Search

The 4D NURBS-based Cardiac-Torso (NCAT) phantom, which provides a realistic model of the normal human anatomy and cardiac and respiratory motions, is used in medical imaging research to evaluate and improve imaging devices and techniques, especially dynamic cardiac applications. One limitation of the phantom is that it lacks the ability to accurately simulate altered functions of the heart that result from cardiac pathologies such as coronary artery disease (CAD). The goal of this work was to enhance the 4D NCAT phantom by incorporating a physiologically based, finite-element (FE) mechanical model of the left ventricle (LV) to simulate both normal and abnormal cardiac motions. The geometry of the FE mechanical model was based on gated high-resolution x-ray multi-slice computed tomography (MSCT) data of a healthy male subject. The myocardial wall was represented as transversely isotropic hyperelastic material, with the fiber angle varying from -90 degrees at the epicardial surface, through 0 degrees at the mid-wall, to 90 degrees at the endocardial surface. A time varying elastance model was used to simulate fiber contraction, and physiological intraventricular systolic pressure-time curves were applied to simulate the cardiac motion over the entire cardiac cycle. To demonstrate the ability of the FE mechanical model to accurately simulate the normal cardiac motion as well abnormal motions indicative of CAD, a normal case and two pathologic cases were simulated and analyzed. In the first pathologic model, a subendocardial anterior ischemic region was defined. A second model was created with a transmural ischemic region defined in the same location. The FE based deformations were incorporated into the 4D NCAT cardiac model through the control points that define the cardiac structures in the phantom which were set to move according to the predictions of the mechanical model. A simulation study was performed using the FE-NCAT combination to investigate how the differences in contractile function between the subendocardial and transmural infarcts manifest themselves in myocardial SPECT images. The normal FE model produced strain distributions that were consistent with those reported in the literature and a motion consistent with that defined in the normal 4D NCAT beating heart model based on tagged MRI data. The addition of a subendocardial ischemic region changed the average transmural circumferential strain from a contractile value of 0.19 to a tensile value of 0.03. The addition of a transmural ischemic region changed average circumferential strain to a value of 0.16, which is consistent with data reported in the literature. Model results demonstrated differences in contractile function between subendocardial and transmural infarcts and how these differences in function are documented in simulated myocardial SPECT images produced using the 4D NCAT phantom. In comparison to the original NCAT beating heart model, the FE mechanical model produced a more accurate simulation for the cardiac motion abnormalities. Such a model, when incorporated into the 4D NCAT phantom, has great potential for use in cardiac imaging research. With its enhanced physiologically-based cardiac model, the 4D NCAT phantom can be used to simulate realistic, predictive imaging data of a patient population with varying whole-body anatomy and with varying healthy and diseased states of the heart that will provide a known truth from which to evaluate and improve existing and emerging 4D imaging techniques used in the diagnosis of cardiac disease.

The performance of the Channelized Hotelling Observer (CHO) was compared to that of human observers for determining optimum parameters for the iterative OS-EM image reconstruction method for the task of defect detection in myocardial SPECT images. The optimum parameters were those that maximized defect detectability in the SPECT images. Low noise, parallel SPECT projection data, with and without an anterior, inferior or lateral LV wall defect, were simulated using the Monte Carlo method. Poisson noise was added to generate noisy realizations. Data were reconstructed using OS-EM at 1 & 4 subsets/iteration and at 1, 3, 5, 7 & 9 iterations. Images were converted to 2D short-axis slices with integer pixel values. The CHO used 3 radially-symmetric, 2D channels, with varying levels of internal observer noise. For each parameter setting, 600 defect-present and 600 defect-absent image vectors were used to calculate the detectability index (dA). The human observers rated the likelihood that a defect was present in a specified location. For each parameter setting, the AUC was estimated from 48 defect-present and 48 defect-absent images. The combined human observer results showed the optimum parameter setting could be in the range 5-36 updates ([number of subsets]/iteration e number of iterations). The CHO results showed the optimum parameter setting to be 4-5 updates. The performance of the CHO was much more sensitive to the reconstruction parameter setting than was that of the human observers. The rankings of the CHO detectability values did not change with varying levels of internal noise.

The objective of this study is to investigate the effects of respiratory motion (RM) on defect detection in Tc-99m sestamibi myocardial perfusion SPECT (MPS) using a phantom population that includes patient variability. Three RM patterns are included, namely breath-hold, slightly enhanced normal breathing, and deep breathing. For each RM pattern, six 4-D NCAT phantoms were generated, each with anatomical variations. Anterior, lateral and inferior myocardial defects with different sizes and contrasts were inserted. Noise-free SPECT projections were simulated using an analytical projector. Poisson noise was then added to generate noisy realizations. The projection data were reconstructed using the OS-EM algorithm with 1 and 4 subsets/iteration and at 1, 2, 3, 5, 7, and 10 iterations. Short-axis images centered at the centroid of the myocardial defect were extracted, and the channelized Hotelling observer (CHO) was applied for the detection of the defect. The CHO results show that the value of the area under the receiver operating characteristics (ROC) curve (AUC) is affected by the RM amplitude. For all the defect sizes and contrasts studied, the highest or optimal AUC values indicate maximum detectability decrease with the increase of the RM amplitude. With no respiration, the ranking of the optimal AUC value in decreasing order is anterior then lateral, and finally inferior defects. The AUC value of the lateral defect drops more severely as the RM amplitude increases compared to other defect locations. Furthermore, as the RM amplitude increases, the AUC values of the smaller defects drop more quickly than the larger ones. We demonstrated that RM affects defect detectability of MPS imaging. The results indicate that developments of optimal data acquisition methods and RM correction methods are needed to improve the defect detectability in MPS. Index Terms--Mathematical observer, myocardial perfusion SPECT, phantom population, channelized Hotelling observer (CHO).

The purpose of this study is to investigate optimal respiratory gating schemes using different numbers of gates and placements within the respiratory cycle for reduction of respiratory motion (RM) artifacts in myocardial SPECT. The 4D NCAT phantom with its realistic respiratory model was used to generate 96 3D phantoms equally spaced over a complete respiratory cycle modeling the activity distribution from a typical Tc-99m Sestamibi study with the maximum movement of the diaphragm set at 2 cm. The 96 time frames were grouped to simulate various gating schemes (1, 3, 6, and 8 equally spaced gates) and different placements of the gates within a respiratory cycle. Projection data, including effects of attenuation, collimator-detector response and scatter, from each respiratory gate and each gating scheme were generated and reconstructed using the OS-EM algorithm with correction for attenuation. Attenuation correction was done with average attenuation maps for each gate and over the entire respiratory cycle. Bull's-eye polar plots generated from the reconstructed images for each gate were analyzed and compared to assess the effect of RM. RM artifacts were found to be reduced the most when going from the ungated to the gated case. No significant difference was found in attenuation compensated images between the use of gated and average attenuation maps. Our results indicate that the extent of RM artifacts is dependent on the placement of the gates in a gating scheme. Artifacts are less prominent in gates near end-expiration and more prominent near end-inspiration. This dependence on gate placement decreases when going to higher numbers of gates (6 and higher). However, it is possible to devise a non-uniform time interval gating scheme with 3 gates that will produce results similar to those using a higher number of gates. We conclude that respiratory gating is an effective way to reduce RM artifacts. Effective implementation of respiratory gating to further improve quantitative myocardial SPECT requires optimization of the gating scheme based on the amount of respiratory motion of the heart during each gate and the placement of the gates within the respiratory cycle. Index Terms--Imaging, motion compensation, simulation, single photon emission computed tomography (SPECT).

With continuing improvements in spatial resolution of positron emission tomography (PET) scanners, small patient movements during PET imaging become a significant source of resolution degradation. This work develops and investigates a comprehensive formalism for accurate motion-compensated reconstruction which at the same time is very feasible in the context of high-resolution PET. In particular, this paper proposes an effective method to incorporate presence of scattered and random coincidences in the context of motion (which is similarly applicable to various other motion correction schemes). The overall reconstruction framework takes into consideration missing projection data which are not detected due to motion, and additionally, incorporates information from all detected events, including those which fall outside the field-of-view following motion correction. The proposed approach has been extensively validated using phantom experiments as well as realistic simulations of a new mathematical brain phantom developed in this work, and the results for a dynamic patient study are also presented. Index Terms--Positron emission tomography (PET) iterative reconstruction, motion correction, system response.

Rotating multisegment slant-hole (RMSSH) single photon emission computed tomography (SPECT) is suitable for detecting small and low-contrast breast lesions since it has much higher detection efficiency than conventional SPECT with a parallel-hole collimator and can image the breast at a closer distance. Our RMSSH SPECT reconstruction extends a previous rotation-shear transformation-based method to include nonuniform attenuation and collimator-detector response (CDR) compensation. To evaluate our reconstruction method, we performed two phantom simulation studies with 1) an isolated breast and 2) a breast phantom attached to the body torso. The reconstructed RMSSH SPECT images with attenuation and CDR compensation showed improved quantitative accuracy and less image artifacts than without. To evaluate the clinical efficacy of RMSSH SPECT mammography, we used a simulation study to compare with planar scintimammography in terms of the signal-to-noise ratio (SNR) value of a breast lesion. The RMSSH SPECT reconstruction images showed higher SNR value than the planar scintimammography images and even more so as we applied compensation for attenuation and collimator detector response. We conclude that attenuation and CDR compensation provide RMSSH SPECT mammography images with improved quality and quantitative accuracy.

To further improve pinhole single photon emission computed tomography (SPECT) imaging, there have been increasing interests in the use of nonstandard collimator designs and/or acquisition geometries. Homogeneous coordinates provide a compact and convenient framework to unify the geometric descriptions of the projection operators for these different imaging geometries, which may facilitate the implementation of iterative reconstruction algorithms and the investigation of crucial geometric calibration problems in pinhole SPECT. In this work, these advantages were demonstrated through three examples, namely, multipinhole SPECT, pinhole SPECT with a helical scanning orbit, and pinhole SPECT with dual detectors. Specifically, we showed adaptable implementations of iterative image reconstruction algorithms and translatable strategies for efficient geometric calibrations through unifying projection operators of the aforementioned imaging geometries. Notably, the unified geometric descriptions of multipinhole and single pinhole projection operators allowed us to derive that one can effectively calibrate a multipinhole geometry using only two point sources without measuring their distance. Experimental studies were performed to demonstrate the validity of our approaches, which may be extended to other pinhole SPECT and cone-beam X-ray computed tomography imaging geometries. Index Terms--Calibration, emission computed tomography, geometry, image reconstruction, single photon.

Receiver operating characteristic (ROC) analysis is well established in the evaluation of systems involving binary classification tasks. However, medical tests often require distinguishing among more than two diagnostic alternatives. The goal of this work was to develop an ROC analysis method for three-class classification tasks. Based on decision theory, we developed a method for three-class ROC analysis. In this method, the objects were classified by making the decision that provided the maximal utility relative to the other two. By making assumptions about the magnitudes of the relative utilities of incorrect decisions, we found a decision model that maximized the expected utility of the decisions when using log-likelihood ratios as decision variables. This decision model consists of a two-dimensional decision plane with log likelihood ratios as the axes and a decision structure that separates the plane into three regions. Moving the decision structure over the decision plane, which corresponds to moving the decision threshold in two-class ROC analysis, and computing the true class 1, 2, and 3 fractions defined a three-class ROC surface. We have shown that the resulting three-class ROC surface shares many features with the two-class ROC curve; i.e., using the log likelihood ratios as the decision variables results in maximal expected utility of the decisions, and the optimal operating point for a given diagnostic setting (set of relative utilities and disease prevalences) lies on the surface. The volume under the three-class surface (VUS) serves as a figure-of-merit to evaluate different data acquisition systems or image processing and reconstruction methods when the assumed utility constraints are relevant. Index Terms--Ideal observers, ROC analysis, three-class classification.

In [sup.201]T1/[sup.99m]Tc dual-isotope simultaneous-acquisition (DISA) myocardial imaging, crosstalk due to Tc photons results in significant contamination of the T1 data. The objective of this work is to seek the acquisition parameters (i.e., energy window width and center) that have the optimal tradeoff between minimizing the crosstalk and maximizing the detection efficiency. The optimization criterion was based on maximizing an ideal observer signal-to-noise ratio (SNR) for the myocardial defect detection task using single-isotope and DISA projection images acquired from a torso phantom. For single-isotope images, the optimal energy windows (width/center: 26 keV/75 keV and 28 keV/165 keV for [sup.201]T1, 30 keV/142 keV for [sup.99m]Tc) are wider than typical windows. For DISA imaging, the optimal windows varied with the [sup.99m]Tc to [sup.201]T1 activity ratio and are thus likely to depend on the uptake ratio in each patient. Using the optimal ratio 2.25-2.75 (148 MBq [sup.201]T1 and 333-407 MBq [sup.99m]Tc) with the corresponding optimal windows (22 keV/72 keV, 24 keV/167 keV, and 24 keV/140 keV) gives [sup.201]T1 images with substantially increased SNRs as well as [sup.99m]Tc images with SNRs same as those of 370 MBq [sup.99m]Tc-only images. However, without the addition of crosstalk compensation, the use of the optimal activity and energy windows alone is likely not sufficient to restore the DISA TI SNR to that of TI-only image. Index Terms--Listmode, optimization, simultaneous dual-isotope imaging, [sup.99m]TC, [sup.201]T1.

Partial volume (PV) effects degrade the quantitative accuracy of SPECT brain images. In this paper, we extended a PV compensation (PVC) method originally developed for brain PET, the geometric transfer matrix (GTM) method, to brain SPECT using iterative reconstruction-based compensations. In the GTM method a linear transform between the true regional activities and the measured results was assumed. Elements of the GTM were calculated by projecting and reconstructing maps with uniform regions representing different structures. However, with iterative reconstruction methods, especially when reconstruction-based compensation for detector response was applied, we found that it was important to treat the region maps as a perturbation to the reconstructed image in the estimation of the GTM. This modified method, termed perturbation-based GTM (pGTM) was evaluated using Monte Carlo (MC) simulated and experimentally acquired data. Results showed great improvement of the quantitative accuracy in brain SPECT imaging. For MC simulated data, PVC using pGTM reduced the underestimation of striatal activities from 30% to less than 1.2%. For experimental data, PVC using pGTM reduced the underestimation of striatal activities from 36% to less than 7.8%. The underestimation of the striatum to background activity ratio was also improved from 31% to 2.7%. Index Terms--Brain SPECT imaging, partial volume effect compensation, quantification.

This study evaluates myocardial SPECT images obtained from parallel-hole (PH) and fan-beam (FB) collimator geometries using both circular-orbit (CO) and noncircular-orbit (NCO) acquisitions. A newly developed 4-D NURBS-based cardiac-torso (NCAT) phantom was used to simulate the [sup.99m]Tc-sestamibi uptakes in human torso with myocardial defects in the left ventricular (LV) wall. Two phantoms were generated to simulate patients with thick and thin body builds. Projection data including the effects of attenuation, collimator-detector response and scatter were generated using SIMSET Monte Carlo simulations. A large number of photon histories were generated such that the projection data were close to noise free. Poisson noise fluctuations were then added to simulate the count densities found in clinical data. Noise-free and noisy projection data were reconstructed using the iterative OS-EM reconstruction algorithm with attenuation compensation. The reconstructed images from noisy projection data show that the noise levels are lower for the FB as compared to the PH collimator due to increase in detected counts. The NCO acquisition method provides slightly better resolution and small improvement in defect contrast as compared to the CO acquisition method in noise-free reconstructed images. Despite lower projection counts the NCO shows the same noise level as the CO in the attenuation corrected reconstruction images. The results from the channelized Hotelling observer (CHO) study show that FB collimator is superior to PH collimator in myocardial defect detection, but the NCO shows no statistical significant difference from the CO for either PH or FB collimator. In conclusion, our results indicate that data acquisition using NCO makes a very small improvement in the resolution over CO for myocardial SPECT imaging. This small improvement does not make a significant difference on myocardial defect detection. However, an FB collimator provides better defect detection than a PH collimator with similar spatial resolution for myocardial SPECT imaging. Index Terms--Channelized Hotelling observer (CHO), fan beam, ROC analysis, SPECT.

The goal of this study was to develop and apply a population of phantoms that realistically models patient variability and use it to optimize and evaluate different compensation methods used during reconstruction process with respect to defect detection in myocardial SPECT images. Various combinations of attenuation, detector response and scatter compensation were used in this study. A major difference between this and previous studies was that the level of realism was significantly increased by inclusion of variability in heart and organ uptakes, in the heart size and orientation, and in the defect size and contrast. In this study we used a population of 24 4-D NCAT phantoms [1] (half male, half female) recently developed with statistical models for organ uptake and organ size based on clinical data. Almost noise-free projection data of the torso, heart, liver, lungs, and other organs were simulated for each phantom using the SIMSET MC simulation code. They were then combined to form 72 sets of projections for each phantom using randomly sampled activity ratios from a clinically realistic distribution. Poisson noise was then added to the projection data. We applied the channelized hotelling observer (CHO) and receiver operating characteristic (ROC) analysis to optimize iteration number for OSEM and cutoff frequency of a 3-D post-reconstruction Butterworth filter. We found that the area under the ROC curve (AUC) values were reduced compared to a previous study that included significantly less phantom variability, even though the defect contrast was higher and noise level was lower. The resulting AUC values were similar to those obtained using patient data. We found, in agreement with the previous study, that including compensation for more effects resulted in improved defect detectability. However, the optimal filter cutoff frequency was increased compared to the previous study. These studies demonstrate the importance of including realistic levels of phantom variability in myocardial perfusion studies using simulated data. Index Terms--Attenuation compensation, mathematical observer, phantom population, scatter compensation, SPECT.

The four-dimensional (4-D) NCAT phantom was developed to realistically model human anatomy based on the visible human data and cardiac and respiratory motions based on 4-D tagged magnetic resonance imaging and respiratory-gated CT data from normal human subjects. Currently, the 4-D NCAT phantom does not include the airway tree or its motion within the lungs. Also, each lung is defined with a single surface; the individual lobes are not distinguished. The authors further the development of the phantom by creating dynamic models for the individual lung lobes and for the airway tree in each lobe. NURBS surfaces for the lobes and an initial airway tree model (~ 4 generations) were created through manual segmentation of the visible human data. A mathematical algorithm with physiological constraints was used to extend the original airway model to fill each lobe. For each parent airway branch inside a lobe, the algorithm extends the airway tree by creating two daughter branches modeled with cylindrical tubes. Parameters for the cylindrical tubes such as diameter, length, and angle are constrained based on flow parameters and available lung space. The bifurcating branches are propagated within a lung lobe until it is filled. Once each lobe is filled, the cylindrical tubes are converted into NURBS surfaces and blended with the original airway tree obtained through segmentation. The respiratory model previously developed using the respiratory-gated CT data is then applied to the surfaces of the lobes and airway tree to create the new 4-D respiratory model. This improved model will provide a useful tool in future studies researching the effects of respiratory motion on lung tumor imaging. It is also an important step in advancing the 4-D NCAT for applications in more high-resolution imaging modalities such as x-ray CT. Index Terms--Airway tree, phantom, respiratory motion, segmentation.

Multisegment slant-hole (MSSH) collimators provide much higher detection efficiency at a single view as compared to conventional low-energy high-resolution (LEHR) parallel-hole collimators with the same spatial resolution. The purpose of this study is to evaluate rotating MSSH (RMSSH) SPECT mammography using Monte Carlo Simulation methods. We modified the SIMIND Monte Carlo code to generate RMSSH projection data including the effects of attenuation, collimator-detector response and scatter from a realistic 4-D NCAT phantom with attached breasts and lesions placed in the central, medial, and lateral regions of the breast. The projection data were reconstructed using analytic and iterative image reconstruction methods. For comparison, we examine planar scintimammography images obtained by placing an LEHR collimator with a resolution of 1 cm at 15.2 cm, which is the same as that of a four-segment slant-hole (4SSH) collimator at the center of its common volume-of-view, in close proximity to the lateral wall of the breast. Also, we simulated the inclusion of a lead shield that covers the body to keep photons emitted from the other organs from reaching the MSSH collimator. Our results indicated that, depending on the lesion location, the spatial resolution of the RMSSH SPECT images may be comparable or slightly poorer than that of the planar scintimammographic images obtained using a LEHR collimator while the lesion contrast is significantly higher at all lesion locations. Shielding of photon emissions from the other organs substantially improves the RMSSH SPECT image quality. We conclude that the Monte Carlo simulation methods are useful in evaluating the application of RMSSH SPECT to breast imaging as compared to conventional planar scintimammography and that RMSSH SPECT provides higher overall image quality. Index Terms--Attenuation, biomedical nuclear imaging, Monte Carlo methods, single photon emission computed tomography (SPECT).

Pinhole collimation of gamma rays to image distributions of radiolabeled tracers is considered promising for use in small animal imaging. The recent availability of transgenic mice, coupled with the development of [sup.125]I and [sup.99m]Tc labeled tracers, has allowed the study of a range of human disease models while creating demand for ultrahigh resolution imaging devices. We have developed a compact gamma camera that, in combination with pinhole collimation, allows for accessible, ultrahigh resolution in vivo single photon emission computed tomography (SPECT) imaging of small animals. The system is based on a pixilated array of NaI(TI) crystals coupled to an array of position sensitive photomultiplier tubes (PSPMTs). Interchangeable tungsten pinholes with diameters ranging from 0.5 to 3 mm are available, allowing the camera to be optimized for a variety of imaging situations. We use a three dimensional maximum likelihood expectation maximization (MLEM) algorithm to reconstruct the images. Our evaluation indicates that high quality, submillimeter spatial resolution images can be achieved in living mice. Reconstructed axial spatial resolution was measured to be 0.53, 0.74, and 0.96 mm full width at half maximum (FWHM) for rotation radii of 1, 2, and 3 cm, respectively, using the 0.5-mm pinhole. In this configuration, sensitivity is comparable to that of a high-resolution parallel hole collimator. SPECT images of hot- and cold-rod phantoms and a highly structured monkey brain phantom illustrate that high quality images can be obtained with the system. Images of living mice demonstrate the ability of the system to obtain high-resolution images in vivo. The effect of object size on the quantitative assessment of isotope distributions in an image was also studied. Index Terms--Animal imaging, pin hole collimation, single photon emission computed tomography (SPECT), three-dimensional (3-D) image reconstruction.

In this paper, we apply the channelized Hotelling observer (CHO) using a defect detection task to the optimization and evaluation of three-dimensional iterative reconstruction-based compensation methods for myocardial perfusion single-photon emission computed tomography (SPECT). We used a population of 24 mathematical cardiac-torso phantoms that realistically model the activity and attenuation distribution in three classes of patients: females, and males with flat diaphragms and raised diaphragms. Projection data were generated and subsequently reconstructed using methods based on the ordered subsets-expectation maximization (OSEM) algorithm. The methods evaluated included compensation for attenuation, detector response blurring, and scatter in various combinations. We applied the CHO to optimize the number of iterations for OSEM and the cutoff frequency and order of a three-dimensional postreconstruction Butterworth filter. Using the optimal parameters, we then compared the compensation methods. The index of comparison in these studies was the area under the receiver operating characteristics curve (AUC) for the CHO. We found that attenuation compensation with either detector response or scatter compensation gave statistically significant increases in the AUC compared to attenuation compensation alone. The greatest increase in the AUC occurred when all three compensations were applied. These results indicate that compensation for detector response and scatter, in addition to attenuation compensation, will improve defect detectability in myocardial SPECT images. Index Terms--Attenuation compensation, mathematical observers, scatter compensation, SPECT.

Respiratory motion can cause artifacts in myocardial single photon emission computed tomography (SPECT) images, which can lead to the misdiagnosis of cardiac diseases. One method to correct for respiratory artifacts is through respiratory gating. We study the effectiveness of respiratory gating through a simulation study using the newly developed four-dimensional (4-D) NURBS-based cardiac-torso (NCAT) phantom. The organ shapes in the 4-D NCAT phantom are formed using nonuniform rational b-splines (NURBS) and are based on detailed human image data. With its basis on actual human data, the 4-D NCAT phantom realistically simulates human anatomy and motions such as the cardiac and respiratory motions. With the 4-D NCAT phantom, we generated 128 phantoms over one respiratory cycle (5 s per cycle) with the diaphragm and heart set to move a total of 4 cm from end-inspiration to end-expiration. The heart was set to beat with a normal contractile motion at a rate of 1 beat per second resulting in a total of five heart cycles. We divide the respiratory cycle into different numbers of respiratory gates (16, 8, and 4) by summing the phantoms. For each gate, we generate its projection data using an analytical projection algorithm simulating the effects of attenuation, scatter, and detector response. We then reconstruct the projections using an iterative OS-EM algorithm compensating for the three effects. The reconstructed images for each gating method were examined for artifacts due to the respiratory motion during that gate. We found that respiratory artifacts are significantly reduced if the respiratory motion of the heart that occurs during a gating time period is 1 cm or less. We conclude that respiratory gating is an effective method for reducing effects due to respiration. The timing of the respiratory gates for reduced image artifacts is dependent on the extent of the heart's motion during respiration. Index Terms--Biomedical image processing, biomedical nuclear imaging, image analysis, motion compensation, respiratory system, simulation software, single photon emission computed tomography (SPECT).

In simultaneous Tl-201 and Tc-99m dual isotope imaging, Pb X-rays are a significant source of contamination in the Tl image. In order to characterize this contamination, we simulated images of line sources in the Tc and Tl photopeak windows using an experimentally verified Monte Carlo program. The line sources were placed at various distances from the collimator face and emitted photons with energies from 88 to 140.5 keV. The Pb X-ray contamination line source response function in the Tl window was fitted well by a Gaussian plus an exponential function. The width of these two functions changed linearly with distance. Parameterization of the Pb X-ray response was done by simultaneously fitting the Pb X-ray response functions at various distances with fitting functions that were determined empirically to model the distance dependence of the Gaussian and exponential components of the Pb X-ray response. The parameterized model of Pb X-ray contamination is useful in developing methods to model Pb X-ray contamination in Tl-201 data by Tc-99 m. This Pb X-ray contamination model can be used in iterative reconstruction-based cross-talk compensation for simultaneous Tc/Tl dual isotope imaging. Index Terms--Monte Carlo N-Particle (MCNP), Monte Carlo simulation, Pb X-ray contamination, simultaneous dual-isotope imaging, Tc-99m, Tl-201.

We develop a realistic computerized heart phantom for use in medical imaging research. This phantom is a hybrid of realistic patient-based phantoms and flexible geometry-based phantoms. The surfaces of heart structures are defined using non-uniform rational B-splines (NURBS), as used in 3D computer graphics. The NURBS primitives define continuous surfaces allowing the phantom to be defined at any resolution. Also, by fitting NURBS to patient data, the phantom is more realistic than those based on solid geometry. An important innovation is the extension of NURBS to the fourth dimension, time, to model heart motion. Points on the surfaces of heart structures were selected from a gated MRI study of a normal patient. Polygon surfaces were fit to the points for each time frame, and smoothed. 3D NURBS surfaces were fit to the smooth polygon surfaces and then a 4D NURBS surface was fit through these surfaces. Each of the principal 4D surfaces (atria, ventricles, inner and outer walls) contains approximately 200 control points. We conclude that 4D NURBS are an efficient and flexible way to describe the heart and other anatomical objects for a realistic phantom.

The purpose of this work was to evaluate lesion detectability with and without nonuniform attenuation compensation (AC) in myocardial perfusion SPECT imaging in women using an anthropomorphic phantom and receiver operating characteristics (ROC) methodology. Breast attenuation causes artifacts in reconstructed images and may increase the difficulty of diagnosis of myocardial perfusion imaging in women. The null hypothesis tested using the ROC study was that nonuniform AC does not change the lesion detectability in myocardial perfusion SPECT imaging in women. We used a filtered backprojection (FBP) reconstruction algorithm and Chang's single iteration method for AC. In conclusion, with our proposed myocardial defect model nuclear medicine physicians demonstrated no significant difference for the detection of the anterior wall defect; however, a greater accuracy for the detection of the inferior wall defect was observed without nonuniform AC than with it (P-value = 0.0034). Medical physicists did not demonstrate any statistically significant difference in defect detection accuracy with or without nonuniform AC in the female phantom.

We study the statistical convergence properties of two fast iterative reconstruction algorithms, the rescaled block-iterative (RBI) and ordered subset (OS) EM algorithms, in the context of cardiac SPECT with 3D detector response modeling. The Monte Carlo method was used to generate nearly noise-free projection data modeling the effects of attenuation, detector response, and scatter from the MCAT phantom. One thousand noise realizations were generated with an average count level approximating a typical Tl-201 cardiac study. Each noise realization was reconstructed using the RBI and OS algorithms for cases with and without detector response modeling. For each iteration up to twenty, we generated mean and variance images, as well as covariance images for six specific locations. Both OS and RBI converged in the mean to results that were close to the noise-free ML-EM result using the same projection model. When detector response was not modeled in the reconstruction, RBI exhibited considerably lower noise variance than OS for the same resolution. When 3D detector response was modeled, the RBI-EM provided a small improvement in the tradeoff between noise level and resolution recovery, primarily in the axial direction, while OS required about half the number of iterations of RBI to reach the same resolution. We conclude that OS is faster than RBI, but may be sensitive to errors in the projection model. Both OS-EM and RBI-EM are effective alternatives to the ML-EM algorithm, but noise level and speed of convergence depend on the projection model used.

The effect of scatter correction on the accuracy of lesion detection in single-photon emission computed tomography (SPECT) imaging requires analysis of observer performance. Experiments were designed to evaluate the class of correction methods that subtract counts. Simulations were used to approximate liver imaging with labeled antibodies. The lesion was a 2.5-cm-diameter, spherical, 'cold' tumor. Ramp-filtered backprojection and noniterative Chang attenuation compensation were used to approximate clinical practice. Perfect scatter rejection, defined as images containing only primary (nonscattered) photons, was selected as the ideal case. These images were compared with uncorrected images for conditions of both low- and high-scatter fractions (SF). The dual photopeak window (DPW) method was also tested to evaluate a practical subtraction correction. Receiver operating characteristic (ROC) experiments were conducted under signal-known-exactly (SKE) conditions, using the area under the curve as the index of accuracy. A statistically significant difference in detection was found only in a few experiments when scatter rejection was compared with no correction. The results suggest that scatter correction does not necessarily assure improved detection accuracy at a statistically significant level. However, corrections that produce conditions similar to ideal scatter rejection may offer such improvement in detection, particularly for cases of high SF's. Index Terms - Lesion detection, ROC analysis, scatter correction.

We investigate the effects of initial conditions in the iterative maximum-likelihood (ML) reconstruction of fan-beam transmission projection data with truncation. In an iterative ML reconstruction, the estimate of the transmission reconstructed image in the previous iteration is multiplied by some factors to obtain the current estimate. Normally, a fiat initial condition (FIC) or an image with equal positive pixel values is used as initial condition for an ML reconstruction. Usage of FIC has also been perceived as a way of preventing any bias on the reconstruction which may have come from the initial condition. When projection data have truncation, we show that using an FIC in an ML iterative reconstruction can introduce a bias to the reconstruction inside the densely sampled region (DSR), whose projection data have no truncation at any angle. To reduce this bias, we propose to use the largest right singular vector (LRSV) of the system matrix as the initial condition, and demonstrate that the bias can be reduced with the LRSV. When data truncation is reduced, the LRSV approaches the FIC. This result does not contradict to the use of FIC when projection data are not truncated. We also demonstrate that the reconstructed transmission image using LRSV as initial condition provides a more accurate attenuation coefficient distribution than that using FIC. However, the improvement is mostly in the area outside the DSR. Index Terms - Fan beam, image reconstruction, initial condition, SPECT, transmission imaging.

In this work singular value decomposition (SVD) techniques are used to investigate how the use of low energy photons and multiple energy windows affects the noise properties of Tc-99m SPECT imaging. We have previously shown that, when modeling scatter in the projector and backprojector of iterative reconstruction algorithms, simultaneous reconstruction from multiple energy window data can result in very different noise characteristics. Further, the properties depend upon the width and number of energy windows used. To investigate this further, we have generated photon transport matrices using models for scatter, an elliptical phantom containing cold rods of various sizes, and a number of multiple energy window acquisition schemes. Transfer matrices were also generated for the cases of perfect scatter rejection and ideal scatter subtraction. The matrices were decomposed using SVD, and signal power and projection space variance spectra were computed using the basis formed by the left singular vectors. Results indicate very different noise levels for the various energy window combinations. The perfect scatter rejection case resulted in the lowest variance spectrum, and reconstruction-based scatter compensation performed better than the scatter subtraction case. When including lower energy photons in reconstruction-based scatter compensation, using a series of multiple energy windows outperformed a single large energy window. One multiple window combination is presented which achieves a lower variance spectrum than the standard 20% energy window, indicating the potential for using low energy photons to improve the noise characteristics of SPECT images.

In simultaneous Tc-99m TCT and T1-201 SPECT data acquisition, the emission images are degraded by not only scatter from the emission source, but also cross-contamination from the transmission photons. Also, the transmission images are degraded by scatter from the transmission source and cross-contamination from the emission photons. In this work, a criterion based on maximizing a Figure-of-Merit (FOM) was used to reduce image degradations by optimizing the position and width of the energy windows used in the data acquisition in a triple-camera fan-beam SPECT system. The FOM determines a compromise between increased detection efficiency of primary photons and reduction of scatter and cross-contamination photons. Scatter in the projection data was modeled using Monte Carlo (MC) simulation methods. Experimental studies were performed to verify the simulations and to estimate the amount of Pb x-rays and scatter in the collimator which were not modeled in the MC program. Results suggest that cross-contamination degrades the emission and transmission data. However, the contamination does not significantly alter the optimal energy window settings, which should be centered at [approximately]77 keV with a window width of [approximately]34% for detecting the Tl emission photons, and centered at 140 keV with a window width of [approximately]20% for detecting the Tc transmission photons.

We have developed a prototype emission-transmission computed tomographic (ETCT) system, which can acquire both radionuclide-emission and dual-energy X-ray transmission data. The dual-energy X-ray data are used to synthesize an object- and energy-specific attenuation map which is incorporated into an ML-EM algorithm to reconstruct an attenuation-corrected SPECT image. The technique was applied to a thorax-like phantom with regions simulating lung, background and cardiac regions, and bone. SPECT images reconstructed with no attenuation correction, or with a uniform water-equivalent map, contained increased count levels in the lung region and reduced count levels in the cardiac region as compared with the radioactivity concentrations contained in the phantom. Use of an object-specific attenuation map improved the accuracy of the reconstruction, but overestimated the activity in the cardiac chamber by about 10%. This error was reduced to about 5% when the attenuation coefficients were modified to include the effects of scattered radiation.

We have compared noise properties of emission tomographic images reconstructed using maximum likelihood-expectation maximization (ML-EM) and filtered-backprojection (FBP) algorithms. Noise comparisons were made in terms of the covariance matrix which gives information on the noise magnitude and noise correlations. We studied noise properties as a function of iteration for ML-EM and as a function of noise apodization filter for FBP. We show that FBP reconstruction spreads noise variance from image regions containing high count densities into regions of low count densities. We also demonstrate at lower FBP filter cutoff frequencies the noise is correlated over relatively long distances and the correlation function has deep negative sidelobes. For ML-EM reconstruction we show that little noise variance is spread from high count density regions into low count regions and that at lower iteration number the noise is correlated over shorter distances than for FBP. Also for ML-EM, the correlation function has no negative side-lobes at low iterations numbers, although as the iteration number increases the function begins to resemble the FBP correlation function. We conclude from these observations that ML-EM reconstruction offers properties which may exceed FBP in terms of detectability for certain emission tomographic imaging situations.

We study the relationship between the choice of parameters for a generalized Gibbs prior for the MAP-EM (Maximum A Posteriori, Expectation Maximization) algorithm, and the model of the projection/backprojection process used in SPECT reconstruction. A realistic phantom, derived from an X-ray CT study and average T1-201 uptake distributions in patients, was used in the investigation. Simulated projection data including nonuniform attenuation, detector response, scatter, and Poisson noise were generated. From this data set, reconstructions were created using a MAP-EM technique with a generalized Gibbs prior, which is designed to smooth noise with minimal smoothing of edge information. The Gibbs prior has three adjustable parameters: one which determines the overall weight placed on the prior in the reconstruction process, and two others which affect the relative smoothing of noise and edges in the reconstructed image estimates. Reconstructions were performed over several different values of the prior parameters for three projector/backprojector models: one with no compensations at all, one incorporating only nonuniform attenuation compensation, and one incorporating both nonuniform attenuation and detector response compensations. Analyzing several measures of image quality in a region-of-interest surrounding the myocardium, we conclude that, for each projection model, there is an optimal value of the weighting parameter which decreases as the projection process is modeled more accurately. We conclude also that the use of the Gibbs prior together with nonuniform attenuation and detector response compensations offers improved quantitative accuracy over Maximum Likelihood EM reconstruction with the same compensations.

Maximum a posteriori (MAP) reconstruction in SPECT has been shown to have significant advantages over traditional maximum likelihood (ML) methods in terms of noise performance, but these advantages are highly dependent on the choice of the distribution used to model the prior knowledge about the solution image. Several Gibbs prior distributions have been proposed in the literature, but there has been relatively little work comparing and contrasting the effects of these prior distributions on actual reconstructions. We demonstrate the effects of several of these prior distributions in terms of noise characteristics, edge sharpness, and overall quantitative accuracy of the final estimates obtained from an iterative MAP procedure applied to data from a realistic chest phantom. We also examine the effects of the adjustable parameters built into the prior distribution on these properties. We find that these parameter values influence the noise and edge characteristics of the final estimate and can generate reconstructions closer to the actual solution than ML. In addition, we find that the choice of the shape of the prior distribution affects the noise characteristics and edge sharpness in the final estimate.

This study investigated several aspects of the use of half-fanbeam collimation with a 90 [degrees] dual-camera system for myocardial single photon emission computed tomography (SPECT) imaging. The detection efficiency, relative to parallel-hole collimation, was evaluated for a range of focal lengths and radii of rotation (ROR) to determine if there was an optimum focal length, which maximized the detection efficiency. In addition, sinograms were constructed and a simulation study was performed to determine if there was an optimal camera system rotation that maximized the total acquired myocardial counts while providing sufficient angular sampling for the myocardial region. Finally, artifacts in images reconstructed from data acquired over various system rotations were evaluated using simulated and experimental data. There existed an optimal collimator focal length for a given ROR; but it varied with ROR. Relative to parallel-hole collimation, the detection efficiency for half-fanbeam collimation was roughly 20% greater, for cast collimators constructed using the same pins and thickness. The theoretical minimum system rotation for sufficient sampling of the myocardial region ranged from 124 [degrees] to 148 [degrees] for RORs ranging from 13 to 25 cm, respectively. The total number of acquired myocardial counts was relatively constant for system rotations of 90 [degrees] to 360 [degrees]. Myocardial SPECT images reconstructed iteratively with attenuation compensation from half-fanbeam data collected over system rotations ranging from 135 [degrees] to 360 [degrees] showed no artifacts in the myocardial region. Based on these results, we concluded that there was no single optimum system rotation, but that a system rotation of 180 [degrees] centered at 45 [degrees] left anterior oblique was a good, practical minimum rotation. Half-fanbeam collimation is a useful alternative system configuration for myocardial SPECT imaging.

The new PET tracer, 18 F-flurpiridaz, with high myocardial extraction allows quantitative myocardial blood flow (MBF) estimation from dynamic PET data and tracer kinetic modeling. The goal of this study is to determine the optimal imaging protocols and parameters using a realistic simulation study. The time activity curves (TACs) of different tissue organs from a 30-s infusion time (IT) of 18 F-flurpiridaz in a dynamic PET study were extracted from a previous study. The TACs at different time points were incorporated in a series of realistic 3D XCAT phantoms from which the parameters of a 2-compartment model and the ‘true’ MBF of 18 F-flurpiridaz were determined. The compartmental model was used to generate TACs from 7 additional ITs. PET projection data from the XCAT phantoms were generated using Monte Carlo simulation. They were reconstructed using an OS-EM reconstruction algorithm with different update number (N) to obtain dynamic PET images. The blood and myocardial TACs were derived from the dynamic images from which the MBF and %MBF error was estimated. The %MBF error decreases with increasing N of the OS-EM and levels off after ∼42. The 30-s IT gave the smallest %MBF error that decreases from ∼0.57% to ∼19.40%. The MBF for 2-min, 4-min, 8-min and 16-min IT were statistically significant different from the MBF for 30-s IT (P < 0.05). Too fast or too slow infusion time gave higher %MBF error. The optimal imaging protocol in dynamic 18 F-flurpiridaz PET for accurate quantitative MBF estimation was 30-s IT and N of ∼42 for the OS-EM. [ABSTRACT FROM AUTHOR]