Software For Tomographic Image Reconstruction Research Articles

PARALLELPROJ—an open-source framework for fast calculation of projections in tomography

In this article, we introduce parallelproj, a novel open-source framework designed for efficient parallel computation of projections in tomography leveraging either multiple CPU cores or GPUs. This framework efficiently implements forward and back projection functions for both sinogram and listmode data, utilizing Joseph’s method, which is further extended to encompass time-of-flight (TOF) PET projections. Our evaluation involves a series of tests focusing on PET image reconstruction using data sourced from a state-of-the-art clinical PET/CT system. We thoroughly benchmark the performance of the projectors in non-TOF and TOF, sinogram, and listmode employing multi CPU-cores, hybrid CPU/GPU, and exclusive GPU mode. Moreover, we also investigate the timing of non-TOF sinogram projections calculated in STIR (Software for Tomographic Image Reconstruction) which recently integrated parallelproj as one of its projection backends. Our results indicate that the exclusive GPU mode provides acceleration factors between 25 and 68 relative to the multi-CPU-core mode. Furthermore, we demonstrate that OSEM listmode reconstruction of state-of-the-art real-world PET data sets is achievable within a few seconds using a single consumer GPU.

Investigation and optimization of PET-guided SPECT reconstructions for improved radionuclide therapy dosimetry estimates

AimsTo investigate and optimize the SPECTRE (Single Photon Emission Computed Theranostic REconstruction) reconstruction approach, using the hybrid kernelized expectation maximization (HKEM) algorithm implemented in the software for tomographic image reconstruction (STIR) software library, and to demonstrate the feasibility of performing algorithm exploration and optimization in 2D. Optimal SPECTRE parameters were investigated for the purpose of improving SPECT-based radionuclide therapy (RNT) dosimetry estimates.MethodsUsing the NEMA IEC body phantom as the test object, SPECT data were simulated to model an early and late imaging time point following a typical therapeutic dose of 8 GBq of 177Lu. A theranostic 68Ga PET-prior was simulated for the SPECTRE reconstructions. The HKEM algorithm parameter space was investigated for SPECT-unique and PET-SPECT mutual features to characterize optimal SPECTRE parameters for the simulated data. Mean and maximum bias, coefficient of variation (COV %), recovery, SNR and root-mean-square error (RMSE) were used to facilitate comparisons between SPECTRE reconstructions and OSEM reconstructions with resolution modelling (OSEM_RM). 2D reconstructions were compared to those performed in 3D in order to evaluate the utility of accelerated algorithm optimization in 2D. Segmentation accuracy was evaluated using a 42% fixed threshold (FT) on the 3D reconstructed data.ResultsSPECTRE parameters that demonstrated improved image quality and quantitative accuracy were determined through investigation of the HKEM algorithm parameter space. OSEM_RM and SPECTRE reconstructions performed in 2D & 3D were qualitatively and quantitatively similar, with SPECTRE showing an average reduction in background COV % by a factor of 2.7 and 3.3 for the 2D case and 3D case respectively. The 42% FT analysis produced an average % volume difference from ground truth of 158% and 26%, for the OSEM_RM and SPECTRE reconstructions, respectively.ConclusionsThe SPECTRE reconstruction approach demonstrates significant potential for improved SPECT image quality, leading to more accurate RNT dosimetry estimates when conventional segmentation methods are used. Exploration and optimization of SPECTRE benefited from both fast reconstruction times afforded by first considering the 2D case. This is the first in-depth exploration of the SPECTRE reconstruction approach, and as such, it reveals several insights for reconstructing SPECT data using PET side-information.

NEMA NU 1-2018 performance characterization and Monte Carlo model validation of the Cubresa Spark SiPM-based preclinical SPECT scanner

BackgroundThe Cubresa Spark is a novel benchtop silicon-photomultiplier (SiPM)-based preclinical SPECT system. SiPMs in SPECT significantly improve resolution and reduce detector size compared to preclinical cameras with photomultiplier tubes requiring highly magnifying collimators. The NEMA NU 1 Standard for Performance Measurements of Gamma Cameras provides methods that can be readily applied or extended to characterize preclinical cameras with minor modifications. The primary objective of this study is to characterize the Spark according to the NEMA NU 1-2018 standard to gain insight into its nuclear medicine imaging capabilities. The secondary objective is to validate a GATE Monte Carlo simulation model of the Spark for use in preclinical SPECT studies.MethodsNEMA NU 1-2018 guidelines were applied to characterize the Spark’s intrinsic, system, and tomographic performance with single- and multi-pinhole collimators. Phantoms were fabricated according to NEMA specifications with deviations involving high-resolution modifications. GATE was utilized to model the detector head with the single-pinhole collimator, and NEMA measurements were employed to tune and validate the model. Single-pinhole and multi-pinhole SPECT data were reconstructed with the Software for Tomographic Image Reconstruction and HiSPECT, respectively.ResultsThe limiting intrinsic resolution was measured as 0.85 mm owing to a high-resolution SiPM array combined with a 3 mm-thick scintillation crystal. The average limiting tomographic resolution was 1.37 mm and 1.19 mm for the single- and multi-pinhole collimators, respectively, which have magnification factors near unity at the center of rotation. The maximum observed count rate was 15,400 cps, and planar sensitivities of 34 cps/MBq and 150 cps/MBq were measured at the center of rotation for the single- and multi-pinhole collimators, respectively. All simulated tests agreed well with measurement, where the most considerable deviations were below 7%.ConclusionsNEMA NU 1-2018 standards determined that a SiPM detector mitigates the need for highly magnifying pinhole collimators while preserving detailed information in projection images. Measured and simulated NEMA results were highly comparable with differences on the order of a few percent, confirming simulation accuracy and validating the GATE model. Of the collimators initially provided with the Spark, the multi-pinhole collimator offers high resolution and sensitivity for organ-specific imaging of small animals, and the single-pinhole collimator enables high-resolution whole-body imaging of small animals.

Evaluation of STIR Library Adapted for PET Scanners with Non-Cylindrical Geometry

Software for Tomographic Image Reconstruction (STIR) is an open source C++ library used to reconstruct single photon emission tomography and positron emission tomography (PET) data. STIR has an experimental scanner geometry modelling feature to accurately model detector placement. In this study, we test and improve this new feature using several types of data: Monte Carlo simulations and measured phantom data acquired from a dedicated brain PET prototype scanner. The results show that the new geometry class applied to non-cylindrical PET scanners improved spatial resolution, uniformity, and image contrast. These are directly observed in the reconstructions of small features in the test quality phantom. Overall, we conclude that the revised “BlocksOnCylindrical” class will be a valuable addition to the next STIR software release with adjustments of existing features (Single Scatter Simulation, forward projection, attenuation corrections) to “BlocksOnCylindrical”.

Initial clinical experience with dedicated multi-pinhole (MPH) collimator for 99mTc-HMPAO brain perfusion SPECT.

Dedicated multi-pinhole (MPH) collimators have been successfully tested in selected clinical investigations. The aim of our work was to report initial experiences with an MPH collimator set designed for brain perfusion single photon emission tomography (SPECT). Ten patients underwent sequential technetium-99m-hexamethylpropyleneamineoxime (99mTc-HMPAO) SPECT with a dual-head SPECT camera equipped with conventional low-energy parallel hole collimators (LEHR), and with a triple-head system equipped with MPH collimators. Low-energy parallel hole collimators data were reconstructed by filtered back projection (FBP), ordered subset expectation maximization (OSEM), software for tomographic image reconstruction (STIR). In addition, both the parallel hole data and MPH data were reconstructed by Tera-TomoTM 3D iterative reconstruction denoted LEHR_TT3D and MPH_TT3D, respectively. Five medical experts visually compared the reconstructed images of the five data sets and defined a ranking sequence from the lowest (1) to the highest (5) image quality. Results were compared using the Friedman test. P values below 0.05 were considered significant. Low-energy parallel hole collimators acquisition resulted in 5 million, while MPH acquisition in 13 million total counts with 30 and 34 minutes of acquisition time, respectively. Mean rank coefficients of the reconstruction methods were 1.96±0.52, 2.66±0.46, 2.86±0.60, 3.62±0.55, 3.9±0.68 for FBP, STIR, LEHR_TT3D, LEHR_OSEM, MPH_TT3D respectively. The differences between MPH_TT3D-FBP (P<0.01); MPH_TT3D-STIR (P<0.05); LEHR_OSEM-FBP (P<0.01) were significant. Image quality provided by MPH collimator is comparable to that provided by conventional LEHR imaging. Higher sensitivity has the potential to shorten acquisition time or to reduce the amount of administered activity.

Evaluation of quantitative, efficient image reconstruction for VersaPET, a compact PET system.

Previously we developed a high-resolution positron emission tomography(PET) system-VersaPET-characterized by a block geometry with relatively large axial and transaxial interblock gaps and a compact geometry susceptible to parallax blurring effects. In this work, we report the qualitative and quantitative evaluation of a graphic processing unit (GPU)-accelerated maximum-likelihood by expectation-maximization (MLEM) image reconstruction framework for VersaPET which features accurate system geometry and projection space point-spread-function (PSF) modeling. We combined the ray-tracing module from software for tomographic image reconstruction (STIR),an open-source PET image reconstruction package, with VersaPET's exact block geometry for the geometric system matrix. Point-spread-function modeling of crystal penetration and scattering was achieved by a custom Monte-Carlo simulation for projection space blurring in all dimensions. We also parallelized the reconstruction in GPU taking advantage of the system's symmetry for PSF computation. To investigate the effects of PSF width, we generated and studied multiple kernels between one that reflects the true LYSO density in the MC simulation and another that reflects geometry only (no PSF). GATE simulations of hot and cold-sphere phantoms with spheres of different sizes, real microDerenzo phantom, and human blood vessel data were used to characterize the quantitative and qualitative performances of the reconstruction. Reconstruction with an accurate system geometry effectively improved image quality compared to STIR (version 3.0) which assumes an idealized system geometry. Reconstructions of GATE-simulated hot-sphere phantom data showed that all PSF kernels achieved superior performance in contrast recovery and bias reduction compared to using no PSF, but may introduce edge artifact and lumped background noise pattern depending on the width of PSF kernels. Cold-sphere phantom simulation results also indicated improvement in contrast recovery and quantification with PSF modeling (compared to no PSF) for 5 and 10mm cold spheres. Real microDerenzo phantom images with the PSF kernel that reflects the true LYSO density showed degraded resolving power of small sectors that could be resolved more clearly by underestimated PSF kernels, which is consistent with recent literature despite differences in scanner geometries and in approaches to system model estimation. The human vessel results resemble those of the hot-sphere phantom simulation with the PSF kernel that reflects the true LYSO density achieving the highest peak in the time activity curve (TAC) and similar lumped noise pattern. We fully evaluated a practical MLEM reconstruction framework that we developed for VersaPET in terms of qualitative and quantitative performance. Different PSF kernels may be adopted for improving the results of specific imaging tasks but the underlying reasons for the variation in optimal kernel for the real and simulation studies requires further study.

PET image reconstruction using physical and mathematical modelling for time of flight PET-MR scanners in the STIR library.

This work demonstrates how computational and physical modelling of the positron emission tomography (PET) image acquisition process for a state-of-the-art integrated PET and magnetic resonance imaging (PET-MR) system can produce images comparable to the manufacturer. The GE SIGNA PET/MR scanner is manufactured by General Electric and has time-of-flight (TOF) capabilities of about 390ps. All software development took place in the Software for Tomographic Image Reconstruction (STIR: http://stir.sf.net) library, which is a widely used open source software to reconstruct data as exported from emission tomography scanners. The new software developments will be integrated into STIR, providing the opportunity for researchers worldwide to establish and expand their image reconstruction methods. Furthermore, this work is of particular significance as it provides the first validation of TOF PET image reconstruction for real scanner datasets using the STIR library. This paper presents the methodology, analysis, and critical issues encountered in implementing an independent reconstruction software package. Acquired PET data were processed via several appropriate algorithms which are necessary to produce an accurate and precise quantitative image. This included mathematical, physical and anatomical modelling of the patient and simulation of various aspects of the acquisition. These included modelling of random coincidences using 'singles' rates per crystals, detector efficiencies and geometric effects. Attenuation effects were calculated by using the STIR's attenuation correction model. Modelling all these effects within the system matrix allowed the reconstruction of PET images which demonstrates the metabolic uptake of the administered radiopharmaceutical. These implementations were validated using measured phantom and clinical datasets. The developments are tested using the ordered subset expectation maximisation (OSEM) and the more recently proposed kernelised expectation maximisation (KEM) algorithm which incorporates anatomical information from MR images into PET reconstruction.

SIRF: Synergistic Image Reconstruction Framework

The combination of positron emission tomography (PET) with magnetic resonance (MR) imaging opens the way to more accurate diagnosis and improved patient management. At present, the data acquired by PET-MR scanners are essentially processed separately, but the opportunity to improve accuracy of the tomographic reconstruction via synergy of the two imaging techniques is an active area of research.In this paper, we present Release 2.1.0 of the CCP-PETMR Synergistic Image Reconstruction Framework (SIRF) software suite, providing an open-source software platform for efficient implementation and validation of novel reconstruction algorithms. SIRF provides user-friendly Python and MATLAB interfaces built on top of C++ libraries. SIRF uses advanced PET and MR reconstruction software packages and tools. Currently, for PET this is Software for Tomographic Image Reconstruction (STIR); for MR, Gadgetron and ISMRMRD; and for image registration tools, NiftyReg. The software aims to be capable of reconstructing images from acquired scanner data, whilst being simple enough to be used for educational purposes.The most recent version of the software can be downloaded from http://www.ccppetmr.ac.uk/downloads and https://github.com/CCPPETMR/. Program summaryProgram Title: Synergistic Image Reconstruction Framework (SIRF)Program Files DOI:http://dx.doi.org/10.17632/s45f5jh55j.1Licensing provisions: GPLv3 and Apache-2.0Programming languages: C++, C, Python, MATLABNature of problem: In current practice, data acquired by PET-MR scanners are processed separately. Methods for improving the accuracy of the tomographic reconstruction using the synergy of the two imaging techniques are actively being investigated by the PET-MR research and development community, however, practical application is heavily reliant on software. Open-source software available to the PET-MR community – such as the PET package (STIR) (Thielemans et al., 2012) and the MR package Gadgetron (Hansen and Sørensen, 2013) – provide a basis for new synergistic PET-MR software. However, these two software packages are independent and have very different software architectures. They are mostly written in C++ but many researchers in the PET-MR community are more familiar with script-style languages, such as Python and MATLAB, which enable rapid prototyping of novel reconstruction algorithms. In the current situation it is difficult for researchers to exploit any synergy between PET and MR data. Furthermore, techniques from one field cannot easily be applied in the other.Solution method: In SIRF, the bulk of computation is performed by available advanced open-source reconstruction and registration software (currently STIR, Gadgetron and NiftyReg) that can use multithreading and GPUs. The SIRF C++ code provides a thin layer on top of these existing libraries. The SIRF layer has unified data-containers and access mechanisms. This C++ layer provides the basis for a simple and intuitive Python and MATLAB interface, enabling users to quickly develop and test their reconstruction algorithms using these scripting languages only. At the same time, advanced users proficient in C++ can directly utilise wider SIRF functionality via the SIRF C++ libraries that we provide.

Implementation and validation of time-of-flight PET image reconstruction module for listmode and sinogram projection data in the STIR library

In this paper, we describe the implementation of support for time-of-flight (TOF) positron emission tomography (PET) for both listmode and sinogram data in the open source software for tomographic image reconstruction (STIR). We provide validation and performance characterization using simulated data from the open source GATE Monte Carlo toolbox, with TOF configurations spanning from 81.2 to 209.6 ps. The coincidence detector resolution was corrected for the timing resolution deterioration due to the contribution of the crystal length. Comparison between the reconstruction of listmode and sinogram data demonstrated good agreement in both TOF and non-TOF cases in terms of relative absolute error. To reduce the reconstruction time, we assessed the truncation of the TOF kernel along lines-of-response (LOR). Rejection of LOR elements beyond four times the TOF standard deviation provides significant acceleration of without compromising the image quality. Further narrowing of the kernel can provide extra time reduction but with the gradual introduction of error in the reconstructed images. As expected, TOF reconstruction performs better than non-TOF in terms of both contrast-recovery-coefficient (CRC) and signal-to-noise ratio (SNR). CRC achieves convergence faster with TOF, at lower noise levels. SNR with TOF was superior for early iterations, but with quick deterioration. Higher timing resolution further improved reconstruction performance, while TOF bin mashing was shown to have only a small impact on reconstructed images.

Information Content in Nuclear Medicine Imaging

The aim of the present study was to investigate the information content of positron emission tomography (PET) images. We used the GATE Monte Carlo package (GEANT4 application for tomographic emission) and reconstructed images, obtained using the software for tomographic image reconstruction (STIR). The case study for the investigation of the PET images information content was the General Electric Discovery-ST (USA) PET scanner. A thin film plane source aluminum (Al) foil, coated with a thin layer of silica and a fluorodeoxyglucose (18F-FDG) bath distribution of (1 MBq) was used in the simulation for the image signal to noise ratio assessment. The influence of the maximum likelihood estimation ordered subsets maximum a posteriori one step late (MLE)-OS-MAP-OSL algorithm, using various subsets (1 to 21) and iterations (1 to 20) was examined. The image information content was assessed in terms of the information capacity (IC). Results showed that the single index information capacity maximized for the range of 8-20 iterations and 3 subsets. In conclusion, our study showed that the image information content of PET scanners can be fully characterized and further improved by investigation of the imaging chain components through Monte Carlo methods. Moreover, new perspectives are created by using the suggested techniques in the context of a global cloud service that could serve as an online quality evaluation metric for the PET scanners and other medical imaging systems.

Information Capacity of Positron Emission Tomography Scanners

Background: The aim of the present study was to assess the upper information content bound of positron emission tomography (PET) images, by means of the information capacity (IC). Methods: The Geant4 Application for the Tomographic Emission (GATE) Monte Carlo (MC) package was used, and reconstructed images were obtained by using the software for tomographic image reconstruction (STIR). The case study for the assessment of the information content was the General Electric (GE) Discovery-ST PET scanner. A thin-film plane source aluminum (Al) foil, coated with a thin layer of silica and with a 18F-fludeoxyglucose (FDG) bath distribution of 1 MBq was used. The influence of the (a) maximum likelihood estimation-ordered subsets-maximum a posteriori probability-one step late (MLE-OS-MAP-OSL) algorithm, using various subsets (1 to 21) and iterations (1 to 20) and (b) different scintillating crystals on PET scanner’s performance, was examined. The study was focused on the noise equivalent quanta (NEQ) and on the single index IC. Images of configurations by using different crystals were obtained after the commonly used 2-dimensional filtered back projection (FBP2D), 3-dimensional filtered back projection re-projection (FPB3DRP) and the (MLE)-OS-MAP-OSL algorithms. Results: Results shown that the images obtained with one subset and various iterations provided maximum NEQ values, however with a steep drop-off after 0.045 cycles/mm. The single index IC data were maximized for the range of 8–20 iterations and three subsets. The PET scanner configuration incorporating lutetium orthoaluminate perovskite (LuAP) crystals provided the highest NEQ values in 2D FBP for spatial frequencies higher than 0.028 cycles/mm. Bismuth germanium oxide (BGO) shows clear dominance against all other examined crystals across the spatial frequency range, in both 3D FBP and OS-MAP-OSL. The particular PET scanner provided optimum IC values using FBP3DRP and BGO crystals (2.4829 bits/mm2). Conclusions: The upper bound of the image information content of PET scanners can be fully characterized and further improved by investigating the imaging chain components through MC methods.

Respiratory motion correction in prostate cancer positron emission tomography: A study on patients and phantom simulation

Introduction: To investigate the effects of breathing cycle and tree diaphragm motions on prostate cancer tumors standard uptake value (SUV) during positron emission tomography (PET) and to correct it. Materials and methods: Respiratory motion traces were simulated on the common patient breathing cycle and tree diaphragm motions using the 4D NCAT phantom. Lesions with different activities and diameters were simulated in prostate. This study was conducted in two phases. In the first phase; the generated PET sinograms taken from GATE (Geant4 Application Tomography Emission) Monte Carlo simulations corrected using computed tomography attenuation correction involving the cycle. In the next phase; on the obtained images, the motion-correction was carried out using software for Tomographic Image Reconstruction (STIR). By considering respiration- averaged computed as a true value, the SUVs were measured and analyzed for different motions. Results: In uncorrected measurements, the mean, minimum and maximum SUVs were 0.4%, 0.09% and 2% respectively, whilst in corrected measurements, these value were reported as 3.5%, 0.7% and 14% respectively. Conclusion: The respiratory motion-induced errors in tumor quantification and delineation. The SUVs depending on the phase of the raspatory. Motion correction could enhance the SUVs in prostate cancer PET.

3D tomographic imaging with the γ-eye planar scintigraphic gamma camera

γ-eye is a desktop planar scintigraphic gamma camera (100 mm × 50 mm field of view) designed by BET Solutions as an affordable tool for dynamic, whole body, small-animal imaging. This investigation tests the viability of using γ-eye for the collection of tomographic data for 3D SPECT reconstruction. Two software packages, QSPECT and STIR (software for tomographic image reconstruction), have been compared. Reconstructions have been performed using QSPECT’s implementation of the OSEM algorithm and STIR’s OSMAPOSL (Ordered Subset Maximum A Posteriori One Step Late) and OSSPS (Ordered Subsets Separable Paraboloidal Surrogate) algorithms. Reconstructed images of phantom and mouse data have been assessed in terms of spatial resolution, sensitivity to varying activity levels and uniformity. The effect of varying the number of iterations, the voxel size (1.25 mm default voxel size reduced to 0.625 mm and 0.3125 mm), the point spread function correction and the weight of prior terms were explored. While QSPECT demonstrated faster reconstructions, STIR outperformed it in terms of resolution (as low as 1 mm versus 3 mm), particularly when smaller voxel sizes were used, and in terms of uniformity, particularly when prior terms were used. Little difference in terms of sensitivity was seen throughout.

Detective quantum efficiency (DQE) in PET scanners: A simulation study

The aim of the present study is to introduce the detective quantum efficiency (DQE) for the image quality assessment of positron emission tomography (PET) scanners. For this purpose, a thin layer chromatography (TLC) plane source was simulated using a previously validated, scanner and source geometry, Monte Carlo (MC) model. The model was developed with the Geant4 application for tomographic emission (GATE) MC package and reconstructed images obtained with the software for tomographic image reconstruction (STIR), with cluster computing. The GE Discovery ST PET scanner was simulated by using a previously validated code. A plane source consisting of a TLC plate, was simulated by a layer of silica gel on aluminum (Al) foil substrate, immersed in 18F-FDG bath solution (1MBq). Image quality was assessed in terms of the modulation transfer function (MTF) and the normalized noise power spectrum (NNPS) in order to obtain the detective quantum efficiency (DQE). MTF curves were estimated from transverse reconstructed images of the plane source, whereas the NNPS data were estimated from the corresponding coronal images. Images were reconstructed by the maximum likelihood estimation ordered subsets maximum a posteriori one step late (MLE)-OS-MAP-OSL algorithm, by using various subsets 1–21) and iterations 1–20). MTF values were found to increase up to the 12th iteration whereas remain almost constant thereafter. However, the range of the increase in the MTF is limited as the number of subsets increases. The noise levels were found to increase with the corresponding increase of both the number of iterations and subsets. The maximum NNPS value (0.517mm2) was observed for the 420 MLEM-equivalent iterations reconstructed image at 0cycles/mm. Finally DQE values were found to increase for spatial frequencies up to 0.038cycles/mm and to decrease thereafter with the corresponding increase in both number of iterations and subsets. The maximum DQE value (0.48 at 0.038cycles/mm) was obtained for the 8 MLEM-equivalent iterations image. The simulated PET evaluation method based on the TLC plane source can be useful in the quality control and in the further development of PET and SPECT scanners though GATE simulations.

Quantitative PET image reconstruction employing nested expectation-maximization deconvolution for motion compensation

Bulk body motion may randomly occur during PET acquisitions introducing blurring, attenuation-emission mismatches and, in dynamic PET, discontinuities in the measured time activity curves between consecutive frames. Meanwhile, dynamic PET scans are longer, thus increasing the probability of bulk motion. In this study, we propose a streamlined 3D PET motion-compensated image reconstruction (3D-MCIR) framework, capable of robustly deconvolving intra-frame motion from a static or dynamic 3D sinogram. The presented 3D-MCIR methods need not partition the data into multiple gates, such as 4D MCIR algorithms, or access list-mode (LM) data, such as LM MCIR methods, both associated with increased computation or memory resources. The proposed algorithms can support compensation for any periodic and non-periodic motion, such as cardio-respiratory or bulk motion, the latter including rolling, twisting or drifting. Inspired from the widely adopted point-spread function (PSF) deconvolution 3D PET reconstruction techniques, here we introduce an image-based 3D generalized motion deconvolution method within the standard 3D maximum-likelihood expectation-maximization (ML-EM) reconstruction framework. In particular, we initially integrate a motion blurring kernel, accounting for every tracked motion within a frame, as an additional MLEM modeling component in the image space (integrated 3D-MCIR). Subsequently, we replaced the integrated model component with a nested iterative Richardson-Lucy (RL) image-based deconvolution method to accelerate the MLEM algorithm convergence rate (RL-3D-MCIR). The final method was evaluated with realistic simulations of whole-body dynamic PET data employing the XCAT phantom and real human bulk motion profiles, the latter estimated from volunteer dynamic MRI scans. In addition, metabolic uptake rate Ki parametric images were generated with the standard Patlak method. Our results demonstrate significant improvement in contrast-to-noise ratio (CNR) and noise-bias performance in both dynamic and parametric images. The proposed nested RL-3D-MCIR method is implemented on the Software for Tomographic Image Reconstruction (STIR) open-source platform and is scheduled for public release.

Impact of muscular uptake and statistical noise on tumor quantification based on simulated FDG-PET studies

PurposeThe aim of this work is to study the effect of physiological muscular uptake variations and statistical noise on tumor quantification in FDG-PET studies. MethodsWe designed a realistic framework based on simulated FDG-PET acquisitions from an anthropomorphic phantom that included different muscular uptake levels and three spherical lung lesions with diameters of 31, 21 and 9mm. A distribution of muscular uptake levels was obtained from 136 patients remitted to our center for whole-body FDG-PET. Simulated FDG-PET acquisitions were obtained by using the Simulation System for Emission Tomography package (SimSET) Monte Carlo package. Simulated data was reconstructed by using an iterative Ordered Subset Expectation Maximization (OSEM) algorithm implemented in the Software for Tomographic Image Reconstruction (STIR) library. Tumor quantification was carried out by using estimations of SUVmax, SUV50 and SUVmean from different noise realizations, lung lesions and multiple muscular uptakes. ResultsOur analysis provided quantification variability values of 17–22% (SUVmax), 11–19% (SUV50) and 8–10% (SUVmean) when muscular uptake variations and statistical noise were included. Meanwhile, quantification variability due only to statistical noise was 7–8% (SUVmax), 3–7% (SUV50) and 1–2% (SUVmean) for large tumors (>20mm) and 13% (SUVmax), 16% (SUV50) and 8% (SUVmean) for small tumors (<10mm), thus showing that the variability in tumor quantification is mainly affected by muscular uptake variations when large enough tumors are considered. In addition, our results showed that quantification variability is strongly dominated by statistical noise when the injected dose decreases below 222MBq. ConclusionsOur study revealed that muscular uptake variations between patients who are totally relaxed should be considered as an uncertainty source of tumor quantification values.

PET iterative reconstruction incorporating an efficient positron range correction method

Positron range is one of the main physical effects limiting the spatial resolution of positron emission tomography (PET) images. If positrons travel inside a magnetic field, for instance inside a nuclear magnetic resonance (MR) tomograph, the mean range will be smaller but still significant. In this investigation we examined a method to correct for the positron range effect in iterative image reconstruction by including tissue-specific kernels in the forward projection operation. The correction method was implemented within STIR library (Software for Tomographic Image Reconstruction). In order to obtain the positron annihilation distribution of various radioactive isotopes in water and lung tissue, simulations were performed with the Monte Carlo package GATE [Jan et al. 2004 [1]] simulating different magnetic field intensities (0 T, 3 T, 9.5 T and 11 T) along the axial scanner direction. The positron range kernels were obtained for 68Ga in water and lung tissue for 0 T and 3 T magnetic field voxellizing the annihilation coordinates into a three-dimensional matrix. The proposed method was evaluated using simulations of material-variant and material-invariant positron range corrections for the HYPERImage preclinical PET-MR scanner. The use of the correction resulted in sharper active region boundary definition, albeit with noise enhancement, and in the recovery of the true activity mean value of the hot regions. Moreover, in the case where a magnetic field is present, the correction accounts for the non-isotropy of the positron range effect, resulting in the recovery of resolution along the axial plane.

The SRT reconstruction algorithm for semiquantification in PET imaging.

The spline reconstruction technique (SRT) is a new, fast algorithm based on a novel numerical implementation of an analytic representation of the inverse Radon transform. The mathematical details of this algorithm and comparisons with filtered backprojection were presented earlier in the literature. In this study, the authors present a comparison between SRT and the ordered-subsets expectation-maximization (OSEM) algorithm for determining contrast and semiquantitative indices of (18)F-FDG uptake. The authors implemented SRT in the software for tomographic image reconstruction (stir) open-source platform and evaluated this technique using simulated and real sinograms obtained from the GE Discovery ST positron emission tomography/computer tomography scanner. All simulations and reconstructions were performed in stir. For OSEM, the authors used the clinical protocol of their scanner, namely, 21 subsets and two iterations. The authors also examined images at one, four, six, and ten iterations. For the simulation studies, the authors analyzed an image-quality phantom with cold and hot lesions. Two different versions of the phantom were employed at two different hot-sphere lesion-to-background ratios (LBRs), namely, 2:1 and 4:1. For each noiseless sinogram, 20 Poisson realizations were created at five different noise levels. In addition to making visual comparisons of the reconstructed images, the authors determined contrast and bias as a function of the background image roughness (IR). For the real-data studies, sinograms of an image-quality phantom simulating the human torso were employed. The authors determined contrast and LBR as a function of the background IR. Finally, the authors present plots of contrast as a function of IR after smoothing each reconstructed image with Gaussian filters of six different sizes. Statistical significance was determined by employing the Wilcoxon rank-sum test. In both simulated and real studies, SRT exhibits higher contrast and lower bias than OSEM at the cold lesions. This improvement is achieved at the expense of increasing the noise in the reconstructed images. For the hot lesions, SRT exhibits a small improvement in contrast and LBR over OSEM with 21 subsets and two iterations; however, this improvement is not statistically significant. As the number of iterations increases, the performance of OSEM improves over SRT but again without statistical significance. The curves of contrast and LBR as a function of IR after Gaussian blurring indicate that the advantage of SRT in the cold regions is maintained even after decreasing the noise level by Gaussian blurring. SRT, at the expense of slightly increased noise in the reconstructed images, reconstructs images of higher contrast and lower bias than the clinical protocol of OSEM. This improvement is particularly evident for images involving cold regions. Thus, it appears that SRT should be particularly useful for the quantification of low-count and cold regions.

Influence of Iterative Reconstruction Algorithms on PET Image Resolution

The aim of the present study was to assess image quality of PET scanners through a thin layer chromatography (TLC) plane source. The source was simulated using a previously validated Monte Carlo model. The model was developed by using the GATE MC package and reconstructed images obtained with the STIR software for tomographic image reconstruction. The simulated PET scanner was the GE DiscoveryST. A plane source consisted of a TLC plate, was simulated by a layer of silica gel on aluminum (Al) foil substrates, immersed in 18F-FDG bath solution (1MBq). Image quality was assessed in terms of the modulation transfer function (MTF). MTF curves were estimated from transverse reconstructed images of the plane source. Images were reconstructed by the maximum likelihood estimation (MLE)-OSMAPOSL, the ordered subsets separable paraboloidal surrogate (OSSPS), the median root prior (MRP) and OSMAPOSL with quadratic prior, algorithms. OSMAPOSL reconstruction was assessed by using fixed subsets and various iterations, as well as by using various beta (hyper) parameter values. MTF values were found to increase with increasing iterations. MTF also improves by using lower beta values. The simulated PET evaluation method, based on the TLC plane source, can be useful in the resolution assessment of PET scanners.

Assessment of the Contrast to Noise Ratio in PET Scanners with Monte Carlo Methods

The aim of the present study was to assess the contrast to noise ratio (CNR) of PET scanners through a thin layer chromatography (TLC) plane source. The source was simulated using a previously validated Monte Carlo model. The model was developed by using the GATE MC package and reconstructed images obtained with the STIR software for tomographic image reconstruction. The PET scanner simulated was the GE DiscoveryST. A plane source consisted of a TLC plate, was simulated by a layer of silica gel on aluminum (Al) foil substrates, immersed in 18F-FDG bath solution. Image quality was assessed in terms of the CNR. CNR was estimated from coronal reconstructed images of the plane source. Images were reconstructed by the maximum likelihood estimation (MLE)-OSMAPOSL. OSMAPOSL reconstruction was assessed by using various subsets (3, 15 and 21) and various iterations (2 to 20). CNR values were found to decrease when both iterations and subsets increase. Two (2) iterations were found to be optimal. The simulated PET evaluation method, based on the TLC plane source, can be useful in image quality assessment of PET scanners.