Table of contents

The development of radiation detectors capable of delivering spatial information about gamma-ray interactions was one of the key enabling technologies for nuclear medicine imaging and, eventually, single-photon emission computed tomography (SPECT). The continuous sodium iodide scintillator crystal coupled to an array of photomultiplier tubes, almost universally referred to as the Anger Camera after its inventor, has long been the dominant SPECT detector system. Nevertheless, many alternative materials and configurations have been investigated over the years. Technological advances as well as the emerging importance of specialized applications, such as cardiac and preclinical imaging, have spurred innovation such that alternatives to the Anger Camera are now part of commercial imaging systems. Increased computing power has made it practical to apply advanced signal processing and estimation schemes to make better use of the information contained in the detector signals. In this review we discuss the key performance properties of SPECT detectors and survey developments in both scintillator and semiconductor detectors and their readouts with an eye toward some of the practical issues at least in part responsible for the continuing prevalence of the Anger Camera in the clinic.

The purpose of this study is to assess the accuracy of day-to-day predictions of liver tumour position using implanted gold markers as surrogates and to compare the method with alternative set-up strategies, i.e. no correction, vertebrae and 3D diaphragm-based set-up. Twenty patients undergoing stereotactic body radiation therapy (SBRT) with abdominal compression for primary or metastatic liver cancer were analysed. We determined the day-to-day correlation between gold marker and tumour positions in contrast-enhanced CT scans acquired at treatment preparation and before each treatment session. The influence of marker–tumour distance on the accuracy of prediction was estimated by introducing a method extension of the set-up error paradigm. The distance between gold markers and the centre of the tumour varied between 5 and 96 mm. Marker-guidance was superior to guiding treatment using other surrogates, although both the random and systematic components of the prediction error SD depended on the tumour–marker distance. For a marker–tumour distance of 4 cm, we observed σ = 1.3 mm and Σ = 1.6 mm. The 3D position of the diaphragm dome was the second best predictor. In conclusion, the tumour position can be predicted accurately using implanted markers, but marker-guided set-up accuracy decreases with increasing distance between implanted markers and the tumour.

Recent acute shortage of medical radioisotopes prompted investigations into alternative methods of production and the use of a cyclotron and ¹⁰⁰Mo(p,2n)^99mTc reaction has been considered. In this context, the production yields of ^99mTc and various other radioactive and stable isotopes which will be created in the process have to be investigated, as these may affect the diagnostic outcome and radiation dosimetry in human studies. Reaction conditions (beam and target characteristics, and irradiation and cooling times) need to be optimized in order to maximize the amount of ^99mTc and minimize impurities. Although ultimately careful experimental verification of these conditions must be performed, theoretical calculations can provide the initial guidance allowing for extensive investigations at little cost. We report the results of theoretically determined reaction yields for ^99mTc and other radioactive isotopes created when natural and enriched molybdenum targets are irradiated by protons. The cross-section calculations were performed using a computer program EMPIRE for the proton energy range 6–30 MeV. A computer graphical user interface for automatic calculation of production yields taking into account various reaction channels leading to the same final product has been created. The proposed approach allows us to theoretically estimate the amount of ^99mTc and its ratio relative to ^99gTc and other radioisotopes which must be considered reaction contaminants, potentially contributing to additional patient dose in diagnostic studies.

An algorithm capable of mitigating respiratory motion blurring artifacts in cone-beam computed tomography (CBCT) lung tumor images based on the motion of the tumor during the CBCT scan is developed. The tumor motion trajectory and probability density function (PDF) are reconstructed from the acquired CBCT projection images using a recently developed algorithm Lewis et al (2010 Phys. Med. Biol.55 2505–22). Assuming that the effects of motion blurring can be represented by convolution of the static lung (or tumor) anatomy with the motion PDF, a cost function is defined, consisting of a data fidelity term and a total variation regularization term. Deconvolution is performed through iterative minimization of this cost function. The algorithm was tested on digital respiratory phantom, physical respiratory phantom and patient data. A clear qualitative improvement is evident in the deblurred images as compared to the motion-blurred images for all cases. Line profiles show that the tumor boundaries are more accurately and clearly represented in the deblurred images. The normalized root-mean-squared error between the images used as ground truth and the motion-blurred images are 0.29, 0.12 and 0.30 in the digital phantom, physical phantom and patient data, respectively. Deblurring reduces the corresponding values to 0.13, 0.07 and 0.19. Application of a −700 HU threshold to the digital phantom results in tumor dimension measurements along the superior–inferior axis of 2.8, 1.8 and 1.9 cm in the motion-blurred, ground truth and deblurred images, respectively. Corresponding values for the physical phantom are 3.4, 2.7 and 2.7 cm. A threshold of −500 HU applied to the patient case gives measurements of 3.1, 1.6 and 1.7 cm along the SI axis in the CBCT, 4DCT and deblurred images, respectively. This technique could provide more accurate information about a lung tumor's size and shape on the day of treatment.

Optimizing targeted radionuclide therapy requires patient-specific estimation of organ doses. The organ doses are estimated from quantitative nuclear medicine imaging studies, many of which involve planar whole body scans. We have previously developed the quantitative planar (QPlanar) processing method and demonstrated its ability to provide more accurate activity estimates than conventional geometric-mean-based planar (CPlanar) processing methods using physical phantom and simulation studies. The QPlanar method uses the maximum likelihood-expectation maximization algorithm, 3D organ volume of interests (VOIs), and rigorous models of physical image degrading factors to estimate organ activities. However, the QPlanar method requires alignment between the 3D organ VOIs and the 2D planar projections and assumes uniform activity distribution in each VOI. This makes application to patients challenging. As a result, in this paper we propose an extended QPlanar (EQPlanar) method that provides independent-organ rigid registration and includes multiple background regions. We have validated this method using both Monte Carlo simulation and patient data. In the simulation study, we evaluated the precision and accuracy of the method in comparison to the original QPlanar method. For the patient studies, we compared organ activity estimates at 24 h after injection with those from conventional geometric mean-based planar quantification using a 24 h post-injection quantitative SPECT reconstruction as the gold standard. We also compared the goodness of fit of the measured and estimated projections obtained from the EQPlanar method to those from the original method at four other time points where gold standard data were not available. In the simulation study, more accurate activity estimates were provided by the EQPlanar method for all the organs at all the time points compared with the QPlanar method. Based on the patient data, we concluded that the EQPlanar method provided a substantial increase in accuracy of organ activity estimates from 24 h planar images compared to the CPlanar using 24 h SPECT as the golden standard. For other time points, where no golden standard is available, better agreement between estimated and measured projections was observed by using the EQPlanar method compared to the QPlanar method. This phenomenon is consistent with the improvement in goodness of fit seen in both simulation data and 24 h patient data. Therefore, this indicates the improved reliability of organ activity estimates obtained though the EQPlanar method.

We aimed to estimate the scattered radiation from dental metallic crowns during head and neck radiotherapy by irradiating a jaw phantom with external photon beams. The phantom was composed of a dental metallic plate and hydroxyapatite embedded in polymethyl methacrylate. We used radiochromic film measurement and Monte Carlo simulation to calculate the radiation dose and dose distribution inside the phantom. To estimate dose variations in scattered radiation under different clinical situations, we altered the incident energy, field size, plate thickness, plate depth and plate material. The simulation results indicated that the dose at the incident side of the metallic dental plate was approximately 140% of that without the plate. The differences between dose distributions calculated with the radiation treatment-planning system (TPS) algorithms and the data simulation, except around the dental metallic plate, were 3% for a 4 MV photon beam. Therefore, we should carefully consider the dose distribution around dental metallic crowns determined by a TPS.

Statistical iterative reconstruction (SIR) algorithms have shown potential to substantially improve low-dose cone-beam CT (CBCT) image quality. The penalty term plays an important role in determining the performance of SIR algorithms. In this work, we quantitatively evaluate the impact of the penalties on the performance of a statistics-based penalized weighted least-squares (PWLS) iterative reconstruction algorithm for improving the image quality of low-dose CBCT. Three different edge-preserving penalty terms, exponential form anisotropic quadratic (AQ) penalty (PWLS-Exp), inverse square form AQ penalty (PWLS-InverseSqr) and total variation penalty (PWLS-TV), were compared against the conventional isotropic quadratic form penalty (PWLS-Iso) using both computer simulation and experimental studies. Noise in low-dose CBCT can be substantially suppressed by the PWLS reconstruction algorithm and edges are well preserved by both AQ- and TV-based penalty terms. The noise-resolution tradeoff measurement shows that the PWLS-Exp exhibits the best spatial resolution of all the three anisotropic penalty terms at matched noise level for reconstructing high-contrast objects. For the reconstruction of low-contrast objects, the TV-based penalty outperforms the AQ-based one with better resolution preservation at matched noise levels. Different penalty terms may be used for better edge preservation at different targeted contrast levels.

The efficacy of high-intensity focused ultrasound (HIFU) for the treatment of a range of different cancers, including those of the liver, prostate and breast, has been demonstrated. As a non-invasive focused therapy, HIFU offers considerable advantages over techniques such as chemotherapy and surgical resection in terms of reduced risk of harmful side effects. Despite this, there are a number of significant challenges which currently hinder its widespread clinical application. One of these challenges is the need to transmit sufficient energy through the rib cage to induce tissue necrosis in the required volume whilst minimizing the formation of side lobes. Multi-element random-phased arrays are currently showing great promise in overcoming the limitations of single-element transducers. Nevertheless, successful treatment of a patient with liver tumours requires a thorough understanding of the way in which the ultrasonic pressure field from a HIFU array is scattered by the rib cage. In order to address this, a boundary element approach based on a generalized minimal residual (GMRES) implementation of the Burton–Miller formulation was used in conjunction with phase conjugation techniques to focus the field of a 256-element random HIFU array behind human ribs at locations requiring intercostal and transcostal treatment. Simulations were carried out on a 3D mesh of quadratic pressure patches generated using CT scan anatomical data for adult ribs 9–12 on the right side. The methodology was validated on spherical and cylindrical scatterers. Field calculations were also carried out for idealized ribs, consisting of arrays of strip-like scatterers, demonstrating effects of splitting at the focus. This method has the advantage of fully accounting for the effect of scattering and diffraction in 3D under continuous wave excitation.

To meet the growing demand for functional imaging technology for use in studying plant biology, we are developing a novel technique that permits simultaneous imaging of escaped positrons and coincidence gammas from annihilation of positrons within an intake leaf. The multi-modality imaging system will include two planar detectors: one is a typical PET detector array and the other is a phoswich imaging detector that detects both beta and gamma. The novel phoswich detector is made of a plastic scintillator, a lutetium oxyorthosilicate (LSO) array, and a position sensitive photomultiplier tube (PS-PMT). The plastic scintillator serves as a beta detector, while the LSO array serves as a gamma detector and light guide that couples scintillation light from the plastic detector to the PMT. In our prototype, the PMT signal was fed into the Siemens QuickSilver electronics to achieve shaping and waveform sampling. Pulse-shape discrimination based on the detectors' decay times (2.1 ns for plastic and 40 ns for LSO) was used to differentiate beta and gamma events using the common PMT signals. Using our prototype phoswich detector, we simultaneously measured a beta image and gamma events (in single mode). The beta image showed a resolution of 1.6 mm full-width-at-half-maximum using F-18 line sources. Because this shows promise for plant-scale imaging, our future plans include development of a fully functional simultaneous beta-and-coincidence-gamma imager with sub-millimeter resolution imaging capability for both modalities.

Dual energy x-ray analysis (DEXA) is investigated using a nonlinear model for the x-ray linear attenuation coefficient μ that is expressed as a function of electron density N_e and the fourth compositional ratio R₄. Nonlinear simultaneous equations are solved using a least-squares algorithm based upon the method of Levenberg and Marquardt. Measurements of μ for low atomic number materials (containing elements hydrogen to calcium) at energies 32–66 keV are used to study DEXA accuracy as a function of sample composition, photon energy and their separation ΔE. Results are presented for ΔE = 5–30 keV, for 2% measurement precision, and the doses involved are quantified. The model is subject to propagation of error analysis and results are presented for the relationship between measurement uncertainties and those for N_e and R₄. The analysis shows how DEXA accuracy is controlled by the fractional compositional cross-product, which represents the contribution of composition to μ, and how this can be optimized by careful selection of beam energies according to the compositional range of interest. Accurate DEXA is achieved over restricted energy and compositional ranges: soft tissues only at approximately 15–25 keV, all tissues at approximately 30–80 keV and, for situations where a higher dose can be tolerated, all tissues at approximately 4–8 MeV.

Magnetoencephalography (MEG) source analysis has largely relied on spherical conductor models of the head to simplify forward calculations of the brain's magnetic field. Multiple- (or overlapping, local) sphere models, where an optimal sphere is selected for each sensor, are considered an improvement over single-sphere models and are computationally simpler than realistic models. However, there is limited information available regarding the different methods used to generate these models and their relative accuracy. We describe a variety of single- and multiple-sphere fitting approaches, including a novel method that attempts to minimize the field error. An accurate boundary element method simulation was used to evaluate the relative field measurement error (12% on average) and dipole fit localization bias (3.5 mm) of each model over the entire brain. All spherical models can contribute in the order of 1 cm to the localization bias in regions of the head that depart significantly from a sphere (inferior frontal and temporal). These spherical approximation errors can give rise to larger localization differences when all modeling effects are taken into account and with more complex source configurations or other inverse techniques, as shown with a beamformer example. Results differed noticeably depending on the source location, making it difficult to recommend a fitting method that performs best in general. Given these limitations, it may be advisable to expand the use of realistic head models.

LNE-LNHB is involved in a European project aiming at establishing absorbed dose-to-water standards for photon-radiation fields down to 2 × 2 cm². This requires the calibration of reference ionization chambers of small volume. Twenty-four ionization chambers of eight different types with volume ranging from 0.007 to 0.057 cm³ were tested in a ⁶⁰Co beam. For each chamber, two major characteristics were investigated: (1) the stability of the measured current as a function of the irradiation time under continuous irradiation. At LNE-LNHB, the variation of the current should be less than ±0.1% in comparison with its first value (over a 16 h irradiation time); (2) the variation of the ionization current with the applied polarizing voltage and polarity. Leakage currents were also measured. Results show that (1) every tested PTW (31015, 31016 and 31014) and Exradin A1SL chambers demonstrate a satisfying stability under irradiation. Other types of chambers have a stability complying with the stability criterion for some or none of them. (2) IBA CC01, IBA CC04 and Exradin A1SL show a proper response as a function of applied voltage for both polarities. PTW, Exradin A14SL and Exradin A16 do not. Only three types of chambers were deemed suitable as reference chambers according to LNE-LNHB requirements and specifications from McEwen (2010 Med. Phys.37 2179–93): Exradin A1SL chambers (3/3), IBA CC04 (2/3) and IBA CC01 (1/3). The Exradin A1SL type with an applied polarizing voltage of 150 V was chosen as an LNE-LNHB reference chamber type in 2 × 2 cm² radiation fields.

A detailed three-dimensional (3D) model of the coronary artery tree with cardiac motion has great potential for applications in a wide variety of medical imaging research areas. In this work, we first developed a computer-generated 3D model of the coronary arterial tree for the heart in the extended cardiac-torso (XCAT) phantom, thereby creating a realistic computer model of the human anatomy. The coronary arterial tree model was based on two datasets: (1) a gated cardiac dual-source computed tomography (CT) angiographic dataset obtained from a normal human subject and (2) statistical morphometric data of porcine hearts. The initial proximal segments of the vasculature and the anatomical details of the boundaries of the ventricles were defined by segmenting the CT data. An iterative rule-based generation method was developed and applied to extend the coronary arterial tree beyond the initial proximal segments. The algorithm was governed by three factors: (1) statistical morphometric measurements of the connectivity, lengths and diameters of the arterial segments; (2) avoidance forces from other vessel segments and the boundaries of the myocardium, and (3) optimality principles which minimize the drag force at the bifurcations of the generated tree. Using this algorithm, the 3D computational model of the largest six orders of the coronary arterial tree was generated, which spread across the myocardium of the left and right ventricles. The 3D coronary arterial tree model was then extended to 4D to simulate different cardiac phases by deforming the original 3D model according to the motion vector map of the 4D cardiac model of the XCAT phantom at the corresponding phases. As a result, a detailed and realistic 4D model of the coronary arterial tree was developed for the XCAT phantom by imposing constraints of anatomical and physiological characteristics of the coronary vasculature. This new 4D coronary artery tree model provides a unique simulation tool that can be used in the development and evaluation of instrumentation and methods for imaging normal and pathological hearts with myocardial perfusion defects.

Dynamic contrast-enhanced magnetic resonance imaging is increasingly applied for tumour diagnosis and early evaluation of therapeutic responses over time. However, the reliability of pharmacokinetic parameters derived from DCE-MRI is highly dependent on the experimental settings. In this study, the effect of sampling frequency (f_s) and duration on the precision of pharmacokinetic parameters was evaluated based on system identification theory and computer simulations. Both theoretical analysis and simulations showed that a higher value of the pharmacokinetic parameter K^trans required an increasing sampling frequency. For instance, for similar results, a relatively low f_s of 0.2 Hz was sufficient for a low K^trans of 0.1 min⁻¹, compared to a high f_s of 3 Hz for a high K^trans of 0.5 min⁻¹. For the parameter v_e, a decreasing value required a higher sampling frequency. A sampling frequency below 0.1 Hz systematically resulted in imprecise estimates for all parameters. For the K^trans and v_e parameters, the sampling duration should be above 2 min, but durations of more than 7 min do not further improve parameter estimates.

The beam orientation optimization (BOO) problem in intensity modulated radiation therapy (IMRT) treatment planning is a nonlinear problem, and existing methods to obtain solutions to the BOO problem are time consuming due to the complex nature of the objective function and size of the solution space. These issues become even more difficult in total marrow irradiation (TMI), where many more beams must be used to cover a vastly larger treatment area than typical site-specific treatments (e.g., head-and-neck, prostate, etc). These complications result in excessively long computation times to develop IMRT treatment plans for TMI, so we attempt to develop methods that drastically reduce treatment planning time. We transform the BOO problem into the classical set cover problem (SCP) and use existing methods to solve SCP to obtain beam solutions. Although SCP is NP-Hard, our methods obtain beam solutions that result in quality treatments in minutes. We compare our approach to an integer programming solver for the SCP to illustrate the speed advantage of our approach.

A high-resolution radiographic method for soft tissues in the small joints of the hand would aid in the study and treatment of rheumatoid arthritis (RA) and osteoarthritis (OA), which often attacks these joints. Of particular interest would be imaging with <100 µm resolution the joint cartilage, whose integrity is a main indicator of disease. Differential phase-contrast (DPC) or refraction-based x-ray imaging with Talbot grating interferometers could provide such a method, since it enhances soft tissue contrast and can be implemented with conventional x-ray tubes. A numerical joint phantom was first developed to assess the angular sensitivity and spectrum needed for a hand DPC system. The model predicts that, due to quite similar refraction indexes for joint soft tissues, the refraction effects are very small, requiring high angular resolution. To compare our model to experiment we built a high-resolution bench-top interferometer using 10 µm period gratings, a W anode tube and a CCD-based detector. Imaging experiments on animal cartilage and on a human finger support the model predictions. For instance, the estimated difference between the index of refraction of cartilage and water is of only several percent at ∼25 keV mean energy, comparable to that between the linear attenuation coefficients. The potential advantage of DPC imaging thus comes mainly from the edge enhancement at the soft tissue interfaces. Experiments using a cadaveric human finger are also qualitatively consistent with the joint model, showing that refraction contrast is dominated by tendon embedded in muscle, with the cartilage layer difficult to observe in our conditions. Nevertheless, the model predicts that a DPC radiographic system for the small hand joints of the hand could be feasible using a low energy quasi-monochromatic source, such as a K-edge filtered Rh or Mo tube, in conjunction with a ∼2 m long 'symmetric' interferometer operated in a high Talbot order.

Crucial to all cancer therapy modalities is a strong correlation between treatment and effect. Predictability of therapy success/failure allows for the optimization of treatment protocol and aids in the decision of whether additional treatment is necessary to prevent tumour progression. This work evaluated the relationship between cancer treatment and effect for intratumoural infusions of liposome-encapsulated ¹⁸⁶Re to head and neck squamous cell carcinoma xenografts of nude rats. Absorbed dose calculations using a dose-point kernel convolution technique showed significant intratumoural dose heterogeneity due to the short range of the beta-particle emissions. The use of three separate tumour infusion locations improved dose homogeneity compared to a single infusion location as a result of a more uniform radioactivity distribution. An improved dose–response correlation was obtained when using effective uniform dose (EUD) calculations based on a generic set of radiobiological parameters (R² = 0.84) than when using average tumour absorbed dose (R² = 0.22). Varying radiobiological parameter values over ranges commonly used for all types of tumours showed little effect on EUD calculations, which suggests that individualized parameter use is of little significance as long as the intratumoural dose heterogeneity is taken into consideration in the dose–response relationship. The improved predictability achieved when using EUD calculations for this cancer therapy modality may be useful for treatment planning and evaluation.

Compact, room temperature x-ray spectroscopy detectors are of interest in many areas including diagnostic x-ray imaging, radiation protection and dosimetry. Room temperature cadmium zinc telluride (CZT) semiconductor detectors are promising candidates for these applications. One of the major problems for CZT detectors is low-energy tailing of the energy spectrum due to hole trapping. Spectral post-correction methods to correct the tailing effect do not work well for a number of reasons; thus it is advisable to eliminate the hole trapping effect in CZT using physical methods rather than correcting an already deteriorated energy spectrum. One method is using a CZT detector with an electrode configuration which modifies the electric field in the CZT volume to decrease low-energy tailing. Another method is to irradiate the CZT surface at a tilted angle, which modifies depth of interaction to decrease low-energy tailing. Neither method alone, however, eliminates the tailing effect. In this work, we have investigated the combination of modified electric field and tilted angle irradiation in a single detector to further decrease spectral tailing. A planar CZT detector with 10 × 10 × 3 mm³ size and CZT detector with 5 × 5 × 5 mm³ size and cap-shaped electrode were used in this study. The cap-shaped electrode (referred to as CAPture technology) modifies the electric field distribution in the CZT volume and decreases the spectral tailing effect. The detectors were investigated at 90° (normal) and 30° (tilted angle) irradiation modes. Two isotope sources with 59.6 and 122 keV photon energies were used for gamma-ray spectroscopy experiments. X-ray spectroscopy was performed using collimated beams at 60, 80 and 120 kVp tube voltages, in both normal and tilted angle irradiation. Measured x-ray spectra were corrected for K x-ray escape fractions that were calculated using Monte Carlo methods. The x-ray spectra measured with tilted angle CAPture detector at 60, 80 and 120 kVp tube voltages were compared to corresponding theoretical spectra. The low-energy tailing was nearly completely eliminated from 59.6 and 122 keV isotope spectra, and 60, 80 and 120 kVp x-ray spectra, when CAPture detector was used with 30° tilted angle irradiation. It is concluded that using a CZT detector with modified electric field in tilted angle configuration resolves problem of the tailing effect in CZT detectors, opening promising possibilities in gamma-ray and x-ray spectroscopy applications.

Quantitative analysis of dynamic contrast enhanced magnetic resonance imaging (DCE-MRI) data requires the accurate determination of the arterial input function (AIF). A novel method for obtaining the AIF is presented here and pharmacokinetic parameters derived from individual and population-based AIFs are then compared. A Philips 3.0 T Achieva MR scanner was used to obtain 20 DCE-MRI data sets from ten breast cancer patients prior to and after one cycle of chemotherapy. Using a semi-automated method to estimate the AIF from the axillary artery, we obtain the AIF for each patient, AIF_ind, and compute a population-averaged AIF, AIF_pop. The extended standard model is used to estimate the physiological parameters using the two types of AIFs. The mean concordance correlation coefficient (CCC) for the AIFs segmented manually and by the proposed AIF tracking approach is 0.96, indicating accurate and automatic tracking of an AIF in DCE-MRI data of the breast is possible. Regarding the kinetic parameters, the CCC values for K^trans, v_p and v_e as estimated by AIF_ind and AIF_pop are 0.65, 0.74 and 0.31, respectively, based on the region of interest analysis. The average CCC values for the voxel-by-voxel analysis are 0.76, 0.84 and 0.68 for K^trans, v_p and v_e, respectively. This work indicates that K^trans and v_p show good agreement between AIF_pop and AIF_ind while there is a weak agreement on v_e.

We have proposed a zoom-in positron emission tomography (PET) system that incorporates a high-resolution detector into an existing PET scanner to obtain high-resolution images of a region of interest. Previously we have shown by computer simulations that the high-resolution detector can improve the overall system performance in terms of spatial resolution and lesion detectability. In this study, we assessed the resolution improvement in a real system by incorporating a high-resolution detector into our existing microPET II scanner. The high-resolution detector consists of a 14 × 28 array of 0.5 × 0.5 × 10 mm³ lutetium oxyorthosilicate scintillator elements and is placed near the center of the microPET II scanner. It is coupled to two 64-channel photomultiplier tubes (PMTs) via tapered optical fiber bundles. The PMT signals were read out by the electronics in the microPET II scanner. A 15 µCi Na-22 point source was positioned at various locations above the high-resolution detector. Images were reconstructed using the data measured by the microPET II scanner alone and the microPET II data combined with the high-resolution detector data. Profiles taken through the reconstructed point sources show substantial reduction in full-width-at-half-maximum along the direction parallel to the face of the high-resolution detector.

Monte Carlo (MC) methods are the gold standard for modeling photon and electron transport in a heterogeneous medium; however, their computational cost prohibits their routine use in the clinic. Cloud computing, wherein computing resources are allocated on-demand from a third party, is a new approach for high performance computing and is implemented to perform ultra-fast MC calculation in radiation therapy. We deployed the EGS5 MC package in a commercial cloud environment. Launched from a single local computer with Internet access, a Python script allocates a remote virtual cluster. A handshaking protocol designates master and worker nodes. The EGS5 binaries and the simulation data are initially loaded onto the master node. The simulation is then distributed among independent worker nodes via the message passing interface, and the results aggregated on the local computer for display and data analysis. The described approach is evaluated for pencil beams and broad beams of high-energy electrons and photons. The output of cloud-based MC simulation is identical to that produced by single-threaded implementation. For 1 million electrons, a simulation that takes 2.58 h on a local computer can be executed in 3.3 min on the cloud with 100 nodes, a 47× speed-up. Simulation time scales inversely with the number of parallel nodes. The parallelization overhead is also negligible for large simulations. Cloud computing represents one of the most important recent advances in supercomputing technology and provides a promising platform for substantially improved MC simulation. In addition to the significant speed up, cloud computing builds a layer of abstraction for high performance parallel computing, which may change the way dose calculations are performed and radiation treatment plans are completed.