic simulation software Sindbad. The simulation software provides a lot of parameters which can be easily modified (spectra, materials, geometry, detector...) so that its use for an optimisation purpose is very practical. Comparison of simulated and experimental radiographies performed under synchrotron radiation microtomography configuration validates the accuracy of our simulation process. Simulated radiographs under several clinical conditions are also presented.

Our work investigates the effect on the zero-spatial-frequency DQE of indirect mammographic image receptors of the combined effects of x-ray beam hardening and light photon transport within the x-ray sensitive medium of the image receptor. Beam hardening, in this context, refers to the preferential absorption of low-energy x-ray photons at the surface of the detector on which the x-ray beam is incident, with deeper layers of the detector absorbing higher and higher average energies. The light photon transport properties of both powder and crystalline/columnar phosphors favor the collection at the light-sensitive element of the detector of light photons generated closest to that element. The net result of these two effects is to perform a weighting of the detected x-ray spectrum. For the standard back-screen configuration used in conventional mammography, low-energy x rays are more heavily weighted, matching the energy dependence of the signal to be detected. For a front-screen configuration, used in all existing indirect-detection digital mammographic systems, just the opposite is true. We have used the optical Monte Carlo simulation code DETECT-II to determine average light collection efficiency and optical pulse distributions for a Min-R screen in both front- and back-screen configurations, and for two thicknesses of CsI, as a function of x-ray energy. These data, along with appropriately hardened x-ray spectra for several anode and filter combinations and a range of tube voltages, were used as input to the task-dependent DQE theory of Tapiovaara and Wagner.

Optimization of the display of digital mammograms is an important challenge and requires knowledge of the characteristics of actual patient images. This work aims to create a description of some of the fundamental statistical properties of a large volume of images acquired on an FDA approved device as used in clinical practice. 4569 digital mammograms (1246 patients) were acquired between October 2001 and August 2002 on a GE Senograph 2000D at Sunnybrook and Women's College Health Sciences Centre. Images were saved in "raw" format. The breast was then segmented from the background on the image using a technique based on thresholding and some connectivity rules. The histogram of pixel values in the breast only is then calculated for both the raw and processed versions of the image. The region of constant thickness, where the breast is in contact with the compression paddle, was also segmented from the CC view raw images. The histogram and statistical properties in this central region were also calculated. Assorted statistical descriptors of the histograms were examined (dynamic range, mean, standard deviations, median and mode). The effect of image processing on the dynamic range in the periphery and central area of the breast was evaluated. The results were compared against the automatic exposure algorithm and acquisition parameters, projection (view) and breast thickness.

Ideally, the gray level changes in a Contrast-Enhanced Digital Mammography (CEDM) sequence reflect the uptake and wash-out of contrast medium in the breast. While insignificant in standard mammography, gray level variations with time caused both by patient and system related factors, have been observed in clinical CEDM sequences. We have acquired phantom image series on digital mammography systems using a Mo/Cu anode-filter combination and a tube voltage between 45 and 49 kVp, in order to derive a model for gray level change with time as a function of system parameters. The gray level variation exhibits a fair degree of inter-series repeatability, and strongly depends on the dose received by the detector and timing of the image acquisition series. Moreover, for tissue-equivalent compositions, the relative gray level change with respect to the first image does not depend on the composition. We designed a calibration procedure that can be used to compensate for the tiny system-dependent signal variation that has been observed. A global reduction of 80-93% of the variation has been demonstrated in sequences acquired on a breast shaped phantom. Local improvement is effective across the whole field of view. When imaging iodine inserts (0.5-2 mg/cm2 concentration), the calibration increases the constancy with time of iodine signal on subtracted sequences by a factor of 4 (median value).

Software has been developed to simulate a cone-beam CT mammography imaging system that consists of an x-ray tube and a flat-panel detector that rotate simultaneously around the pendant breast. The simulation uses an analytical expression or ray-tracing to generate projection sets of breast phantoms at 1 keV intervals dictated by the input x-ray energy spectra. The x-ray focal spot was modeled as having a Gaussian distribution. The detector was modeled as an amorphous silicon (aSi:H) flat-panel imager that uses a structured CsI scintillator. Noise propagation through the detector was simulated by modeling statistical variations of the projection images at each energy interval as a scaled Poisson process. Scintillator blurring was simulated by using an empirically determined modulation transfer function. After introducing noise and detector blur, projection sets simulated at each energy were then combined and reconstructed using Feldkamp's cone-beam reconstruction algorithm. Using this framework, the effects of a number of acquisition and reconstruction parameters can be investigated. Some examples are shown including the impact of the kVp setting and the number of projection angles on the reconstructed image.

This paper describes a unique system for constructing a three-dimensional volume from a set of two-dimensional (2D) x-ray projection images based on optical aperture theory. This proprietary system known as Tuned-Aperture Computed Tomography (TACT) is novel in that only a small number of projections acquired from arbitrary or task-specific projection angles is required for the reconstruction process. We used TACT to reconstruct a simulated phantom from seven 2D projections made with the x-ray source positioned within 30 degrees of perpendicular to a detector array. The distance from the x-ray source was also varied to change the amount of perspective distortion in each projection. Finally, we determined the reconstruction accuracy of TACT and compared it to that of a conventional tomosynthesis system. We found the reconstructed volumetric data sets computed with TACT to be geometrically accurate and contain significantly less visible blurring than a similar data set computed with the control technique.

Radiology-based lung-cancer detection is a high-contrast imaging task, consisting of the detection of a small mass of tissue within much lower density lung parenchyma. This imaging task requires removal of confounding image details, fast image acquisition (< 0.1 s for pericardial region), low dose (comparable to a chest x-ray), high resolution (< 0.25 mm in-plane) and patient positioning flexibility. We present an investigation of tomosynthesis, implemented using the Scanning-Beam Digital X-ray System (SBDX), to achieve these goals. We designed an image-based computer model of tomosynthesis using a high-resolution (0.15-mm isotropic voxels), low-noise CT volume image of a lung phantom, numerically added spherical lesions and convolution-based tomographic blurring. Lesion visibility was examined as a function of half-tomographic angle for 2.5 and 4.0 mm diameter lesions. Gaussian distributed noise was added to the projected images. For lesions 2.5 mm and 4.0 mm in diameter, half-tomographic angles of at least 6° and 9° respectively were necessary before visualization of the lesions improved. The addition of noise for a dose equivalent to 1/10 that used for a standard chest radiograph did not significantly impair lesion detection. The results are promising, indicating that lung-cancer detection using a modified SBDX system is possible.

Digital tomosynthesis is a method that enables the retroactive reconstruction of arbitrary tomographic planes in an object from a finite series of digital projection radiographs, acquired with limited angle tube movement. Conventional tomosynthesis suffers from the presence of blurring artifacts, created by objects located outside of each reconstructed plane. Matrix inversion tomosynthesis (MITS) utilizes known acquisition geometry to solve directly for the unwanted out-of-plane blur artifacts, thus enabling their removal. This paper examines practical strategies for the implementation of MITS in a clinical setting, on a flat-panel fast-readout detector, with the aim of minimizing procedure time and image reconstruction artifacts concurrently. Topics include a comparison of continuous vs. incremental tube motion, the presence of reconstruction artifacts due to error in computing the x-ray tube location, the effect of scrubbing the detector between projections to reduce image retention, and a method for accounting for data that gets projected off the detector. We conclude that MITS is robust enough to be clinically applicable, even under less-than-ideal conditions. Rapid image acquisition with continuous tube movement and no detector scrubbing is clinically desirable for MITS imaging of the chest, where patient motion is a concern. Knowledge of the source-detector geometry can be satisfactorily determined via either a lead fiducial marker placed on the patient, or a tube motion device with sufficient precision and accuracy. Extrapolation of data at the top and bottom of projection images provides excellent amelioration of image truncation artifacts.

An information-theoretic framework for assessing and predicting ultrasound system performance for detection tasks is outlined. Current models of image quality for ultrasound detection tasks make some stringent assumptions, including large target area, that place limits on the applicability of the theory. New models of image quality for ultrasound systems are proposed based on the ideal observer that account for noise, system, and object properties. One result is an expression for the ideal observer detectability that is a generalization of Noise-Equivalent Quanta (NEQ), a measure used by photon imaging modalities. The detection signal-to-noise ratio is shown to be an integration of the generalized NEQ weighted by the spectral variance of the target signal (in contrast to the squared magnitude of the Fourier transform of the signal, as is the case with other modalities). This reflects that ultrasound systems are sensitive not to the magnitude of the medium parameters but rather the variance (spatial fluctuations) of these quantities. The resulting framework is amenable to measurement and prediction of system performance. The theory is used to predict the information content of the ultrasonic beam at various field points. New strategies are revealed for processing RF data that could improve detection of lesions.

Air trapping is a prominent finding in small airway disease (SAD) of the lungs. To investigate the feasibility of accurate, automated assessment of air-trapping from low-dose CT, we compare visual scoring by expert radiologists to a conventional method of automated assessment as well as two novel methods. The conventional method, the "density mask" method, has been reported to correlate weakly but significantly with visual scoring on normal-dose CT. While we were unable to reproduce these results on our low-dose scans, our two novel methods showed some promise. More study on larger data sets is required to determine the optimal analysis method.

End-to-end system studies in the digital acquisition of photon-limited images inevitably lead to the question of setting technical design goals in order to achieve overall linearity. But this itself poses the specific question of which key print/display variables should, in ideal circumstances, be linear with which key scene-acquisition parameters. Within the digital context, intuitive answers gleaned from traditional analog approaches to this question can vary from the confusing to the misleading, and digital attempts at direct emulation of analog systems may in fact off-set the potential advantages implicit in the newer digital technologies. The author has used a digital end-to-end signal-to-noise ratio model to consider some aspects of this linearity question, and to offer solutions based on the combination of SNR optimization appropriate during image acquisition, and the human visual-response characteristics appropriate during display. Some results and conclusions are discussed here, especially those relating to the optimum digital strategy for scene-acquisition.

There are many aspects that influence and deteriorate the detection of pathologies in X-ray images. Some of those are due to effects taking place in the stage of forming the X-ray intensity pattern in front of the x-ray detector. These can be described as motion blurring, depth blurring, anatomical background, scatter noise and structural noise. Structural noise results from an overlapping of fine irrelevant anatomical structures. A method for measuring the combined effect of structural noise and scatter noise was developed and will be presented in this paper. This method is based on the consideration that within a pair of projections created after rotation of the object with a small angle (which is within the typical uncertainty in positioning the patient) both images would show the same relevant structures whereas the projection of the fine overlapping structures will appear quite differently in the two images. To demonstrate the method two X-ray radiographs of a lung phantom were produced. The second radiograph was achieved after rotating the lung by an angle of about 3. Dyadic wavelet representations of both images were regarded. For each value of the wavelet scale parameter the corresponding pair of approximations was matched using the cross correlation matching technique. The homologous regions of approximations were extracted. The image containing only those structures that appear in both images simultaneously was then reconstructed from the wavelet coefficients corresponding to the homologous regions. The difference between one of the original images and the noise-reduced image contains the structural noise and the scatter noise.

The modulation transfer function (MTF) is a fundamental measure of spatial resolution of an imaging system, and can be measured by imaging a slit, edge or a bar-pattern. In portal imaging, the MTF has been measured using the slit and edge techniques, requiring very thick collimation to minimize the effect of megavoltage scatter and laborious alignment procedures. A simpler and quicker method for measuring MTF is presented: the bar-pattern. This method has been successfully used in diagnostic imaging. In portal imaging, this method is sensitive to the measurement of MTF(0) due to lateral scattering of megavoltage x-rays. A lack of a precise measurement of MTF(0) can lead to an over-estimation of MTF. The slit and bar-pattern techniques were studied using Monte Carlo simulations on a kinestatic charge detector (KCD), which uses a slot photon beam and a scanning high-pressure gas multi-ion chamber. The experimental condition for measuring MTF(0) was determined. MTF measurements using the slit and bar techniques, as well as those from Monte Carlo simulations, were subsequently observed to be in good agreement (i.e. one standard deviation of measurement). The bar-pattern method, being easier and simpler than the slit or edge techniques, provides a fast MTF measurement.

We have developed a full-scale image quality (IQ) simulation model as a tool for X-ray system design, image quality optimization and patient dose reduction. The IQ model supports the (de-)composition of system level requirements and simulates several types of automatic X-ray control technique. The model is implemented in LabVIEW. The X-ray system is modeled in distinguishable components and processes, which allows isolation of sub-systems and exclusion of devices. All relevant patient dose and IQ items such as contrast, sharpness, lag and noise are calculated and additionally combined in IQ "figures of merit" (FOM). Some characteristic application examples will be presented: In a general image magnification study we compare magnification techniques, such as geometric enlargement, image intensifier zooming and digital processing. In an optimization study we apply a new IQ FOM that contains not only imaging properties of the system, but also detail information in terms of material, size and thickness. Combining the IQ simulation model with a Pareto trade-off algorithm appears to be a promising optimization approach. In addition to the mentioned employment, the IQ simulation model is also suitable for comparison studies on the performance of flat detectors versus image intensifier television detectors, application related studies and fine tuning of specific settings and adjustments, design of test objects and development of measuring methods.

The first demonstrations of terahertz imaging in biomedicine were made several years ago, but few data are available on the optical properties of human tissue at terahertz frequencies. A catalogue of these properties has been established to estimate variability and determine the practicality of proposed medical applications in terms of penetration depth, image contrast and reflection at boundaries. A pulsed terahertz imaging system with a useful bandwidth 0.5-2.5 THz was used. Local ethical committee approval was obtained. Transmission measurements were made through tissue slices of thickness 0.08 to 1 mm, including tooth enamel and dentine, cortical bone, skin, adipose tissue and striated muscle. The mean and standard deviation for refractive index and linear attenuation coefficient, both broadband and as a function of frequency, were calculated. The measurements were used in simple models of the transmission, reflection and propagation of terahertz radiation in potential medical applications. Refractive indices ranged from 1.5 ± 0.5 for adipose tissue to 3.06 ± 0.09 for tooth enamel. Significant differences (P < 0.05) were found between the broadband refractive indices of a number of tissues. Terahertz radiation is strongly absorbed in tissue so reflection imaging, which has lower penetration requirements than transmission, shows promise for dental or dermatological applications.

In this paper I show with phantom and animal experiments a non-invasive and quantitative method for measuring the conductivity and dielectric distributions based on high field magnetic resonance imaging. High field MRI is accompanied by significant RF wave propagation effects. They are observed as phase and magnitude variations of the image that cannot be removed by optimizing the static field homogeneity, or by improving the RF coils. These variations reflect the RF field distribution in the sample, and in fact obey a modified Helmholtz equation. By mapping both the phase and magnitude of the field with MRI techniques, both the conductivity and the dielectric constant are determined non-invasively. In phantom experiments at 1.5 tesla, conductivity values were measured at 4 mm resolution to 0.5 S/m accuracy. At 4.7 tesla, the accuracy was improved to 0.2 S/m, and the dielectric constant was measured to an accuracy of 5 (relative to vacuum) for 2cm regions.

In recent years, indirect detection active matrix flat-panel imagers (AMFPIs) have become the gold standard in radiotherapy imaging. The excellent imaging performance of AMFPI-based electronic portal imaging devices (EPIDs) can be attributed to their ability to perform x-ray quantum limited imaging under radiotherapy conditions. However, all current commercial (AMFPI and non-AMFPI) EPIDs use only approximately 1% to 2% of the incident radiation. In this work, strategies to significantly improve the overall performance of indirect detection AMFPI-based EPIDs through novel designs, are presented. Specifically, the focus of this work is on thick, structured scintillators, which achieve high x-ray detection efficiency while simultaneously maintaining spatial resolution. Fundamental signal and noise measurements using prototype, small area, structured scintillators are reported. In addition, theoretical calculations were performed in order to estimate the signal and noise properties of various structured scintillator configurations. Results from theory and experiments were used to estimate the frequency-dependent detective quantum efficiency (DQE). Results suggest that these designs could yield DQE values that are significantly higher than those for current commercial EPID technologies.

A novel CT detector based on CMOS photodiodes has been developed. A detector module comprises two identical photosensor arrays mounted to a ceramic substrate. Each sensor has a matrix of 20 by 10 pixels. Pixels are 1 mm (channel direction) x 1.8 mm (slice) large and consist of a photodiode, charge integration unit and a sample and hold stage. An automated switching between a low and a high sensitivity mode allows for a dynamic range of 17 bits. The integrated signals are read out, transferred to a printed circuit board (at a rate of 2463 Hz per pixel) and here converted into a digital data stream. The structured cadmium tungstate scintillator features lead stripes between pixels to reduce x-ray crosstalk and to shield the underlying in-pixel electronics. During assembling care was taken to ensure that the lead stripes of the scintillator entirely cover the pixel electronics underneath. Several prototype modules have been assembled and their performance concerning linearity, noise, crosstalk, and temperature dependence has been evaluated.

We have been investigating whether a microlens layer placed between the phosphor and the photodetector can improve indirect detection x-ray imagers. Using a simulation study, we analyzed the light collection properties of the proposed imager taking into account Fresnel reflection and transmission properties of the lenses and the screen. A digital x-ray imager combining an 82-μm-thick Gd₂O₂S:Tb phosphor screen, a fused silica microlens layer, and a 127-μm pixel pitch photodetector (optical fill factor of 57%) were modeled. The light collection for the prototypes varied from 53% to 69% for lens thicknesses ranging from 10 to 50 μm. The full-width half-maximum (FWHM) of the light spread function ranged from 177-192 μm. 4-8% of the light was reflected back into the phosphor screen when correctly taking into account Fresnel reflections for these prototype imagers. In comparison, 56% of the light was collected and the FWHM of the light spread function was 174 μm for a conventional imager with the screen in direct contact with the photodetector. We observed that the light collection was overestimated by 6-9% but the spread functions were basically unaffected when the Fresnel assumption was not utilized in the simulations. This study shows that a properly designed microlens layer can more than offset Fresnel losses, thereby producing an improved digital x-ray imager.

Megavoltage x-ray imaging suffers from relatively poor contrast and spatial resolution compared to diagnostic kilovoltage x-ray imaging due to the dominant Compton scattering in the former. Recently available amorphous silicon/selenium based flat-panel imagers overcome many of the limitations of poor contrast and spatial resolution that affect conventional video based electronic portal imaging devices (EPIDs). An alternative technology is presented here: kinestatic charge detection (KCD). The KCD uses a slot photon beam, high-pressure gas (xenon, 100 atm) and a multi-ion rectangular chamber in scanning mode. An electric field is used to regulate the cation drift velocity. By matching the scanning speed with that of the cation drift, the cations remain static in the object frame of reference, allowing temporal integration of the signal. KCD imaging is characterized by reduced scatter and a high signal-to-noise ratio. Measurements and Monte Carlo simulations of modulation transfer function (MTF), noise power spectrum (NPS) and the detective quantum efficiency (DQE) of a prototype small field of view KCD detector (384 channels, 0.5 mm spacing) were carried out. Measurements yield DQE[0]=0.19 and DQE[0.5cy/mm]=0.01. KCD imaging is compared to film and commercial EPID systems using phantoms, with the KCD requiring an extremely low dose (0.1 cGy) per image. A proposed cylindrical chamber design with a higher ion-collection depth is expected to further improve image quality (DQE[0]>0.25).

This article reviews the state of the Noise Power Standard being drafted by Task Group No. 16 for the American Association of Physicists in Medicine. The Standard is intended to represent a consensus on acceptable practices in the measurement and reporting of noise power spectra for digital radiographic imaging devices based on single projections and to contain informative sections which will be of use to those not completely familiar with the measurement and interpretation of noise power spectra. Several of the issues considered by the committee are reviewed, including issues of conditioning and windowing data, issues specific to several modalities, and various methods of data presentation. A note on the historical background of noise power measurements and a brief discussion of possible avenues for future research is included.

We have applied phase imaging on digital mammography to investigate adequate contrast of printed images for digital phase contrast mammography using a practical molybdenum X-ray tube. Phase contrast mammography procedures were performed with defined air gap (e.g., 0.6 m) configuration using customized mammography equipment and a computed radiography (CR) system. Magnified (x2) phase contrast images acquired with 0.0875mm per pixel were mapped onto the laser imager resolution at 0.04375mm per pixel for printing life-size object on wet processing silver halide recording film. For contact mammography of screen-film system, we used MinR2000 system as a baseline method. ACR 156 phantom printed images with contrasts of 2.8, 3.7, 4.9, 5.7 and 6.7 were evaluated by five radiologists. The ACR scores for the life-size image based on the 2 times magnified phase contrast image acquired by the computed radiography were higher than the scores of MinR2000 image, when the contrast of printed images for both methods was 3.7. The ACR scores were lower in the low contrast images (i.e., 2.8) than its higher contrast counterparts (i.e., >= 3.7) for all techniques used. The detectability improvement should be due to higher spatial resolution and lower noise in the phase contrast images.

Columnar CsI(Tl) screens are now routinely used for digital x-ray imaging in a wide variety of applications such as mammography, dental radiography, and non-destructive testing. While commercially available CsI(Tl) screens exhibit excellent properties, it is possible to significantly improve their performance. Here, we report on a new design of a columnar CsI(Tl) screen. Specifically, columnar CsI(Tl) screens were subjected to mechanical pixelation for minimizing the long range spread of scintillation light within the film, thus enhancing spatial and contrast resolution, and increasing the detective quantum efficiency (DQE(f)) of the digital imaging detector. To date we have fabricated up to 200 μm thick pixelated CsI(Tl) screens for mammography, and characterized their performance using a CCD camera. This paper presents a comparison of the new pixelated CsI(Tl) screens, conventional columnar CsI(Tl) screens, and Gd₂O₂S(Tb) screens. The data show that pixelated screens substantially improve the DQE(f) of the digital imaging system.

Sectra Microdose is the first single photon counting mammography detector. An edge-on crystalline silicon detector is connected to application specific integrated circuits that individually process each photon. The detector is scanned across the breast and the rejection of scattered radiation exceeds 97% without the use of a Bucky. Processing of each x-rays individually enables an optimization of the information transfer from the x-rays to the image in a way previously not possible. Combined with an almost absence of noise from scattered radiation and from electronics we foresee a possibility to reduce the radiation dose and/or increase the image quality. We will discuss fundamental features of the new direct photon counting technique in terms of dose efficiency and present preliminary measurements for a prototype on physical parameters such as Noise Power Spectra (NPS), MTF and DQE.

The GE Senographe 2000D, the first full field digital mammography system based on amorphous Silicon (a-Si) flat panel arrays and a Cesium-Iodide (CsI) scintillator, has been in clinical use for several years. The purpose of this paper is to demonstrate and quantify improvements in the detective quantum efficiency (DQE) for both typical screening and ultra-low exposure levels for this technology platform. A new figure of merit, the electronic noise factor, is introduced to explicitly quantify the influence of the electronic noise, conversion factor, modulation transfer function (MTF), and pixel pitch towards the reduction of DQE at low exposure levels. Methods to improve the DQE through an optimization of both the flat panel design and the scintillator deposition process are discussed. The results show a substantial improvement in the DQE(f) at all frequencies and demonstrate the potential for DQE(0) to exceed 80%. The combination of high DQE at ultra low exposures and the inherent fast read-out capability makes this technology platform ideal for both current clinical procedures and advanced applications that may use multiple projections (tomosynthesis) or contrast media to enhance digital mammography.

Direct conversion detectors have the potential to provide very high resolution and high detective quantum efficiency (DQE). Selection of a material that is appropriate for the task is dictated by the material properties. A linear cascaded systems analysis of DQE is used to predict the performance of several detector materials such as amorphous Se, CdZnTe, and PbI₂. A model is used to predict the spatial frequency-dependent DQE(f) for each material. This model includes: (1) x-ray absorption, (2) K fluorescence, (3) conversion gain, and (4) incomplete charge collection. A depth-dependent approach is used to account for gain variations and charge transport characteristics that change throughout the detector. In the model a parallel cascade, and non-elementary stages are used to model the effect of K-fluorescence reabsorption followed by incomplete charge collection. The DQE(f) is determined across an x-ray energy range of 10 to 100 keV for each material under typical bias conditions ranging from 0.1 V/μm to 10 V/μm. K-fluorescence escape and reabsorption blurring can cause marked reductions in the DQE(f). It is further reduced by incomplete charge collection which can theoretically decrease the DQE(f) by as much as 50% in extreme situations. This model will help determine key factors that will influence material selection for direct conversion x-ray systems.

The purpose of this study was to determine the optimum beam quality in terms of kVp and filtration for a tungsten-target x-ray source for a newly developed cone beam computed mammotomography application. The optimized beam is expected to yield enhanced image quality along with a low dose, equal to or less than that of dual view x-ray mammography. X-ray spectra were computer generated for a range of tube potentials, filter materials, and filter and breast thicknesses. The uncompressed breast was modeled from 8, 12, and 16 cm thick tissue, and breast lesions were modeled as a 0.5 cm thick striated muscle mass and a 0.02 cm thick microcalcification. The detector was modeled as a digital flat-panel detector with a 0.06 cm thick CsI x-ray absorption layer. Figures of merit computed included the ratio of the mean beam energy post-breast to pre-breast as an index of beam hardening, ratio of lesion contrasts with and without filtering, and SNR/exposure. Tube potentials between 50 and 70 kVp provided continuous spectra that, when filtered with Z-filters between 56 and 62 yielded quasi-monochromatic x-ray spectra with optimal SNR/exposure and contrast while providing minimum beam hardening. Figures of merit improve with increasing filter thickness, but diminishing returns are seen beyond the 500th value attenuation layer. While uncompressed breast thickness affected absolute values of the measurement criteria, they had little effect on beam hardening and contrast ratio and did not alter the optimal operating range. Filter thickness near the 500th value layer is expected to be well within the operating range for an intended commercial x-ray tube, indicating that these highly attenuating filters can provide superior performance for mammotomography.