J. Eric Tkaczyk

Pulse pileup statistics for energy discriminating photon counting x‐ray detectors

PURPOSE Energy discriminating photon counting x-ray detectors can be subject to a wide range of flux rates if applied in clinical settings. Even when the incident rate is a small fraction of the detectors maximum periodic rate No, pulse pileup leads to count rate losses and spectral distortion. Although the deterministic effects can be corrected, the detrimental effect of pileup on image noise is not well understood and may limit the performance of photon counting systems. Therefore, the authors devise a method to determine the detector count statistics and imaging performance. METHODS The detector count statistics are derived analytically for an idealized pileup model with delta pulses of a nonparalyzable detector. These statistics are then used to compute the performance (e.g., contrast-to-noise ratio) for both single material and material decomposition contrast detection tasks via the Cramdr-Rao lower bound (CRLB) as a function of the detector input count rate. With more realistic unipolar and bipolar pulse pileup models of a nonparalyzable detector, the imaging task performance is determined by Monte Carlo simulations and also approximated by a multinomial method based solely on the mean detected output spectrum. Photon counting performance at different count rates is compared with ideal energy integration, which is unaffected by count rate. RESULTS The authors found that an ideal photon counting detector with perfect energy resolution outperforms energy integration for our contrast detection tasks, but when the input count rate exceeds 20% N0, many of these benefits disappear. The benefit with iodine contrast falls rapidly with increased count rate while water contrast is not as sensitive to count rates. The performance with a delta pulse model is overoptimistic when compared to the more realistic bipolar pulse model. The multinomial approximation predicts imaging performance very close to the prediction from Monte Carlo simulations. The monoenergetic image with maximum contrast-to-noise ratio from dual energy imaging with ideal photon counting is only slightly better than with dual kVp energy integration, and with a bipolar pulse model, energy integration outperforms photon counting for this particular metric because of the count rate losses. However, the material resolving capability of photon counting can be superior to energy integration with dual kVp even in the presence of pileup because of the energy information available to photon counting. CONCLUSIONS A computationally efficient multinomial approximation of the count statistics that is based on the mean output spectrum can accurately predict imaging performance. This enables photon counting system designers to directly relate the effect of pileup to its impact on imaging statistics and how to best take advantage of the benefits of energy discriminating photon counting detectors, such as material separation with spectral imaging.

Performance of a flat panel cardiac detector

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%.

Simulation of CT dose and contrast-to-noise as function of bowtie shape

Dose is becoming increasingly important for computed tomography clinical practice. It is of general interest to understand the impact that system design can have on dose and image quality. This study addresses the effect of bowtie shape on the dose and contrast-to-noise across the field of view. Simulation of the CT acquisition is used to calculate the energy deposition throughout a numerical phantom for a set of relevant system operating parameters and bowtie shapes. Mean absorbed dose is calculated by summing over the phantom volume and is compared with other typical dose specifications. A more aggressive attenuation profile of the bowtie which offers higher attenuation in the periphery of the field of view can offer the benefit of lower dose but at the expense of reduced contrast-to-noise at the edge of the cross-sectional image.

Thickness-dependent scatter correction algorithm for digital mammography

We have implemented a scatter-correction algorithm (SCA) for digital mammography based on an iterative restoration filter. The scatter contribution to the image is modeled by an additive component that is proportional to the filtered unattenuated x-ray photon signal and dependent on the characteristics of the imaged object. The SCAs result is closer to the scatter-free signal than when a scatter grid is used. Presently, the SCA shows improved contrast-to-noise performance relative to the scatter grid for a breast thickness up to 3.6 cm, with potential for better performance up to 6 cm. We investigated the efficacy of our scatter-correction method on a series of x-ray images of anthropomorphic breast phantoms with maximum thicknesses ranging from 3.0 cm to 6.0 cm. A comparison of the scatter-corrected images with the scatter-free signal acquired using a slit collimator shows average deviations of 3 percent or less, even in the edge region of the phantoms. These results indicate that the SCA is superior to a scatter grid for 2D quantitative mammography applications, and may enable 3D quantitative applications in X-ray tomosynthesis.

Quantization of liver tissue in dual kVp computed tomography using linear discriminant analysis

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

Dual kVp material decomposition using flat-panel detectors

In addition to a conventional Computed Tomography (CT) image, dual energy (dual kVp) imaging can be used to generate an image of the same anatomy that represents the equivalent density of a particular material, for example, calcium, iodine, water, etc. This image can be used to improve the differentiation of materials as well as improve the accuracy of absolute density measurements in a cross-sectional image. It is important to understand the certainty of the estimation of the density of the material. Both simulations and measurements are used to quantify these errors. Data are acquired using a flat-panel based volumetric CT system, by taking two scans and adjusting the maximum energy of the source spectrum (kVp). Physics based simulations are used to compare with the measurements. After validating the simulation algorithms, the accuracy of the dual kVp method is determined using the simulations in a perturbation study.

Modeling the x-ray energy characteristics of DQE for full-field digital mammography

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.

Fast kVp switching CT imaging of a dynamic cardiac phantom

Dual energy CT cardiac imaging is challenging due to cardiac motion and the resolution requirements of clinical applications. In this paper we investigate dual energy CT imaging via fast kVp switching acquisitions of a novel dynamic cardiac phantom. The described cardiac phantom is realistic in appearance with pneumatic motion control driven by an ECG waveform. In the reported experiments the phantom is driven off a 60 beats per minute simulated ECG waveform. The cardiac phantom is inserted into a phantom torso cavity. A fast kVp switching axial step and shoot acquisition is detailed. The axial scan time at each table position exceeds one heart cycle so as to enable retrospective gating. Gating is performed as a mechanism to mitigate the resolution impact of heart motion. Processing of fast kVp data is overviewed and the resulting kVp, material decomposed density, and monochromatic reconstructions are presented. Imaging results are described in the context of potential clinical cardiac applications.

Low cost, thick CZT spectroscopic detectors by sensor-pack construction of multiple tile pieces

Feasibility of a potentially low-cost, large volume radiation detector device is demonstrated. The spectral response performance is mapped per pixel for CZT sensor pack modules build from assemblies of 5mm thick CZT tiles. Device height between 10 and 20mm are compared to modules built with monolithic, 10mm thick parts. Module quality improved over several trials and passing of one years time due to improved assembly process and CZT tile quality. A scaling analysis of the yield and cost suggest that tall devices are optimally built from sensor pack tiles. However, as quality of the process improves and defect density decreases there is a cross-over point where the monolithic detector becomes more economical due to a decrease in sidewall preparation process.

Detectability and image quality metrics based on robust statistics: following non-linear, noise-reduction filters

Non-linear image processing and reconstruction algorithms that reduced noise while preserving edge detail are currently being evaluated in medical imaging research literature. We have implemented a robust statistics analysis of four widely utilized methods. This work demonstrates consistent trends in filter impact by which such non-linear algorithms can be evaluated. We calculate observer model test statistics and propose metrics based on measured non-Gaussian distributions that can serve as image quality measures analogous to SDNR and detectability. The filter algorithms that vary significantly in their approach to noise reduction include median (MD), bilateral (BL), anisotropic diffusion (AD) and total-variance regularization (TV). It is shown that the detectability of objects limited by Poisson noise is not significantly improved after filtration. There is no benefit to the fraction of correct responses in repeated n-alternate forced choice experiments, for n=2-25. Nonetheless, multi-pixel objects with contrast above the detectability threshold appear visually to benefit from non-linear processing algorithms. In such cases, calculations on highly repeated trials show increased separation of the object-level histogram from the background-level distribution. Increased conspicuity is objectively characterized by robust statistical measures of distribution separation.