Maximum Likelihood Expectation Maximization Research Articles

Acceleration of PET-based industrial internal defect image reconstruction using FPGA

Positron emission tomography (PET) can be used to measure the internal defects of industrial parts. However, PET requires a long execution time of image reconstruction, which hinders its practical usage in industrial measurements. A novel parallel scheme based on field-programmable gate arrays (FPGAs) is proposed in this study to accelerate PET image reconstruction. A fast maximum-likelihood expectation–maximization iteration reconstruction algorithm with prior estimation is implemented on the FPGA. This method can achieve satisfactory PET images with limited iteration times. The resources in the FPGA are divided into several groups, and each group supports the image reconstruction for a single sinogram. Thus, several sinograms can be processed in parallel. Two internal defect detection experiments are conducted to apply the proposed method to industrial measurements. Results show that the inner structure can be detected, whereas the inner defects can be visualized. A group of 104 slice images is reconstructed in parallel on FPGAs, and the final 3D PET image of the inner defects is acquired in 10 s.

DPIR-Net: Direct PET Image Reconstruction Based on the Wasserstein Generative Adversarial Network

Positron emission tomography (PET) is an advanced medical imaging technique widely used in various clinical applications, such as tumor detection and neurologic disorders. Reducing the radiotracer dose is desirable in PET imaging because it decreases the patient’s radiation exposure. However, reducing the dose can also increase noise, affecting the image quality. Therefore, an advanced image reconstruction algorithm based on low-dose PET data is needed to improve the quality of the reconstructed image. In this article, we propose the use of a direct PET image reconstruction network (DPIR-Net) using an improved Wasserstein generative adversarial network (WGAN) framework to enhance image quality. This article provides two main findings. First, our network uses sinogram data as input and outputs high-quality PET images direct, resulting in shorter reconstruction times compared with traditional model-based reconstruction networks. Second, we combine perceptual loss, mean square error, and the Wasserstein distance as the loss function, which effectively solves the problem of excessive smoothness and loss of detailed information in traditional network image reconstruction. We performed a comparative study using maximum-likelihood expectation maximization (MLEM) with a post-Gaussian filter, a total variation (TV)-norm regularization, a nonlocal means (NLMs) denoising method, a neural network denoising method, a traditional deep learning PET reconstruction network, and our proposed DPIR-Net method and evaluated the proposed method using both human and mouse data. The mouse data were obtained from a small animal PET prototype system developed by our laboratory. The quantitative and qualitative results show that our proposed method outperformed the conventional methods.

Tomographic-based 3D scintillation dosimetry using a three-view plenoptic imaging system.

To demonstrate the feasibility of a three-plenoptic camera projection, scintillation-based dosimetry system for measuring three-dimensional (3D) dose distributions of static photon radiation fields. Static x-ray photon beams were delivered to a cubic plastic scintillator volume embedded within acrylic blocks. For each beam, three orthogonal projections of the scintillating light emission were recorded using a multifocus plenoptic camera. Experimental 3D reconstructions of the light distribution were obtained using an iterative maximum likelihood-expectation maximization (ML-EM) algorithm. For this purpose, the elements of the system matrix representing the contribution of the scintillator volume voxels to the camera sensor pixels were calculated using optical design software. A reconstruction-specific correction was applied to light reconstructions to account for scintillating light imaged by the camera but not directly resulting from dose deposition. Cross beam profiles (CBPs) and percentage depth dose (PDD) curves were compared to treatment planning system data for square fields. Three-dimensional and 3D gamma analyses were performed for concave-shaped dose distributions and the Pearson correlation coefficient and reconstruction error were employed to assess the quality of the measured relative 3D dose distributions. A full and accurate model of the plenoptic camera-based scintillation dosimetry system was implemented using the light ray tracing capabilities of optical design software. With this model, light distributions were successfully reconstructed over a volume of 60×60×60mm at a resolution of 2mm. For relative 3D measurements of square radiation fields of , and compared with treatment planning system reference distributions, the maximum root-mean-square error of the CBPs evaluated at two different depths was of 3.2%, 1.2%, and 1.1%, respectively; as for the corresponding linearly fitted PDDs of the square fields, the slopes of the reconstructed dose distributions overestimated those of the reference distributions by at most 0.2%/cm. The 2D gamma passing rate with a criterion of 2%/2mm for the concave-shaped photon field was of 61.6%, 66.1%, and 76.4% using one, two, and three plenoptic projections; the respective success rates become 77.1%, 87.5%, and 94.9% using a criterion of 3%/3mm. The 3D correlation coefficient for the corresponding reconstructions was of 0.688, 0.905, and 0.976, respectively. Three-dimensional light distributions emitted from within a plastic scintillator volume were successfully recovered using optical design software to establish a complete tomographic model of a plenoptic camera-based prototype. The tomographic model can equivalently extend to dynamic dose delivery measurements, providing temporal resolution limited by the camera's exposure time. This feasibility study enables a simplified design-to-implementation process for volumetric scintillation dosimetry prototypes toward fully meeting the clinical needs of 3D dose measurements for static and dynamic delivery techniques.

Image reconstruction using multi-energy system matrices with a scintillator-based gamma camera for nuclear security applications

The system response of a gamma camera is dependent on the photon energy, thus the energy-dependent response function needs to be considered to improve the quality and fidelity of reconstructed images for identifying radionuclides in security applications. In this study, two reconstruction strategies using the maximum-likelihood expectation maximization (MLEM) algorithm with the multi-energy system matrices calculated by Monte Carlo simulations are proposed. The difference between the two is in data acquisition; one uses the sum of all events into a single projection image while the other sorts them into separate energy windows. Various radiation images of gamma-ray sources were simulated with a Monte Carlo code, and an actual image was acquired with a gamma camera. Both simulation and experiment results demonstrated the feasibility of the presented multi-energy reconstruction strategies in the detection of orphan sources.

Estimate of the 225Ac Radioactive Isotope Distribution by Means of DOI Compton Imaging in Targeted Alpha Radiotherapy: A Monte Carlo Simulation

For successful targeted alpha radiotherapy (TAT), verifying the accurate position and distribution of a targeted radiotherapeutic agent in a patient or phantom is important. This paper, describes our investigation of depth-of-interaction (DOI) Compton imaging for the two γ-rays emitted during TAT with the 225Ac radioactive isotope. We optimized the design parameters of the DOI Compton camera, for example, the inter-detector distance, based on the figure of merit (FOM). The performance of DOI Compton imaging for TAT was improved because Doppler broadening and the energy uncertainty are inversely proportional to the radiation energy and the position uncertainty of the depth information is decreased. After the contrast phantom and the resolution phantom had been designed, two reconstruction algorithms, the filtered back-projection (FBP) algorithm and the maximum-likelihood expectation maximization (MLEM) algorithm, were applied to each reconstructed phantom image, and the qualities of the reconstructed images for the two γ-rays (218 keV and 440 keV) were compared. In the quantitative evaluation of the reconstructed images, the MLEM reconstruction algorithm performed better than the FBP algorithm. Based on Monte Carlo simulation studies, the DOI Compton images of the 225Ac radioactive isotope emitting two γ-rays demonstrated the capability of imaging a targeted radiotherapeutic agent in TAT.

Effect of Reconstruction Algorithm on the Identification of 3D Printing Polymers Based on Hyperspectral CT Technology Combined with Artificial Neural Network.

Hyperspectral X-ray CT (HXCT) technology provides not only structural imaging but also the information of material components therein. The main purpose of this study is to investigate the effect of various reconstruction algorithms on reconstructed X-ray absorption spectra (XAS) of components shown in the CT image by means of HXCT. In this paper, taking 3D printing polymer as an example, seven kinds of commonly used polymers such as thermoplastic elastomer (TPE), carbon fiber reinforced polyamide (PA-CF), acrylonitrile butadiene styrene (ABS), polylactic acid (PLA), ultraviolet photosensitive resin (UV9400), polyethylene terephthalate glycol (PETG), and polyvinyl alcohol (PVA) were selected as samples for hyperspectral CT reconstruction experiments. Seven kinds of 3D printing polymer and two interfering samples were divided into a training set and test sets. First, structural images of specimens were reconstructed by Filtered Back-Projection (FBP), Algebra Reconstruction Technique (ART) and Maximum-Likelihood Expectation-Maximization (ML-EM). Secondly, reconstructed XAS were extracted from the pixels of region of interest (ROI) compartmentalized in the images. Thirdly, the results of principal component analysis (PCA) demonstrated that the first four principal components contain the main features of reconstructed XAS, so we adopted Artificial Neural Network (ANN) trained by the reconstructed XAS expressed by the first four principal components in the training set to identify that the XAS of corresponding polymers exist in both of test sets from the training set. The result of ANN displays that FBP has the best performance of classification, whose ten-fold cross-validation accuracy reached 99%. It suggests that hyperspectral CT reconstruction is a promising way of getting image features and material features at the same time, which can be used in medical imaging and nondestructive testing.

Micro-Networks for Robust MR-Guided Low Count PET Imaging.

Noise suppression is particularly important in low count positron emission tomography (PET) imaging. Post-smoothing (PS) and regularization methods which aim to reduce noise also tend to reduce resolution and introduce bias. Alternatively, anatomical information from another modality such as magnetic resonance (MR) imaging can be used to improve image quality. Convolutional neural networks (CNNs) are particularly well suited to such joint image processing, but usually require large amounts of training data and have mostly been applied outside the field of medical imaging or focus on classification and segmentation, leaving PET image quality improvement relatively understudied. This article proposes the use of a relatively low-complexity CNN (micro-net) as a post-reconstruction MR-guided image processing step to reduce noise and reconstruction artefacts while also improving resolution in low count PET scans. The CNN is designed to be fully 3-D, robust to very limited amounts of training data, and to accept multiple inputs (including competitive denoising methods). Application of the proposed CNN on simulated low (30 M) count data (trained to produce standard (300 M) count reconstructions) results in a 36% lower normalized root mean squared error (NRMSE, calculated over ten realizations against the ground truth) compared to maximum-likelihood expectation maximization (MLEM) used in clinical practice. In contrast, a decrease of only 25% in NRMSE is obtained when an optimized (using knowledge of the ground truth) PS is performed. A 26% NRMSE decrease is obtained with both RM and optimized PS. Similar improvement is also observed for low count real patient datasets. Overfitting to training data is demonstrated to occur as the network size is increased. In an extreme case, a U-net (which produces better predictions for training data) is shown to completely fail on test data due to overfitting to this case of very limited training data. Meanwhile, the resultant images from the proposed CNN (which has low training data requirements) have lower noise, reduced ringing, and partial volume effects, as well as sharper edges and improved resolution compared to conventional MLEM.

Limited Angle Tomography reconstruction for non-standard MBI system by means of parallel-hole and pinhole optics

The purpose of the present work is the study of reconstruction properties of a new Molecular Breast Imaging (MBI) device for the early diagnosis of breast cancer, in Limited Angle Tomography (LAT), by using two asymmetric detector heads with different collimators. The detectors face each other in anti-parallel viewing direction and, mild-compressing the breast phantom, they are able to reconstruct the inner tumour of the phantoms with only a limited number of projections using a dedicated maximum-likelihood expectation maximization algorithm. The phantoms, the MBI system, as well as the Monte Carlo simulator, are briefly described. The MBI system's model has been implemented in IDL, in order to evaluate the best LAT configuration of the system and its reconstruction ability by varying tumour's size, depth and uptake. The LAT setup in real and simulated configurations, as well as the maximum-likelihood method and the preliminary reconstruction results, are discussed.

Discrimination of hydroxyl radical yields due to thermal neutrons, fast neutrons, and gamma rays in accelerator-based neutron fields

We proposed a depth-intensity curve discrimination (DCD) method for assessing absorbed dose contributions of hydroxyl radicals (HO·) due to thermal neutrons, fast neutrons, and gamma rays in accelerator-based mixed radiation fields, that are relevant to boron neutron capture therapy (BNCT). The HO· yields of thermal neutrons, fast neutrons, and gamma-rays measured by coumarin-3-carboxylic acid solutions were discriminately evaluated using depth-intensity curves of their radiations and maximum likelihood-expectation maximization algorithm. The discrimination accuracies of the three types of radiation were 2.8%, 4.2%, and 1.0%, respectively. We demonstrated that the DCD method is appropriate to determine absorbed dose contributions of HO· in complex accelerator-based neutron fields for BNCT.

Capability of MLEM and OE to Detect Range Shifts With a Compton Camera in Particle Therapy

To identify range deviations by using Compton cameras (CCs), tomographic image reconstruction of CC data is needed. Within this context, image reconstruction is usually performed using maximum likelihood expectation maximization (MLEM), and more recently, the origin ensemble (OE) algorithm. In this article, we investigate how MLEM and OE affect the precision and accuracy of estimated range deviations. In particular, we focus on the effects of data selection, statistical fluctuations, and artifact reduction. The use of external information of the beam path through a hodoscope was also explored. Additionally, two methods to calculate range deviations were tested. To this aim, realistic proton beams were simulated using GATE and data from single spots as well as from seven contiguous spots of an energy layer were reconstructed. MLEM and OE reacted differently to the poor data statistics. In general, both algorithms were able to detect range shifts for single spots, particularly when multiple coincidences were also considered. Selection of events corresponding to the most relevant energy peaks decreased the identification performance due to the lower statistics. When data from several contiguous spots were jointly reconstructed, the accuracy of the results degraded significantly, and nonzero shifts were assigned when no shifts had occurred. The limited size of the cameras and the subsequent restriction in the orientation and aperture of detected cones, as well as in the number of detected events are major challenges. Future efforts should be devoted to noise regularization and compensation for data truncation.

Efficient CT Image Reconstruction in a GPU Parallel Environment.

Computed tomography is nowadays an indispensable tool in medicine used to diagnose multiple diseases. In clinical and emergency room environments, the speed of acquisition and information processing are crucial. CUDA is a software architecture used to work with NVIDIA graphics processing units. In this paper a methodology to accelerate tomographic image reconstruction based on maximum likelihood expectation maximization iterative algorithm and combined with the use of graphics processing units programmed in CUDA framework is presented. Implementations developed here are used to reconstruct images with clinical use. Timewise, parallel versions showed improvement with respect to serial implementations. These differences reached, in some cases, 2 orders of magnitude in time while preserving image quality. The image quality and reconstruction times were not affected significantly by the addition of Poisson noise to projections. Furthermore, our implementations showed good performance when compared with reconstruction methods provided by commercial software. One of the goals of this work was to provide a fast, portable, simple, and cheap image reconstruction system, and our results support the statement that the goal was achieved.

English

English

Near-field artifacts reduction in coded aperture push-broom Compton scatter imaging

Recently, we have devised a novel method called coded aperture push-broom Compton scatter imaging (CAPCSI) to improve the use of backscattered X-rays. Based on a correlation algorithm, near-field artifacts are obvious in the CAPCSI, although they could be reduced partially by averaging dual acquisitions of a mask and an antimask. In this paper, we introduce a maximum-likelihood expectation–maximization (MLEM) algorithm into the CAPCSI to improve the artifacts without dual acquisitions. First, a theoretical system matrix (SM) and a simulative SM were derived from the generation of near-field artifacts. Then, simulation for ideal sources was carried out to compare different algorithms of the correlation and the MLEM. Further, a CAPCSI prototype was set up to confirm the simulation. Because of the system nonuniformity, an experimental SM was derived in advance by a ray-driven way. Finally, imaging for character phantoms was presented. Preliminary results showed the MLEM performed much better for the point source and could obtain a cleaner background for complex objects than the correlation. In the simulation, the theoretical SM and the simulative SM were no essential difference. However, in the experiment, the experimental SM performed essentially better than the theoretical SM, because it took the near-field effects and the system nonuniformity into a comprehensive consideration. This work will help enhance the application potential of the CAPCSI.

A novel MLEM stopping criterion for unfolding neutron fluence spectra in radiation therapy

The spectrum of secondary neutrons generated by a medical linear accelerator (linac) during high-energy radiation therapy must be accurately determined in order to assess the carcinogenic risk that these neutrons pose to patients. Neutron spectrometers such as the Nested Neutron Spectrometer (NNS) can be used to measure neutron fluence spectra but the raw measured data must be deconvolved (unfolded) with the detector’s response functions. The iterative Maximum-Likelihood Expectation–Maximization (MLEM) algorithm can be used to unfold the raw data, however it lacks an objective stopping criterion and produces an increasingly noisy solution as it iterates. In this work, we describe an objective stopping criterion that terminates MLEM unfolding of secondary neutron spectra in radiation therapy after solution convergence but prior to significant accumulation of noise. We validated the robustness of our stopping criterion by using it to unfold NNS measurements spanning a wide range of neutron fluence rates that were acquired around two linacs. We found that these unfolded spectra demonstrate a high level of agreement with the corresponding ideal unfolded spectra (obtained using Monte Carlo simulated spectra) and are relatively free of noise. Thus, use of our stopping criterion increases confidence in experimentally unfolded neutron spectra and can aid in improving carcinogenic risk estimates for patients receiving radiation therapy.

Quantitative X-ray fluorescence imaging of gold nanoparticles using joint L1 and total variation regularized reconstruction.

This work proposed a joint L1 and total variation (TV) regularized reconstruction method for X-ray fluorescence tomography (XFT), and investigated the performance of this method in quantitative imaging of gold nanoparticles (GNPs). We developed a dual-modality XFT/CT imaging system which consisted of a benchtop X-ray source, a translation/rotation stage, a silicon drift detector for X-ray fluorescence (XRF) detection, and a flat panel detector for transmission X-ray detection. A pencil-beam collimator was 3D printed with steel and employed in sample excitation. The sensitivity of the XFT imaging system was determined by imaging water phantoms with multiple inserts containing GNP solutions of various concentrations (0.02-0.16 wt.%). A joint L1 and total variation (TV) regularized algorithm was developed for XFT reconstruction, where the L1 regularization was used to reduce image artifacts and the TV regularization was used to preserve the shape of targets. Nonlinear conjugate gradient (NCG) descent algorithm with backtracking line search was adopted to solve the reconstruction problem. We compared the L1 + TV regularization method with filtered back projection (FBP), maximum likelihood expectation maximization (ML-EM), L1 regularization, and TV regularization methods. Contrast-to-noise ratio (CNR), Dice similarity coefficient (DSC) and localization error (LE) metrics were used to compare the performance of different methods. The CT and XFT imaging doses were also measured using EBT2 radiochromic films. The 3D printed pencil-beam collimator shaped an excitation beam with a 2 mm full width at half maximum at the imaging isocenter. Based on the phantom imaging experiments, the joint L1 and TV regularization method performed better than FBP, ML-EM, L1 regularization and TV regularization methods, with higher localization accuracy (offset <0.6 mm), CNR and DSC values. Compared with CT, XFT with L1 + TV regularized reconstruction demonstrated higher sensitivity in GNP imaging, and could detect GNP at a concentration of 0.02 wt.% or lower. Moreover, there existed a significant linear correlation (R2>0.99) between the reconstructed and true GNP concentration. The estimated XFT imaging dose is about 41.22 cGy under current setting. The joint L1 + TV regularized reconstruction algorithm performed better in noise suppression and shape preservation. Using the L1 + TV regularized reconstruction, the XFT system is able to localize GNP targets with submillimeter accuracy and quantify GNP distribution at a concentration of 0.02 wt.% or lower.

Computationally Efficient System Matrix Calculation Techniques in Computed Tomography Iterative Reconstruction.

Background:Relative to classical methods in computed tomography, iterative reconstruction techniques enable significantly improved image qualities and/or lowered patient doses. However, the computational speed is a major concern for these iterative techniques. In the present study, we present a method for fast system matrix calculation based on the line integral model (LIM) to speed up the computations without compromising the image quality. In addition, we develop a hybrid line–area integral model (AIM) that highlights the advantages of both LIM and AIMs.Methods:The contributing detectors for a given pixel and a given projection view, and the length of corresponding intersection lines with pixels, are calculated using our proposed algorithm. For the hybrid method, the respective narrow-angle fan beam was modeled by multiple equally spaced lines. The computed system matrix was evaluated in the context of reconstruction using the simultaneous algebraic reconstruction technique (SART) as well as maximum likelihood expectation maximization (MLEM).Results:The proposed LIM offers a considerable reduction in calculation times compared to the standard Siddon algorithm: 2.9 times faster. Differences in root mean square error and peak signal-to-noise ratio were not significant between the proposed LIM and the Siddon algorithm for both SART and MLEM reconstruction methods (P > 0.05). Meanwhile, the proposed hybrid method resulted in significantly improved image qualities relative to LIM and the Siddon algorithm (P < 0.05), though computations were 4.9 times more intensive than the proposed LIM.Conclusion:We have proposed two fast algorithms to calculate the system matrix. The first is based on LIM and was faster than the Siddon algorithm, with matched image quality, whereas the second method is a hybrid LIM–AIM that achieves significantly improved images though with its computational requirements.

Estimation of optimized timely system matrix with improved image quality in iterative reconstruction algorithm: A simulation study

The system matrix (SM) being a main part of statistical image reconstruction algorithms establishes relationship between the object and projection space. The aim was to determine it in a short duration time, towards obtaining the best quality of contrast images. In this study, a new analytical method based on Cavalieri's principle as subdividing common regions has been proposed in which the precision of the amounts of estimated areas was improved by increasing the number of divisions (NOD), and consequently the total SM's time was increased. An important issue is the tradeoff between the NODs and computational time. For this purpose, a Monte Carlo simulated Jaszczak phantom study was performed by the Monte Carlo N-Particle transport code version 5 (MCNP5) in which the tomographic images of resolution and contrast phantoms were reconstructed by maximum likelihood expectation maximization (MLEM) algorithm, and the influence of NODs variations was investigated. The results show that the lowest and best quality have been obtained at the NODs of 0 and 8, respectively and in the optimum case, the SM's total time at NOD of 8 was 925 s, which was much lower than those of the conventional Monte Carlo simulations and experimental test.

Multi-Schlieren CT Measurements of Supersonic Microjets from Circular and Square Micro Nozzles

Instantaneous three-dimensional (3D) density distributions of a shock-cell structure of perfectly and imperfectly expanded supersonic microjets escaping into an ambient space are measured. For the 3D observation of supersonic microjets, non-scanning 3D computerized tomography (CT) technique using a 20-directional quantitative schlieren optical system with flashlight source is employed for simultaneous schlieren photography. The 3D density distributions data of the microjets are obtained by 3D-CT reconstruction of the projection’s images using maximum likelihood-expectation maximization. Axisymmetric convergent-divergent (Laval) circular and square micro nozzles with operating nozzle pressure ratio 5.0, 4.5, 4.0, 3.67, and 3.5 have been studied. This study examines perfectly expanded, overexpanded, and underexpanded supersonic microjets issued from micro nozzles with fully expanded jet Mach numbers Mj ranging from 1.47 - 1.71, where the design Mach number is Md = 1.5. A complex phenomenon for free square microjets called axis switching is clearly observed with two types “upright” and “diagonal” of “cross-shaped”. The initial axis-switching is 45° within the first shock-cell range. In addition, from the symmetry and diagonal views of square microjets for the first shock-cells, two different patterns of shock waves are viewed. The shock-cell spacing and supersonic core length for all nozzle pressure ratios are investigated and reported.

A spectral reconstruction algorithm for two-plane Compton cameras

One factor limiting the current applicability extent of hadron therapy is the lack of a reliable method for real time treatment monitoring. The use of Compton imaging systems as monitors requires the correct reconstruction of the distribution of prompt gamma productions during patient irradiation. In order to extract the maximum information from all the measurable events, we implemented a spectral reconstruction method that assigns to all events a probability of being either partial or total energy depositions. The method, implemented in a list-mode maximum likelihood expectation maximization algorithm, generates a four dimensional image in the joint spatial-spectral domain, in which the voxels containing the emission positions and energies are obtained. The analytical model used for the system response function is also employed to derive an analytical expression for the sensitivity, which is calculated via Monte Carlo integration. The performance of the method is evaluated through reconstruction of various experimental and simulated sources with different spatial and energy distributions. The results show that the proposed method can recover the spectral and spatial information simultaneously, but only under the assumption of ideal measurements. The analysis of the Monte Carlo simulations has led to the identification of two important degradation sources: the mispositioning of the gamma interaction point and the missing energy recorded in the interaction. Both factors are related to the high energy transferred to the recoil electrons, which can travel far from the interaction point and even escape the detector. These effects prevent the direct application of the current method in more realistic scenarios. Nevertheless, experimental point-like sources have been accurately reconstructed and the spatial distributions and spectral emission of complex simulated phantoms can be identified.

Joint Optimization of Collimator and Reconstruction Parameters in X-Ray Fluorescence Computed Tomography Using Analytical Point Spread Function and Model Observer.

To jointly optimize collimator design and image reconstruction algorithm in X-ray Fluorescence Computed Tomography (XFCT) for imaging low concentrations of high atomic number (Z) elements in small animal models. Single pinhole (SPH) collimator and three types of multi-pinhole (MPH) collimators were evaluated. MPH collimators with 5, 7, and 9 pinholes using lead, stainless steel and brass were considered. A digital cylindrical phantom (0.5mm3 voxels) of 25mm diameter and 25mm height with a central 5mm diameter and 12.5mm height cylindrical insert containing gold nanoparticles (2:1 insert: background concentration) was modeled. A 125kVp, 2mm Sn filtered x-ray spectrum (0.5 cGy/projection) for gold K-shell XFCT was considered. System matrices were implemented using analytical point spread functions (PSF) for each pinhole collimator. Poisson noise was added to the projection data (16 equiangular views) before image reconstruction using Maximum-Likelihood Expectation-Maximization (ML-EM) algorithm. Signal-present and signal-absent images were generated for the detection task performed by a channelized Hotelling observer (CHO) with 10 Dense Difference-of-Gaussian channels. The area under the curve (AUC) in receiver operating characteristic was used as the image quality metric. A stainless steel focusing type MPH with 7 pinholes and 20 iterations of ML-EM provided the highest AUC. MPH collimators outperformed SPH collimators for XFCT and consistently high AUCs were observed with focusing type MPH designs with 7 pinholes. The combinations of collimator design and image reconstruction parameters that maximized AUC were identified, which could improve the performance of XFCT.