Computer modeling of the spatial resolution properties of a dedicated breast CT system

Optimization of 3D imaging time reduction by assessing spatial resolution in the slice selective direction using the ladder method

PURPOSE

Assessing spatial resolution in MRI is challenging due to non-linearity. Despite the widespread use of 3D imaging in clinical practice for lesion detection and multi-planar reconstruction (MPR), the extended acquisition time poses a shortcoming. To address this, the "Slice resolution" parameter is utilized; however, its impact on MPR images is unclear. This study aims to assess spatial resolution using the ladder method, investigate the effects of diverse slice resolution settings in various imaging sequences, and propose optimal conditions.

METHODS

Images were acquired using various 3D imaging sequences-SPACE T1WI, SPACE T2WI, and VIBE T1WI-with different slice resolutions. Axial cross-section images were acquired and reconstructed into coronal cross-sections. The ladder method was employed for objective evaluation, including spatial frequency analysis. Additionally, visual evaluation was conducted and compared with ladder method results.

RESULTS

For three imaging sequences, the evaluated value of ladder method remained relatively constant from 100 % to 80 % slice resolution. However, the evaluated value decreased in low-spatial frequency for slice resolution below 70 %.

CONCLUSIONS

Results from both ladder method and visual evaluations indicated image quality remained stable when the slice resolution was decreased to 80 %, potentially enabling a 20 % reduction in imaging time while preserving resolution in other cross-sections reconstructed by MPR.

Image quality evaluation in deep-learning-based CT noise reduction using virtual imaging trial methods: Contrast-dependent spatial resolution

BACKGROUND

Deep-learning-based image reconstruction and noise reduction methods (DLIR) have been increasingly deployed in clinical CT. Accurate image quality assessment of these methods is challenging as the performance measured using physical phantoms may not represent the true performance of DLIR in patients since DLIR is trained mostly on patient images.

PURPOSE

In this work, we aim to develop a patient-data-based virtual imaging trial framework and, as a first application, use it to measure the spatial resolution properties of a DLIR method.

METHODS

The patient-data-based virtual imaging trial framework consists of five steps: (1) insertion of lesions into projection domain data using the acquisition geometry of the patient exam to simulate different lesion characteristics; (2) insertion of noise into projection domain data using a realistic photon statistical model of the CT system to simulate different dose levels; (3) creation of DLIR-processed images from projection or image data; (4) creation of ensembles of DLIR-processed patient images from a large number of noise and lesion realizations; and (5) evaluation of image quality using ensemble DLIR images. This framework was applied to measure the spatial resolution of a ResNet based deep convolutional neural network (DCNN) trained on patient images. Lesions in a cylindrical shape and different contrast levels (-500, -100, -50, -20, -10 HU) were inserted to the lower right lobe of the liver in a patient case. Multiple dose levels were simulated (50%, 25%, 12.5%). Each lesion and dose condition had 600 noise realizations. Multiple reconstruction and denoising methods were used on all the noise realizations, including the original filtered-backprojection (FBP), iterative reconstruction (IR), and the DCNN method with three different strength setting (DCNN-weak, DCNN-medium, and DCNN-strong). Mean lesion signal was calculated by performing ensemble averaging of all the noise realizations for each lesion and dose condition and then subtracting the lesion-present images from the lesion absent images. Modulation transfer functions (MTFs) both in-plane and along the z-axis were calculated based on the mean lesion signals. The standard deviations of MTFs at each condition were estimated with bootstrapping: randomly sampling (with replacement) all the DLIR/FBP/IR images from the ensemble data (600 samples) at each condition. The impact of varying lesion contrast, dose levels, and denoising strengths were evaluated. Statistical analysis with paired t-test was used to compare the z-axis and in-plane spatial resolution of five algorithms for five different contrasts and three dose levels.

RESULTS

The in-plane and z-axis spatial resolution degradation of DCNN becomes more severe as the contrast or radiation dose decreased, or DCNN denoising strength increased. In comparison with FBP, a 59.5% and 4.1% reduction of in-plane and z-axis MTF (in terms of spatial frequencies at 50% MTF), respectively, was observed at low contrast (-10 HU) for DCNN with the highest denoising strength at 25% routine dose level. When the dose level reduces from 50% to 12.5% of routine dose, the in-plane and z-axis MTFs reduces from 92.1% to 76.3%, and from 98.9% to 95.5%, respectively, at contrast of -100 HU, using FBP as the reference. For most conditions of contrasts and dose levels, significant differences were found among the five algorithms, with the following relationship in both in-plane and cross-plane spatial resolution: FBP > DCNN-Weak > IR > DCNN-Medium > DCNN-Strong. The spatial resolution difference among algorithms decreases at higher contrast or dose levels.

CONCLUSIONS

A patient-data-based virtual imaging trial framework was developed and applied to measuring the spatial resolution properties of a DCNN noise reduction method at different contrast and dose levels using real patient data. As with other non-linear image reconstruction and post-processing techniques, the evaluated DCNN method degraded the in-plane and z-axis spatial resolution at lower contrast levels, lower radiation dose, and higher denoising strength.

Flat-panel conebeam CT in the clinic: history and current state

Research into conebeam CT concepts began as soon as the first clinical single-slice CT scanner was conceived. Early implementations of conebeam CT in the 1980s focused on high-contrast applications where concurrent high resolution ( < 200    μ m ), for visualization of small contrast-filled vessels, bones, or teeth, was an imaging requirement that could not be met by the contemporaneous CT scanners. However, the use of nonlinear imagers, e.g., x-ray image intensifiers, limited the clinical utility of the earliest diagnostic conebeam CT systems. The development of consumer-electronics large-area displays provided a technical foundation that was leveraged in the 1990s to first produce large-area digital x-ray detectors for use in radiography and then compact flat panels suitable for high-resolution and high-frame-rate conebeam CT. In this review, we show the concurrent evolution of digital flat panel (DFP) technology and clinical conebeam CT. We give a brief summary of conebeam CT reconstruction, followed by a brief review of the correction approaches for DFP-specific artifacts. The historical development and current status of flat-panel conebeam CT in four clinical areas-breast, fixed C-arm, image-guided radiation therapy, and extremity/head-is presented. Advances in DFP technology over the past two decades have led to improved visualization of high-contrast, high-resolution clinical tasks, and image quality now approaches the soft-tissue contrast resolution that is the standard in clinical CT. Future technical developments in DFPs will enable an even broader range of clinical applications; research in the arena of flat-panel CT shows no signs of slowing down.

Location and direction dependence in the 3D MTF for a high-resolution CT system

PURPOSE

The purpose of this study was to quantify location and direction-dependent variations in the 3D modulation transfer function (MTF) of a high-resolution CT scanner with selectable focal spot sizes and resolution modes.

METHODS

The Aquilion Precision CT scanner (Canon Medical Systems) has selectable 0.25 mm or 0.5 mm detectors (by binning) in both the axial (x-y) and detector array width (z) directions. For the x-y and z orientations, detectors are configured (x-y) = 0.5 mm/(z) = 0.5 mm for normal resolution (NR), 0.25/0.5 mm for high resolution (HR), and 0.25/0.25 mm for super high resolution (SHR). Six focal spots (FS1-FS6) range in size from 0.4 (x-y) × 0.5 mm (z) for FS1 to 1.6 × 1.4 mm for FS6. Phantoms fabricated from spherical objects were positioned at radial distances of 0, 4.0, 7.5, 11.0, 14.5, and 18.5 cm. Axial and helical acquisitions were utilized and reconstructed using filtered back projection with the FC18 "Body," FC30 "Bone," and FC81 "Bone Sharp" kernels. The reconstructions were used to measure a 1D slice of the 3D MTF by oversampling the 3D ESF in the axial plane [MTF(f_r ); φ = 0°)], 45° out of the axial plane [MTF(f_r ); φ = 45°)], in the longitudinal direction [MTF(f_r ); φ = 80°)], and along the radial and azimuthal directions within the axial plane.

RESULTS

The MTF(f_r ); φ = 45°) drops to 10% (f₁₀ ) at 1.20, 1.45, and 2.06 mm^-1 for NR, HR, and SHR, respectively, for a helical acquisition with FS1, FC30, and r = 4 cm from the isocenter. The MTF(f_r ); φ = 45°) includes contributions of both the axial-plane MTF (f₁₀  = 1.10, 2.04, and 2.01 mm^-1 ) and the longitudinal MTF (f₁₀  = 1.17, 1.18, and 1.82 mm^-1 ) for the NR, HR, and SHR modes, respectively. For SHR, the axial scan mode showed a 15-25% improvement over helical mode in the longitudinal resolution. Helical pitch, ranging from 0.569 to 1.381, did not appreciably affect the 3D resolution (<2%). The radial MTFs across the axial field of view (FOV) showed dependencies on the focal spot length in z; for example, for SHR with FS2 (0.6 × 0.6 mm), f₁₀ at  r = 11 cm was within 17% of the value at r = 4 cm, but for SHR with FS3 (0.6 × 1.3), the reduction in f₁₀ was 46% from 4 to 11 cm from the isocenter. The azimuthal MTF also decreased as r increased but less so for longer gantry rotation times and smaller focal spot dimensions in the axial plane. The longitudinal MTF was minimally affected (<11%) by position in the FOV and was principally affected by the focal spot length in the z-dimension.

CONCLUSIONS

The 3D MTF was measured throughout the FOV of a high-resolution CT scanner, quantifying the advantages of different resolution modes and focal spot sizes on the axial-plane and longitudinal MTF. Reconstruction kernels were shown to impact axial-plane resolution, imparting non-isotropic 3D resolution characteristics. Focal spot size (both in x-y and in z) and gantry rotation time play important roles in preserving the high-resolution characteristics throughout the field of view for this new high-resolution CT scanner technology.

Reduction of scatter in breast CT yields improved microcalcification visibility

The inadequate visibility of microcalcifications-small calcium deposits that cue radiologists to early stages of cancer-is a major limitation in current designs of dedicated breast computed tomography (bCT). This limitation has previously been attributed to the constituent components, spatial resolution, and utilized dose. Scattered radiation has been considered an occurrence with low-frequency impacts that can be compensated for in post-processing. We hypothesized, however, that the acquisition of scattered radiation has a far more detrimental impact on clinically relevant features than has previously been understood. Critically, acquisition of scatter leads to the reduced visibility of microcalcifications. This hypothesis was investigated and supported via mathematical derivations and simulation studies. We conducted a series of comparative studies in which four bCT systems were simulated under iso-dose and iso-resolution conditions, characterizing the dependencies of microcalcification contrast on accumulated scatter. Included among the simulated systems is a novel bCT design-narrow beam bCT (NB-bCT)-that captures nearly zero scatter. We find that current bCT systems suffer from significant levels of scatter. As validated in theory, depending on the system and size of microcalcifications, between 25% and over 70% of contrast resolution is lost due to scatter. The results in NB-bCT, however, provide evidence that by removing scatter build-up in projections, the contrast of microcalcifications in a bCT image is preserved, regardless of their size or location in the breast.

Reconstruction-of-difference (RoD) imaging for cone-beam CT neuro-angiography

Timely evaluation of neurovasculature via CT angiography (CTA) is critical to the detection of pathology such as ischemic stroke. Cone-beam CTA (CBCT-A) systems provide potential advantages in the timely use at the point-of-care, although challenges of a relatively slow gantry rotation speed introduce tradeoffs among image quality, data consistency and data sparsity. This work describes and evaluates a new reconstruction-of-difference (RoD) approach that is robust to such challenges. A fast digital simulation framework was developed to test the performance of the RoD over standard reference reconstruction methods such as filtered back-projection (FBP) and penalized likelihood (PL) over a broad range of imaging conditions, grouped into three scenarios to test the trade-off between data consistency, data sparsity and peak contrast. Two experiments were also conducted using a CBCT prototype and an anthropomorphic neurovascular phantom to test the simulation findings in real data. Performance was evaluated primarily in terms of normalized root mean square error (NRMSE) in comparison to truth, with reconstruction parameters chosen to optimize performance in each case to ensure fair comparison. The RoD approach reduced NRMSE in reconstructed images by up to 50%-53% compared to FBP and up to 29%-31% compared to PL for each scenario. Scan protocols well suited to the RoD approach were identified that balance tradeoffs among data consistency, sparsity and peak contrast-for example, a CBCT-A scan with 128 projections acquired in 8.5 s over a 180°  +  fan angle half-scan for a time attenuation curve with ~8.5 s time-to-peak and 600 HU peak contrast. With imaging conditions such as the simulation scenarios of fixed data sparsity (i.e. varying levels of data consistency and peak contrast), the experiments confirmed the reduction of NRMSE by 34% and 17% compared to FBP and PL, respectively. The RoD approach demonstrated superior performance in 3D angiography compared to FBP and PL in all simulation and physical experiments, suggesting the possibility of CBCT-A on low-cost, mobile imaging platforms suitable to the point-of-care. The algorithm demonstrated accurate reconstruction with a high degree of robustness against data sparsity and inconsistency.

Generation of polychromatic projection for dedicated breast computed tomography simulation using anthropomorphic numerical phantom

Numerical simulations are fundamental to the development of medical imaging systems because they can save time and effort in research and development. In this study, we developed a method of creating the virtual projection images that are necessary to study dedicated breast computed tomography (BCT) systems. Anthropomorphic software breast phantoms of the conventional compression type were synthesized and redesigned to meet the requirements of dedicated BCT systems. The internal structure of the breast was randomly constructed to develop the proposed phantom, enabling the internal structure of a naturally distributed real breast to be simulated. When using the existing monochromatic photon incidence assumption for projection-image generation, it is not possible to simulate various artifacts caused by the X-ray spectrum, such as the beam hardening effect. Consequently, the system performance could be overestimated. Therefore, we considered the polychromatic spectrum in the projection image generation process and verified the results. The proposed method is expected to be useful for the development and optimization of BCT systems.

Modelling the transport of optical photons in scintillation detectors for diagnostic and radiotherapy imaging

Computational modelling of radiation transport can enhance the understanding of the relative importance of individual processes involved in imaging systems. Modelling is a powerful tool for improving detector designs in ways that are impractical or impossible to achieve through experimental measurements. Modelling of light transport in scintillation detectors used in radiology and radiotherapy imaging that rely on the detection of visible light plays an increasingly important role in detector design. Historically, researchers have invested heavily in modelling the transport of ionizing radiation while light transport is often ignored or coarsely modelled. Due to the complexity of existing light transport simulation tools and the breadth of custom codes developed by users, light transport studies are seldom fully exploited and have not reached their full potential. This topical review aims at providing an overview of the methods employed in freely available and other described optical Monte Carlo packages and analytical models and discussing their respective advantages and limitations. In particular, applications of optical transport modelling in nuclear medicine, diagnostic and radiotherapy imaging are described. A discussion on the evolution of these modelling tools into future developments and applications is presented.

Investigation of spatial resolution characteristics of an in vivo micro computed tomography system

The spatial resolution characteristics of an in vivo micro computed tomography (CT) system was investigated in the in-plane (x-y), cross plane (z) and projection imaging modes. The micro CT system utilized in this study employs a flat panel detector with a 127 μm pixel pitch, a micro focus x-ray tube with a focal spot size ranging from 5-30 μm, and accommodates three geometric magnifications (M) of 1.72, 2.54 and 5.10. The in-plane modulation transfer function (MTF) curves were measured as a function of the number of projections, geometric magnification (M), detector binning and reconstruction magnification (M_Recon). The in plane cutoff frequency (10% MTF) ranged from 2.31 lp/mm (M=1.72, 2×2 binning) to 12.56 lp/mm (M=5.10, 1×1 binning) and a bar pattern phantom validated those measurements. A slight degradation in the spatial resolution was observed when comparing the image reconstruction with 511 and 918 projections, whose effect was visible at the lower frequencies. Small value of M_Recon has little or no impact on the in-plane spatial resolution owning to a stable system. Large value of M_Recon has implications on the spatial resolution and it was evident when comparing the bar pattern images reconstructed with M_Recon=1.25 and 2.5. The cross plane MTF curves showed that the spatial resolution increased as the slice thickness decreased. The cutoff frequencies in the projection imaging mode yielded slightly higher values as compared to the in-plane and cross plane modes at all the geometric magnifications (M). At M=5.10, the cutoff resolution of the projection and cross plane on an ultra-high contrast resolution bar chip phantom were 14.9 lp/mm and 13-13.5 lp/mm. Due to the finite focal spot size of the x-ray tube, the detector blur and the reconstruction kernel functions, the system's spatial resolution does not reach the limiting spatial resolution as defined by the Nyquist's detector criteria with an ideal point source. The geometric magnification employed in the micro CTs provide a tradeoff between field of view and spatial resolution for a wide range of applications.

Task-based optimization of dedicated breast CT via Hotelling observer metrics

PURPOSE

The purpose of this work is to develop and demonstrate a set of practical metrics for CT systems optimization. These metrics, based on the Hotelling observer (HO) figure of merit, are task-based. The authors therefore take the specific example of optimizing a dedicated breast CT system, including the reconstruction algorithm, for two relevant tasks, signal detection and Rayleigh discrimination.

METHODS

A dedicated breast CT system is simulated using specifications in the literature from an existing prototype. The authors optimize configuration and image reconstruction algorithm parameters for two tasks: the detection of simulated microcalcifications and the discrimination of two adjacent, high-contrast signals, known as the Rayleigh discrimination task. The effects on task performance of breast diameter, signal location, image grid size, projection view number, and reconstruction filter were all investigated. Two HO metrics were evaluated: the percentage of correct decisions in a two-alternative forced choice experiment (equivalent to area under the ROC curve or AUC), and the HO efficiency, defined as the squared ratio of HO signal-to-noise ratio (SNR) in the reconstructed image to HO SNR in the projection data.

RESULTS

The ease and efficiency of the HO metric computation allows a rapid high-resolution survey of many system parameters. Optimization of a range of system parameters using the HO results in images that subjectively appear optimal for the tasks investigated. Further, the results of assessment through the HO reproduce closely many existing results in the literature regarding the impact of parameter selection on image quality.

CONCLUSIONS

This study demonstrates the utility of a task-based approach to system design, evaluation, and optimization. The methodology presented is equally applicable to determining the impact of a wide range of factors, including patient parameters, system and acquisition design, and the reconstruction algorithm. The results demonstrate the versatility of the proposed HO formalism by not only generating a set of parameters that are optimal for a given task but also by qualitatively reproducing many existing results from the breast CT literature. Meanwhile, the implementation of the proposed methodology is straightforward and entirely simulation-based. This is an attractive feature for many system optimization problems, where the goal is to analyze the individual system components such as the image reconstruction algorithm. Final assessment of the system as a whole should be based also on real data studies.

Simulated lesion, human observer performance comparison between thin-section dedicated breast CT images versus computed thick-section simulated projection images of the breast

The objective of this study was to compare the lesion detection performance of human observers between thin-section computed tomography images of the breast, with thick-section (>40 mm) simulated projection images of the breast. Three radiologists and six physicists each executed a two alterative force choice (2AFC) study involving simulated spherical lesions placed mathematically into breast images produced on a prototype dedicated breast CT scanner. The breast image data sets from 88 patients were used to create 352 pairs of image data. Spherical lesions with diameters of 1, 2, 3, 5, and 11 mm were simulated and adaptively positioned into 3D breast CT image data sets; the native thin section (0.33 mm) images were averaged to produce images with different slice thicknesses; average section thicknesses of 0.33, 0.71, 1.5 and 2.9 mm were representative of breast CT; the average 43 mm slice thickness served to simulate simulated projection images of the breast.The percent correct of the human observer's responses were evaluated in the 2AFC experiments. Radiologists lesion detection performance was significantly (p < 0.05) better in the case of thin-section images, compared to thick section images similar to mammography, for all but the 1 mm lesion diameter lesions. For example, the average of three radiologist's performance for 3 mm diameter lesions was 92% correct for thin section breast CT images while it was 67% for the simulated projection images. A gradual reduction in observer performance was observed as the section thickness increased beyond about 1 mm. While a performance difference based on breast density was seen in both breast CT and the projection image results, the average radiologist performance using breast CT images in dense breasts outperformed the performance using simulated projection images in fatty breasts for all lesion diameters except 11 mm. The average radiologist performance outperformed that of the average physicist observer, however trends in performance were similar. Human observers demonstrate significantly better mass-lesion detection performance on thin-section CT images of the breast, compared to thick-section simulated projection images of the breast.

Optimization of the design of thick, segmented scintillators for megavoltage cone-beam CT using a novel, hybrid modeling technique

PURPOSE

Active matrix flat-panel imagers (AMFPIs) incorporating thick, segmented scintillators have demonstrated order-of-magnitude improvements in detective quantum efficiency (DQE) at radiotherapy energies compared to systems based on conventional phosphor screens. Such improved DQE values facilitate megavoltage cone-beam CT (MV CBCT) imaging at clinically practical doses. However, the MV CBCT performance of such AMFPIs is highly dependent on the design parameters of the scintillators. In this paper, optimization of the design of segmented scintillators was explored using a hybrid modeling technique which encompasses both radiation and optical effects.

METHODS

Imaging performance in terms of the contrast-to-noise ratio (CNR) and spatial resolution of various hypothetical scintillator designs was examined through a hybrid technique involving Monte Carlo simulation of radiation transport in combination with simulation of optical gain distributions and optical point spread functions. The optical simulations employed optical parameters extracted from a best fit to measurement results reported in a previous investigation of a 1.13 cm thick, 1016 μm pitch prototype BGO segmented scintillator. All hypothetical designs employed BGO material with a thickness and element-to-element pitch ranging from 0.5 to 6 cm and from 0.508 to 1.524 mm, respectively. In the CNR study, for each design, full tomographic scans of a contrast phantom incorporating various soft-tissue inserts were simulated at a total dose of 4 cGy.

RESULTS

Theoretical values for contrast, noise, and CNR were found to be in close agreement with empirical results from the BGO prototype, strongly supporting the validity of the modeling technique. CNR and spatial resolution for the various scintillator designs demonstrate complex behavior as scintillator thickness and element pitch are varied--with a clear trade-off between these two imaging metrics up to a thickness of ~3 cm. Based on these results, an optimization map indicating the regions of design that provide a balance between these metrics was obtained. The map shows that, for a given set of optical parameters, scintillator thickness and pixel pitch can be judiciously chosen to maximize performance without resorting to thicker, more costly scintillators.

CONCLUSIONS

Modeling radiation and optical effects in thick, segmented scintillators through use of a hybrid technique can provide a practical way to gain insight as to how to optimize the performance of such devices in radiotherapy imaging. Assisted by such modeling, the development of practical designs should greatly facilitate low-dose, soft tissue visualization employing MV CBCT imaging in external beam radiotherapy.

Location dependency of the spatial resolution of cone beam computed tomography for dental use

OBJECTIVE

This study evaluates the spatial resolution of cone beam computed tomography for dental use (CBCT) via modulation transfer function (MTF) analyses.

STUDY DESIGN

Two models of CBCT system, 3DX FPD8 and FineCube v.12, were used. MTF analysis was applied to each CBCT system using a thin tungsten wire technique.

RESULTS

The MTF curves in the radial direction on the XY-plane were concordant regardless of position, whereas the curves in the azimuthal direction tended to decrease as the distance from the rotation center increased. In the Z-axis direction, the MTF curve of the medial level of the field of view was superior to that of any other level.

CONCLUSION

The spatial resolution of CBCT systems depends on the location within the field of view. Because the spatial resolution was the highest in the medial level and rotation center position, an object should be placed at this position during a CBCT examination.

A nonlinear lag correction algorithm for a-Si flat-panel x-ray detectors

PURPOSE

Detector lag, or residual signal, in a-Si flat-panel (FP) detectors can cause significant shading artifacts in cone-beam computed tomography reconstructions. To date, most correction models have assumed a linear, time-invariant (LTI) model and correct lag by deconvolution with an impulse response function (IRF). However, the lag correction is sensitive to both the exposure intensity and the technique used for determining the IRF. Even when the LTI correction that produces the minimum error is found, residual artifact remains. A new non-LTI method was developed to take into account the IRF measurement technique and exposure dependencies.

METHODS

First, a multiexponential (N = 4) LTI model was implemented for lag correction. Next, a non-LTI lag correction, known as the nonlinear consistent stored charge (NLCSC) method, was developed based on the LTI multiexponential method. It differs from other nonlinear lag correction algorithms in that it maintains a consistent estimate of the amount of charge stored in the FP and it does not require intimate knowledge of the semiconductor parameters specific to the FP. For the NLCSC method, all coefficients of the IRF are functions of exposure intensity. Another nonlinear lag correction method that only used an intensity weighting of the IRF was also compared. The correction algorithms were applied to step-response projection data and CT acquisitions of a large pelvic phantom and an acrylic head phantom. The authors collected rising and falling edge step-response data on a Varian 4030CB a-Si FP detector operating in dynamic gain mode at 15 fps at nine incident exposures (2.0%-92% of the detector saturation exposure). For projection data, 1st and 50th frame lag were measured before and after correction. For the CT reconstructions, five pairs of ROIs were defined and the maximum and mean signal differences within a pair were calculated for the different exposures and step-response edge techniques.

RESULTS

The LTI corrections left residual 1st and 50th frame lag up to 1.4% and 0.48%, while the NLCSC lag correction reduced 1st and 50th frame residual lags to less than 0.29% and 0.0052%. For CT reconstructions, the NLCSC lag correction gave an average error of 11 HU for the pelvic phantom and 3 HU for the head phantom, compared to 14-19 HU and 2-11 HU for the LTI corrections and 15 HU and 9 HU for the intensity weighted non-LTI algorithm. The maximum ROI error was always smallest for the NLCSC correction. The NLCSC correction was also superior to the intensity weighting algorithm.

CONCLUSIONS

The NLCSC lag algorithm corrected for the exposure dependence of lag, provided superior image improvement for the pelvic phantom reconstruction, and gave similar results to the best case LTI results for the head phantom. The blurred ring artifact that is left over in the LTI corrections was better removed by the NLCSC correction in all cases.

Effect of slice thickness on detectability in breast CT using a prewhitened matched filter and simulated mass lesions

PURPOSE

Dedicated breast CT (bCT) is an emerging technology with the potential to improve the detection of breast cancer in screening and diagnostic capacities. Typically, the 3D volume reconstructed from the scanner is displayed as sectional images. The purpose of this study was to evaluate the effect of section thickness on the detectability of simulated masses using a prewhitened matched filter (PWMF) as a model observer.

METHODS

A breast CT scanner has been designed and fabricated in the authors' laboratory with more than 200 women imaged in IRB-approved phase I and phase II trials to date. Of these, 151 bilateral data sets were selected on the basis of low artifact content, sufficient breast coverage, and excluding cases with breast implants. BIRADS breast density ratings were available for 144 of these patients. Spherical mass lesions of diameter 1, 2, 3, 5, 11, and 15 mm were mathematically generated and embedded at random locations within the parenchymal region of each bCT volume. Microcalcifications were not simulated in this study. For each viewing plane (sagittal, axial, and coronal) and section thickness (ranging from 0.3 to 44 mm), section images of the breast parenchyma containing the lesion were generated from the reconstructed bCT data sets by averaging voxels over the length of the section. Using signal known exactly (SKE) model observer methodology, receiver operating characteristic (ROC) curve analysis was performed on each generated projected image using a PWMF based model observer. ROC curves were generated for each breast data set, and the area under the ROC curve (AUC) was evaluated as well as the sensitivity at 95% specificity.

RESULTS

For all lesion sizes, performance rises modestly to a peak before falling off substantially as section thickness increases over the range of the study. We find that the optimal section thickness tracks the size of the lesion to be detected linearly with a small positive offset and slopes ranging from 0.27 to 0.44. No significant differences were observed between left and right breasts. Performance measures are negatively correlated with measures of breast density, with an average correlation coefficient of -0.48 for the BIRADS breast density score and -0.81 for the proportion of glandular tissue in the breast interior.

CONCLUSIONS

This study shows quantitatively how PWMF detection performance of a known lesion size is influenced by section thickness in dedicated breast CT. While the optimal section thickness is tuned to the size of the lesion being detected, overall performance is more robust for thin section images compared to thicker images.

Effect of detector lag on CT noise power spectra

PURPOSE

The authors examined the effect of detector lag on the noise power spectrum (NPS) of CT images reconstructed with filtered backprojection (FBP).

METHODS

The authors derived an analytical expression of the NPS with detector lag, and then verified it using computer simulations with parallel beam and fan beam geometries. The dependence of the NPS on the amount of lag, location within the scanned field of view (FOV), and the number of views used in the reconstruction (samples per rotation) was investigated using constant and view dependent noise in the raw data.

RESULTS

Detector lag introduces noise correlation in the azimuthal direction. The effect on the NPS is a frequency dependent reduction in amplitude. In small regions of the image, the effect is primarily in the frequencies corresponding to the azimuthal direction. The noise blurring and NPS filtering increases with increasing radial distance, and therefore regions at larger radial distances have lower noise power. With the same detector lag response function, the amount of noise correlation and NPS filtering decreases with increasing number of views.

CONCLUSIONS

The shape of the NPS depends on the detector lag coefficients, location of the region, and the number of views used in the reconstruction. In general, the noise correlation caused by detector lag decreased the amplitude of the NPS.

Investigation into the optimal linear time-invariant lag correction for radar artifact removal

PURPOSE

Detector lag, or residual signal, in amorphous silicon (a-Si) flat-panel (FP) detectors can cause significant shading artifacts in cone-beam computed tomography (CBCT) reconstructions. To date, most correction models have assumed a linear, time-invariant (LTI) model and lag is corrected by deconvolution with an impulse response function (IRF). However, there are many ways to determine the IRF. The purpose of this work is to better understand detector lag in the Varian 4030CB FP and to identify the IRF measurement technique that best removes the CBCT shading artifact.

METHODS

We investigated the linearity of lag in a Varian 4030CB a-Si FP operating in dynamic gain mode at 15 frames per second by examining the rising step-response function (RSRF) followed by the falling step-response function (FSRF) at ten incident exposures (0.5%-84% of a-Si FP saturation exposure). We implemented a multiexponential (N = 4) LTI model for lag correction and investigated the effects of various techniques for determining the IRF such as RSRF versus FSRF, exposure intensity, length of exposure, and spatial position. The resulting IRFs were applied to (1) the step-response projection data and (2) CBCT acquisitions of a large pelvic phantom and acrylic head phantom. For projection data, 1st and 50th frame lags were measured pre- and postcorrection. For the CBCT reconstructions, four pairs of ROIs were defined and the maximum and mean errors within each pair were calculated for the different exposures and step-response edge techniques.

RESULTS

A nonlinearity greater than 50% was observed in the FSRF data. A model calibrated with RSRF data resulted in overcorrection of FSRF data. Conversely, models calibrated with FSRF data applied to RSRF data resulted in undercorrection of the RSRF. Similar effects were seen when LTI models were applied to data collected at different incident exposures. Some spatial variation in lag was observed in the step-response data. For CBCT reconstructions, an average error range of 3-21 HU was observed when using IRFs from different techniques. For our phantoms and FP, the lowest average error occurred for the FSRF-based techniques at exposures of 1.6 or 3.4% a-Si FP saturation, depending on the phantom used.

CONCLUSIONS

The choice of step-response edge (RSRF versus FSRF) and exposure intensity for IRF calibration could leave large residual lag in the step-response data. For the CBCT reconstructions, IRFs derived from FSRF data at low exposure intensities (1.6 and 3.4%) best removed the CBCT shading artifact. Which IRF to use for lag correction could be selected based on the object size.

An X-Ray computed tomography/positron emission tomography system designed specifically for breast imaging

Mammography has served the population of women who are at-risk for breast cancer well over the past 30 years. While mammography has undergone a number of changes as digital detector technology has advanced, other modalities such as computed tomography have experienced technological sophistication over this same time frame as well. The advent of large field of view flat panel detector systems enable the development of breast CT and several other niche CT applications, which rely on cone beam geometry. The breast, it turns out, is well suited to cone beam CT imaging because the lack of bones reduces artifacts, and the natural tapering of the breast anteriorly reduces the x-ray path lengths through the breast at large cone angle, reducing cone beam artifacts as well. We are in the process of designing a third prototype system which will enable the use of breast CT for image guided interventional procedures. This system will have several copies fabricated so that several breast CT scanners can be used in a multi-institutional clinical trial to better understand the role that this technology can bring to breast imaging.

Methodology for generating a 3D computerized breast phantom from empirical data

The initial process for creating a flexible three-dimensional computer-generated breast phantom based on empirical data is described. Dedicated breast computed-tomography data were processed to suppress noise and scatter artifacts in the reconstructed image set. An automated algorithm was developed to classify the breast into its primary components. A preliminary phantom defined using subdivision surfaces was generated from the segmented data. To demonstrate potential applications of the phantom, simulated mammographic image data were acquired of the phantom using a simplistic compression model and an analytic projection algorithm directly on the surface model. The simulated image was generated using a model for a polyenergetic cone-beam projection of the compressed phantom. The methods used to create the breast phantom generate resulting images that have a high level of tissue structure detail available and appear similar to actual mammograms. Fractal dimension measurements of simulated images of the phantom are comparatively similar to measurements from images of real human subjects. A realistic and geometrically defined breast phantom that can accurately simulate imaging data may have many applications in breast imaging research.

Effect of oblique X-ray incidence in flat-panel computed tomography of the breast

We quantify the variation in resolution due to anisotropy caused by oblique X-ray incidence in indirect flat-panel detectors for computed tomography breast imaging systems. We consider a geometry and detector type utilized in breast computed tomography (CT) systems currently being developed. Our methods rely on mantis, a combined X-ray, electron, and optical Monte Carlo transport open source code. The physics models are the most accurate available in general-purpose Monte Carlo packages in the diagnostic energy range. We consider maximum-obliquity angles of 10 ( degrees ) and 13 ( degrees ) at the centers of the 30 and 40 cm detector edges, respectively, and 16 ( degrees ) at the corner of the detector. Our results indicate that blur is asymmetric and that the resolution properties vary significantly with the angle (or location) of incidence. Our results suggest that the asymmetry can be as high as a factor of 2.6 between orthogonal directions. Anisotropy maps predicted by mantis provide an understanding of the effect that such variations have on the imaging system and allow more accurate modeling and optimization of breast CT systems. These maps of anisotropy across the detector could lead to improved reconstruction and help motivate physics-based strategies for computer detection of breast lesions.

Experimental validation of a three-dimensional linear system model for breast tomosynthesis

A three-dimensional (3D) linear model for digital breast tomosynthesis (DBT) was developed to investigate the effects of different imaging system parameters on the reconstructed image quality. In the present work, experimental validation of the model was performed on a prototype DBT system equipped with an amorphous selenium (a-Se) digital mammography detector and filtered back-projection (FBP) reconstruction methods. The detector can be operated in either full resolution with 85 microm pixel size or 2 x 1 pixel binning mode to reduce acquisition time. Twenty-five projection images were acquired with a nominal angular range of +/- 20 degrees. The images were reconstructed using a slice thickness of 1 mm with 0.085 x 0.085 mm in-plane pixel dimension. The imaging performance was characterized by spatial frequency-dependent parameters including a 3D noise power spectrum (NPS) and in-plane modulation transfer function (MTF). Scatter-free uniform x-ray images were acquired at four different exposure levels for noise analysis. An aluminum (Al) edge phantom with 0.2 mm thickness was imaged to measure the in-plane presampling MTF. The measured in-plane MTF and 3D NPS were both in good agreement with the model. The dependence of DBT image quality on reconstruction filters was investigated. It was found that the slice thickness (ST) filter, a Hanning window to limit the high-frequency components in the slice thickness direction, reduces noise aliasing and improves 3D DQE. An ACR phantom was imaged to investigate the effects of angular range and detector operational modes on reconstructed image quality. It was found that increasing the angular range improves the MTF at low frequencies, resulting in better detection of large-area, low-contrast mass lesions in the phantom. There is a trade-off between noise and resolution for pixel binning and full resolution modes, and the choice of detector mode will depend on radiation dose and the targeted lesion.

Noise power properties of a cone-beam CT system for breast cancer detection

The noise power properties of a cone-beam computed tomography (CT) system dedicated for breast cancer detection were investigated. Uniform polyethylene cylinders of various diameters were scanned under different system acquisition conditions. Noise power spectra were calculated from difference data generated by subtraction between two identical scans. Multidimensional noise power spectra (NPS) were used as the metric to evaluate the noise properties of the breast CT (bCT) under different system acquisition and reconstruction conditions. A comprehensive investigation of the noise properties was performed in regard to system acquisition parameters including kVp, mA, number of cone-beam projection images used, cone angle, and object size. The influence on reconstruction parameters including interpolation method, reconstruction filter, field of view, matrix size, and slice thickness were also studied. Under certain conditions, the zero-dimensional NPS (image variance) was used as a quantitative index to compare the influence from different scan parameters, especially the radiation dose. If the total scan dose is changed by linearly changing the total number of projection images while the dose per frame is kept constant, the noise power has a linear relationship with the reciprocal of the total dose. If the total scan dose is changed by linearly changing the dose per frame while the total number of projection images is kept constant, the noise power has a quadratic relationship with the reciprocal of the total dose. With the same amount of total dose, using fewer projection images results in lower image noise power in the CT image. Quantitative results from this noise power analysis provide guidance for the bCT system operation, optimization, and data reconstruction.

Dedicated breast computed tomography: volume image denoising via a partial-diffusion equation based technique

Dedicated breast computed tomography (CT) imaging possesses the potential for improved lesion detection over conventional mammograms, especially for women with dense breasts. The breast CT images are acquired with a glandular dose comparable to that of standard two-view mammography for a single breast. Due to dose constraints, the reconstructed volume has a non-negligible quantum noise when thin section CT slices are visualized. It is thus desirable to reduce noise in the reconstructed breast volume without loss of spatial resolution. In this study, partial diffusion equation (PDE) based denoising techniques specifically for breast CT were applied at different steps along the reconstruction process and it was found that denoising performed better when applied to the projection data rather than reconstructed data. Simulation results from the contrast detail phantom show that the PDE technique outperforms Wiener denoising as well as adaptive trimmed mean filter. The PDE technique increases its performance advantage relative to Wiener techniques when the photon fluence is reduced. With the PDE technique, the sensitivity for lesion detection using the contrast detail phantom drops by less than 7% when the dose is cut down to 40% of the two-view mammography. For subjective evaluation, the PDE technique was applied to two human subject breast data sets acquired on a prototype breast CT system. The denoised images had appealing visual characteristics with much lower noise levels and improved tissue textures while maintaining sharpness of the original reconstructed volume.