Showing papers on "Ordered subset expectation maximization published in 2006"

Systematic and Distributed Time-of-Flight List Mode PET Reconstruction

Abstract: Philips has recently released the time-of-flight (TOF) PET/CT GEMINI-TF scanner. It uses 4 times 4 times 22 mm3 LYSO crystals, which has 600 ps timing resolution, 12% energy resolution and 4.8 mm spatial resolution. This paper describes the system design and general approach of TOF reconstruction in Philips' GEMINI-TF scanner. A relaxed list mode ordered subset expectation maximization (LMOSEM) algorithm is used in the reconstruction, with chronologically ordered subsets. The sensitivity and emission object in LMOSEM are computed in the spherically symmetric basis function (blob) space. The multiplicative correction factors of detector normalization, isotope decay, system dead-time and crystal timing are pre-corrected for each list mode event. Attenuation, scatter and randoms are corrected in the reconstruction system matrix. A TOF-dependent single scatter simulation (SSS) is implemented for TOF scatter estimation. To accelerate reconstruction, the sensitivity calculation and list mode reconstruction are distributed to multiple processors, using a dynamic load balancing scheme. For this paper, a uniform cylinder phantom with cold and hot cylinder inserts, a NEMA body IEC phantom and a patient study are reconstructed with both TOF and non-TOF reconstructions. We have demonstrated that TOF reconstruction converges faster than non-TOF, and controls noise well than non-TOF. It has better contrast-noise trade-offs than non-TOF for cold regions and small hot lesions.

3D-OSEM iterative image reconstruction for high-resolution PET using precalculated system matrix

Abstract: An efficient iterative image reconstruction methodology is presented, adapted to high-resolution flat-head 3D positron emission tomography cameras. It is based on the ordered subsets expectation maximization algorithm and applies to systems with axial symmetry. The associated system matrix is calculated off-line, including a model of the γ-event detection in the crystal, taking into account photoelectric effect and Compton scattering interactions. The nonzero elements of the sparse system matrix are stored in disc in an efficient way that allows the fast sequential access to the matrix elements during the reconstruction. A detailed calculation is performed for the voxels corresponding to central plane within the field of view (FOV) of the camera and the remaining values of the system matrix are obtained via translations based on the symmetries of the system along the axial dimension. GATE-based simulations have been used for the validation of the results.

Local computed tomography via iterative deblurring

Abstract: X-ray computed tomography is a major imaging modality. An iterative deblurring method is adapted for local reconstruction in parallel-beam and cone-beam geometries, utilizing only x-rays passing through a region of interest. The feasibility is demonstrated in numerical simulation with noise-free and noisy projection data. The iterative deblurring method has the theoretical advantages of maintaining nonnegativity, converging monotonically and minimizing Csiszar's I-divergence.

Fast 3D Iterative Reconstruction of PET Images Using PC Graphics Hardware

Abstract: Using iterative reconstruction algorithms in 3D positron emission tomography (PET) studies produce images with superior quality, however the run time is too long for these algorithms to be used routinely, especially for dynamic studies. Recently several new hardware architectures are available to speedup 3D iterative reconstructions, including the graphics processing unit (GPU), which is very attractive due to its fast performance improvement and low cost. In this study, we investigate the possibility of implementing fast iterative 3D PET reconstruction using commercial graphics hardware. A 3D ordered subset expectation maximization (OSEM) algorithm was implemented and a GPU used to calculate the geometric forward and back projections using a line-integral model. Results show that the GPU can be used for fast 3D iterative PET image reconstructions. Future work would include exploring the possibility of implementing more accurate geometric models on the GPU.

SPECT Image Quality and Quantification

Abstract: Accurate reconstruction methods for SPECT are a pre-requirement for absolute tracer quantification. Exact activity numbers can be extracted from SPECT images by use of iterative reconstruction with appropriate corrections for attenuation, scatter and detector resolution. We have developed an analytical model for SPECT image generation, including attenuation, scatter and reconstruction by filtered back-projection (FBP). Image quality is described in terms of signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR). MatLab simulations were added to include an evaluation of ordered-subset expectation maximization (OSEM) as an example of statistical reconstruction, and for a quantitative assessment of lesion signal recovery. As expected, classical reconstruction without attenuation or scatter correction provides incorrect contrast values, and standard corrections like the Chang method cannot mitigate this problem. Statistical reconstruction allows implementing all corrections and results in better image quality than FBP. Our analysis shows, however, that for adequate signal recovery the contrast-to-noise ratio for OSEM is the same as for FBP. We explain the trade-off between SPECT image quality and quantification by describing reconstruction in terms of signal power spectra (SPS), noise power spectra (NPS) and detective quantum efficiency (DQE).

Parallelization and Runtime Prediction of the ListMode OSEM Algorithm for 3D PET Reconstruction

Abstract: For high-resolution PET (Positron Emission Tomography) image reconstructions, the LM OSEM (ListMode Ordered Subset Expectation Maximization) algorithm proves to be quite appropriate, but it is very time-consuming. In order to improve its runtime, we parallelized the algorithm and implemented it on different classes of parallel computer architectures: with shared, distributed and hybrid memory. These implementations reduce the reconstruction time from more than two hours to six minutes. We suggest an analytical model for predicting parallel LM OSEM runtimes on distributed-memory machines, and verify our model in runtime experiments on different reconstruction problems, which demonstrate a prediction error of less than 10 %. The model allows the user to achieve a desired reconstruction quality while minimizing resource usage.

A new approach of 3D SPECT reconstruction for near-field coded aperture imaging

Abstract: Near-field coded aperture imaging is known to have superior image resolution and count sensitivity over conventional parallel-hole collimated nuclear imaging. There have been several studies in image reconstruction for two-dimensional planar objects using the coded aperture imaging technology. However, coded aperture imaging for three-dimensional (3D) objects has not been extensively investigated. In this paper, a 3D reconstruction method for near-field coded aperture imaging is presented. We first introduce the "out-of-focus" correction factor into the generic expectation maximization (EM) algorithm for 3D near-field coded aperture images with the assumption that the photon emissions of coded aperture projections follow the Poisson statistics. The ordered subset expectation maximization (OSEM) method is then adapted for full 3D coded aperture image reconstruction. A 3D capillary tube phantom filled with 99m Tc radioactive solution was used to evaluate the performance of our methods. A dual-head SPECT camera, one head quipped with a coded aperture module and the other with a parallel-hole collimator, was utilized for image acquisitions. Images were reconstructed using the modified EM and OSEM methods associated with the depth-dependent out-of-focus correction. The preliminary phantom results showed that our methods may have potential of reconstructing 3D near-field coded aperture images and also providing superior image resolution as compared to conventional parallel-hole collimated images.

Fast forward projection and backward projection algorithm using SIMD

Abstract: Recent developments in PET scanners such as the HRRT (High Resolution Research Tomograph) developed by Siemens greatly enhanced their resolution as well as sensitivity, but they increased coincidence lines of response more than 4.5 times 10 generated by as many nuclear detectors as 120,000. This formidable amount of data poses a real problem in the image reconstruction and its applications. It also has been the major bottleneck in further developments of the higher resolution PET scanners. To remedy this problem in the image reconstruction, we developed a new algorithm based on the SIMD (Single Instruction Multiple Data) technique incorporated with the symmetry properties of the projection and backprojection processes, especially in the 3D OSEM algorithm. We refer to this technique as the SSP (Symmetry and SIMD based Projection-backprojection) algorithm. As a demonstration, the algorithm was applied to the OSEM (Ordered Subset Expectation Maximization) 3D algorithm with HRRT data and it effectively reduced the total image reconstruction time to 80 folds.

ROC analysis of PET imaging performance of two PET/CT scanners based on lesion detectability in a torso phantom

Abstract: Positron emission tomography (PET) and computed tomography (CT) are established imaging tools. Integrated PET/CT scanners are now commonly used in radiation treatment planning to delineate accurate tumor margins, to identify nodal involvement, and in post-treatment follow-up assessments. The use of PET/CT images for quantitative selection of dose prescriptions is also being considered. The usefulness of the images produced for these applications depends strongly on the interplay between the performance of the imaging system and the skill of the diagnostician. The purpose of this study was to evaluate the performance of the PET component of two PET/CT scanners using receiver operating characteristic (ROC) analysis of the ability of human observers to identify the presence of simulated tumors in PET images of an anthropomorphic torso phantom. We also assessed the effects of PET acquisition and reconstruction parameters (such as acquisition length and reconstruction method) on the observers' performances and the relationship of performance to physical system factors (such as resolution and sensitivity). Image data of 18F-fluorodeoxyglucose distribution in the phantom and water-filled acrylic spheres were acquired with the PET/CT scanners; the spheres simulate tumors. Data were obtained for a variety of acquisition and reconstruction parameters available on the scanners. Positive and negative images were generated from the acquired image data, representing up to two simulated tumors per image as visualized with the various system parameters. The images were rated for the presence of tumors by "trained" observers (physicians who use radiologic images daily in patient management) and "untrained" observers (medical physicists, post-doctoral researchers, and medical physics graduate students who do not use these images routinely). ROC curves were generated to assess readers' performances relative to the scanners' parameters as characterized by area under the curve (Az). Overall, the untrained group provided little discriminating power in this study, perhaps because of its heterogeneous levels of experience. Trained observers consistently outperformed untrained observers. As well, trained observers performed significantly better using Reveal HD system images than with images generated in the Discovery ST two-dimensional (2D) mode (p < 0.01); however, differences between other pairings of scanners were not significant. Of the three reconstruction algorithms investigated (ordered-subset expectation maximization [OSEM], filtered backprojection, and direct-inversion Fourier transform), OSEM produced the smallest Az for trained observers. Because of its higher sensitivity and higher resolution, the fully 3D Reveal HD facilitated better performance by the trained observers than the Discovery ST 2D mode. To further evaluate the use of PET/CT in radiation therapy treatment planning, we will plan to expand this study to include fused PET/CT images and specific radiation therapy treatment planning tasks, such as drawing clinical target volume margins around tumors.

An improved ordered subsets expectation maximization reconstruction

Abstract: Positron emission computerized tomography (PET) based on Ordered Subsets Expectation Maximization (OS-EM) is usually used to computerized tomography, which imposes the radionuclide to emission positron But it also has some disadvantage as costing long time and bad reconstruction quality Filtered Back Projection (FBP) that has many advantages such as simple structure and short reconstruction time is firstly introduced to the initialization stage to accelerate reconstruction Then, the smoothness method is introduced to improve the reconstruction quality after OS-EM algorithm The results show that this method has the advantages of fast reconstruction speed and good quality

General formulation for X-ray computed tomography

Abstract: Over recent years, various exact cone-beam reconstruction algorithms have been proposed. The derivations of these algorithms are quite complicated, and often difficult to see the fundamental connections among these methods and their key steps. In this paper, we present a straightforward perspective based on the Fourier transform, which is a universal principle for parallel and divergent beam computed tomography (CT). The formulas in this paper are not only consistent with the latest findings in the field but also valid under more general conditions.

WE‐D‐ValB‐01: Effects of PET Reconstruction Parameters On the Delineation of Heterogeneous Target Volumes

Abstract: Purpose As emerging radiotherapy techniques incorporate biological targeting of sub‐tumor volumes, steps must be taken to ensure the validity of the assumed substructures. This study measures the effects of PET image reconstruction on heterogeneous target definitions both in vivo and in a phantom. Method and Materials: A known heterogeneous phantom composed of Y‐86 and Ge‐68 spheres with an F‐18 background altering signal‐to‐background ratios tested the accuracy of reconstructions using ordered subset expectation maximization (OSEM) with varying numbers of iterations and filtered backprojection (FBP) with Hanning, Shepp‐Logan, and ramp filters. In vivo measurements used heterogeneously proliferating tumorimages obtained from a canine tumorimaged using [F‐18]FLT at three stages of treatment using the same reconstruction methods. Difference images and standard deviations were used to assess the reconstruction differences. A three‐dimensional form of the Moran I(d) spatial statistic was used to assess global heterogeneity at various correlation distances. Results: Absolute difference images from FBP and 2 iteration OSEM reconstructions showed internal tumor voxel clusters deviating by more than 10% of the maximum SUV of the reference image (OSEM20) and relative voxel values varying by as much as 40% in tumor periphery. Image differences in OSEM reconstructions significantly decreased after 10 iterations, accompanied by decreases in the standard deviation of differences and slight increases in heterogeneity as global I(d) values decreased. FBP reconstructions both underestimated (Hanning, Shepp‐Logan) and overestimated (ramp) global heterogeneity I(d) relative to reference values, but large standard deviations of absolute difference indicated images compared poorly to the reference. Conclusion:Tumor heterogeneity obtained through PET may vary by at least 10% internally with larger variability at the periphery, greatly affecting both tumor volume delineation and internal heterogeneity. Prescriptions for dose painting based on proliferation measures can vary widely with the reconstruction algorithm.

Effects of collimator dependency and correction methods on I-123 SPECT images using X-ray-based attenuation map

Abstract: Low energy (LE) collimator is generally used in I-123 SPECT imaging. However, the septal penetration and scattering of photons emitted with energy above 159 keV will affect the image contrast and quantitative accuracy of images. To reduce this effect, medium energy (ME) collimator has been used with the cost of lower counting statistics and spatial resolution. The effects of collimator dependency on the quantitative accuracy of attenuation corrected (AC) I-123 SPECT images using X-ray-based attenuation map was investigated. Both brain and heart/thorax phantoms were used to evaluate different degree of attenuation effect between brain and thorax. Experiments were performed at different target-to-background ratios to simulate different object contrast. Both photopeak and scatter projections were collected for dual-energy window scatter correction (SC). Images were reconstructed and compared using different reconstruction methods, which included FBP (filtered backprojection), and OSEM (ordered subset expectation maximization) without corrections, with AC, with SC, and with AC and SC. In both phantom studies, the image contrast and quantitative accuracy were both improved with the use of CT-based transmission map for AC. The image contrast provided by ME collimator was better than that of LE collimator, especially in the region with void activity of thorax phantom. From the results of phantom studies, medium energy collimator with CT-based transmission map showed improvements in image contrast and quantitative accuracy of I-123 SPECT images.

Modified expectation maximization algorithm for transmission tomography

Abstract: A Bayesian reconstruction algorithm of transmission tomographic images is presented. The reconstruction is based on maximum aposteriori (MAP) estimation using the Expectation Maximization (EM) algorithm. The Gibbs prior with sigmoidal potential function that smoothes out high differences in neighboring pixel values while retaining sharp-edged features of images is employed. The Gibbs prior improves the image quality as well as the convergence properties of the EM algorithm. The Bayesian reconstruction problem itself is solved through the One-Step-Late Expectation-Maximization algorithm (EM-OSL) proposed by Green. Approximations are introduced to make the problem more tractable for parallel computation. Computer simulated phantoms are used in this study. Images reconstructed from this algorithm are compared to those reconstructed from currently available reconstruction algorithms in terms of image quality and convergence rate. The reconstruction algorithm works quite well for low-photon-count and low-contrast cases in which the widely used convolution backprojection algorithm performs rather poorly.