Tobias Funk

Radiation dose estimate in small animal SPECT and PET.

Calculations of radiation dose are important in assessing the medical and biological implications of ionizing radiation in medical imaging techniques such as SPECT and PET. In contrast, radiation dose estimates of SPECT and PET imaging of small animals are not very well established. For that reason we have estimated the whole-body radiation dose to mice and rats for isotopes such as 18F, 99mTc, 201Tl, (111)In, 123I, and 125I that are used commonly for small animal imaging. We have approximated mouse and rat bodies with uniform soft tissue equivalent ellipsoids. The mouse and rat sized ellipsoids had a mass of 30 g and 300 g, respectively, and a ratio of the principal axes of 1:1:4 and 0.7:1:4. The absorbed fractions for various photon energies have been calculated using the Monte Carlo software package MCNP. Using these values, we then calculated MIRD S-values for two geometries that model the distribution of activity in the animal body: (a) a central point source and (b) a homogeneously distributed source, and compared these values against S-value calculations for small ellipsoids tabulated in MIRD Pamphlet 8 to validate our results. Finally we calculated the radiation dose taking into account the biological half-life of the radiopharmaceuticals and the amount of activity administered. Our calculations produced S-values between 1.06 x 10(-13) Gy/Bq s and 2.77 x 10(-13) Gy/Bq s for SPECT agents, and 15.0 x 10(-13) Gy/Bq s for the PET agent 18F, assuming mouse sized ellipsoids with uniform source distribution. The S-values for a central point source in an ellipsoid are about 10% higher than the values obtained for the uniform source distribution. Furthermore, the S-values for mouse sized ellipsoids are approximately 10 times higher than for the rat sized ellipsoids reflecting the difference in mass. We reviewed published data to obtain administered radioactivity and residence times for small animal imaging. From these values and our computed S-values we estimated that the whole body dose in small animals ranges between 6 cGy and 90 cGy for mice and between about 1 cGy and 27 cGy for rats. The whole body dose in small animal imaging can be very high in comparison to the lethal dose to mice (LD50/30 approximately 7 Gy). For this reason the dose in small animal imaging should be monitored carefully and the administered activity should be kept to a minimum. These results also underscore the need of further development of instrumentation that improves detection efficiency and reduces radiation dose in small animal imaging.

A multipinhole small animal SPECT system with submillimeter spatial resolution

Single photon emission computed tomography (SPECT) is an important technology for molecular imaging studies of small animals. In this arena, there is an increasing demand for high performance imaging systems that offer improved spatial resolution and detection efficiency. We have designed a multipinhole small animal imaging system based on position sensitive avalanche photodiode (PSAPD) detectors with the goal of submillimeter spatial resolution and high detection efficiency, which will allow us to minimize the radiation dose to the animal and to shorten the time needed for the imaging study. Our design will use 8 x 24 mm2 PSAPD detector modules coupled to thallium-doped cesium iodide [CsI(Tl)] scintillators, which can achieve an intrinsic spatial resolution of 0.5 mm at 140 keV. These detectors will be arranged in rings of 24 modules each; the animal is positioned in the center of the 9 stationary detector rings which capture projection data from the animal with a cylindrical tungsten multipinhole collimator. The animal is supported on a bed which can be rocked about the central axis to increase angular sampling of the object. In contrast to conventional SPECT pinhole systems, in our design each pinhole views only a portion of the object. However, the ensemble of projection data from all of the multipinhole detectors provide angular sampling that is sufficient to reconstruct tomographic data from the object. The performance of this multipinhole PSAPD imaging system was simulated using a ray tracing program that models the appropriate point spread functions and then was compared against the performance of a dual-headed pinhole SPECT system. The detection efficiency of both systems was simulated and projection data of a hot rod phantom were generated and reconstructed to assess spatial resolution. Appropriate Poisson noise was added to the data to simulate an acquisition time of 15 min and an activity of 18.5 MBq distributed in the phantom. Both sets of data were reconstructed with an ML-EM reconstruction algorithm. In addition, the imaging performance of both systems was evaluated with a uniformity phantom and a realistic digital mouse phantom. Simulations show that our proposed system produces a spatial resolution of 0.8 mm and an average detection efficiency of 630 cps/MBq. In contrast, simulations of the dual-headed pinhole SPECT system produce a spatial resolution of 1.1 mm and an average detection efficiency of 53 cps/MBq. These results suggest that our novel design will achieve high spatial resolution and will improve the detection efficiency by more than an order of magnitude compared to a dual-headed pinhole SPECT system. We expect that this system can perform SPECT with submillimeter spatial resolution, high throughput, and low radiation dose suitable for in vivo imaging of small animals.

Computed tomographic metal artifact reduction for the detection and quantitation of small features near large metallic implants: a comparison of published methods.

Computed tomographic imaging of tissue surrounding metallic implants is often limited by metal artifacts. This paper compares 3 existing metal artifact reduction techniques that are based on segmentation of metal-affected regions in native images, followed by reprojection of segmented areas into original Radon space, removal of metal trace(s), and renewed reconstruction: Detector row-wise linear interpolation, 2-dimensional interpolation, and combination of row-wise linear interpolation and adaptive filtering. For each method, improvements of CT number accuracy and signal-noise as well as contrast-noise ratios near the prosthesis and in the image periphery over the values found for native images were evaluated in a phantom experiment simulating osteolytic bone lesions of different size and density around a Chrome-Cobalt hip prosthesis stem. Improvement in diagnostic usability was evaluated as lesion detectability by size. Quantitative and qualitative results showed that the linear interpolation and the combination method removed the artifacts most effectively. The mean accuracy error over different regions of interest placed in the direct vicinity of the metal and in the periphery of the object decreased 10-fold with linear interpolation. These methods increased contrast-noise ratio up to 68% of that measured on artifact-free images for the least dense lesion. Qualitatively, the linear interpolation and the combination method improved the lesion detectability and enabled differentiation of different lesion densities. However, in proximity to the stem, some artifacts remained for all methods. We conclude that published algorithms for metal artifact reduction substantially improve image quality for CT imaging of a metallic object and may be adequate for quantitative measurements except for the direct vicinity of the metallic object.

A new CdZnTe-based gamma camera for high resolution pinhole SPECT

We have developed and tested a pixelated detector for pinhole SPECT. The 80/spl times/80 detector has a pixel size of 2.5/spl times/2.5 mm/sup 2/ and incorporates the room-temperature solid-state semiconductor CdZnTe which offers direct charge conversion and good stopping power for radionuclides used for small animal SPECT. To optimize low energy photon detection, the detector is operated at a temperature of 15/spl plusmn/1/spl deg/C, and has been designed to minimize absorption losses for photons entering the active detector material. The CdZnTe detector demonstrates good uniformity, spatial resolution and energy resolution. Transmission images obtained with an /sup 125/I point source demonstrate its low-energy performance, while dual isotope images were obtained using /sup 99m/Tc and /sup 201/Tl. The detector therefore offers good performance for high-resolution small animal SPECT.

Modeling and Correction of Spatial Distortion in Position-Sensitive Avalanche Photodiodes

Position-sensitive avalanche photodiodes (PSAPDs) are a promising alternative to photomultiplier tubes for the development of a new generation of gamma imagers. They offer compactness, high gain and superior quantum efficiency. PSAPDs having a sensitive surface of up to 28times28 mm2 have been fabricated. However, unlike pixellated imaging devices having a similar configuration, each PSAPD can achieve submillimeter position sensing over its surface with only four readout channels. This key feature is obtained by Anger-logic event positioning from the signals of four corner anodes printed on a resistive layer covering one of the PSAPD surfaces. This readout scheme provides high degree of multiplexing for reading position and energy information from the device, but leads to pincushion distortion in the spatial information due to the nonlinear charge sharing pattern associated with the potential across the resistive layer. We have developed a method to reproduce and correct this distortion based on finite-element simulations of the readout configuration. The resistive layer and the anodes are represented by a two-dimensional array of resistors and this circuit is numerically solved to obtain the signal on the four anodes for different current injection nodes. The relation between the injection positions and the resulting Anger positions is modeled and then used to correct experimental data. The algorithm was tested on 99mTc flood images of a 16times16 array of 0.4times0.4times4 mm3 CsI(Tl) crystals and successfully restored the regular pattern. The correction procedure is fast and robust, and constitutes a step toward the realization of a low-cost, high-resolution gamma camera based on PSAPDs

The feasibility of an inverse geometry CT system with stationary source arrays.

PURPOSE Inverse geometry computed tomography (IGCT) has been proposed as a new system architecture that combines a small detector with a large, distributed source. This geometry can suppress cone-beam artifacts, reduce scatter, and increase dose efficiency. However, the temporal resolution of IGCT is still limited by the gantry rotation time. Large reductions in rotation time are in turn difficult due to the large source array and associated power electronics. We examine the feasibility of using stationary source arrays for IGCT in order to achieve better temporal resolution. We anticipate that multiple source arrays are necessary, with each source array physically separated from adjacent ones. METHODS Key feasibility issues include spatial resolution, artifacts, flux, noise, collimation, and system timing clashes. The separation between the different source arrays leads to missing views, complicating reconstruction. For the special case of three source arrays, a two-stage reconstruction algorithm is used to estimate the missing views. Collimation is achieved using a rotating collimator with a small number of holes. A set of equally spaced source spots are designated on the source arrays, and a source spot is energized when a collimator hole is aligned with it. System timing clashes occur when multiple source spots are scheduled to be energized simultaneously. We examine flux considerations to evaluate whether sufficient flux is available for clinical applications. RESULTS The two-stage reconstruction algorithm suppresses cone-beam artifacts while maintaining resolution and noise characteristics comparable to standard third generation systems. The residual artifacts are much smaller in magnitude than the cone-beam artifacts eliminated. A mathematical condition is given relating collimator hole locations and the number of virtual source spots for which system timing clashes are avoided. With optimization, sufficient flux may be achieved for many clinical applications. CONCLUSIONS IGCT with stationary source arrays could be an imaging platform potentially capable of imaging a complete 16-cm thick volume within a tenth of a second.

Progress of focusing x-ray and gamma-ray optics for small animal imaging

Significant effort is currently being devoted to the development of noninvasive imaging systems that allow in vivo assessment of biological and biomolecular interactions in mice and other small animals. Ideally, one would like to discern these functional and metabolic relationships with in vivo radionuclide imaging at spatial resolutions approaching those that can be obtained using the anatomical imaging techniques (i.e., <100 μm, which would help to answer outstanding questions in many areas of biomedicine. In this paper, we report progress on our effort to develop high resolution focusing X-ray and gamma-ray optics for small-animal radionuclide imaging. The use of reflective optics, in contrast to methods that rely on absorptive collimation like single- or multiple-pinhole cameras, decouples spatial resolution from sensitivity (efficiency). Our feasibility studies have refined and applied ray tracing routines to design focusing optics for small animal studies. We also have adopted a replication technique to manufacture the X-ray mirrors, and which in experimental studies have demonstrated a spatial resolution of ~190 μm. We conclude that focusing optics can be designed and fabricated for gamma-ray energies, and with spatial resolutions, and field of view suitable for in vivo biological studies. While the efficiency of a single optic is limited, fabrication methods now are being developed that may make it possible to develop imaging systems with multiple optics that could collect image data over study times that would be practical for performing radionuclide studies of small animals.

Small-animal radionuclide imaging with focusing gamma-ray optics

Significant effort currently is being devoted to the development of noninvasive imaging systems that allow in vivo assessment of biological and biomolecular interactions in mice and other small animals. While physiological function in small animals can be localized and imaged using conventional radionuclide imaging techniques such as single-photon emission tomography (SPECT) and positron emission tomography (PET), these techniques inherently are limited to spatial resolutions of 1-2 mm. For this reason, we are developing a small animal radionuclide imaging system (SARIS) using grazing incidence optics to focus gamma-rays emitted by 125I and other radiopharmaceuticals. We have developed a prototype optic with sufficient accuracy and precision to focus the 27.5 keV photons from 125I onto a high-resolution imaging detector. Experimental measurements from the prototype have demonstrated that the optic can focus X-rays from a microfocus X-ray tube to a spot having physical dimensions (approximately 1500 microns half-power diameter) consistent with those predicted by theory. Our theoretical and numerical analysis also indicate that an optic can be designed and build that ultimately can achieve 100 μm spatial resolution with sufficient efficiency to perform it in vivo single photon emission imaging studies in small animal.

Monte Carlo simulations of compact gamma cameras based on avalanche photodiodes

Avalanche photodiodes (APDs), and in particular position-sensitive avalanche photodiodes (PSAPDs), are an attractive alternative to photomultiplier tubes (PMTs) for reading out scintillators for PET and SPECT. These solid-state devices offer high gain and quantum efficiency, and can potentially lead to more compact and robust imaging systems with improved spatial and energy resolution. In order to evaluate this performance improvement, we have conducted Monte Carlo simulations of gamma cameras based on avalanche photodiodes. Specifically, we investigated the relative merit of discrete and PSAPDs in a simple continuous crystal gamma camera. The simulated camera was composed of either a 4 x 4 array of four channels 8 x 8 mm2 PSAPDs or an 8 x 8 array of 4 x 4 mm2 discrete APDs. These configurations, requiring 64 channels readout each, were used to read the scintillation light from a 6 mm thick continuous CsI:Tl crystal covering the entire 3.6 x 3.6 cm2 photodiode array. The simulations, conducted with GEANT4, accounted for the optical properties of the materials, the noise characteristics of the photodiodes and the nonlinear charge division in PSAPDs. The performance of the simulated camera was evaluated in terms of spatial resolution, energy resolution and spatial uniformity at 99mTc (140 keV) and 125I ( approximately 30 keV) energies. Intrinsic spatial resolutions of 1.0 and 0.9 mm were obtained for the APD- and PSAPD-based cameras respectively for 99mTc, and corresponding values of 1.2 and 1.3 mm FWHM for 125I. The simulations yielded maximal energy resolutions of 7% and 23% for 99mTc and 125I, respectively. PSAPDs also provided better spatial uniformity than APDs in the simple system studied. These results suggest that APDs constitute an attractive technology especially suitable to build compact, small field of view gamma cameras dedicated, for example, to small animal or organ imaging.

Simulation and Validation of Point Spread Functions in Pinhole SPECT Imaging

Modeling and assessment of point spread functions (PSFs) of pinhole collimators is essential for the design of small animal single photon emission computed tomography (SPECT) imaging systems, and are gaining increasing importance with the advent of multipinhole imaging techniques. PSFs also can be used in resolution recovery methods implemented in reconstruction algorithms. Therefore, we have developed and validated a ray-tracing approach to calculate PSFs and absolute detection efficiency of pinhole collimators in radionuclide imaging. The PSFs were calculated for user defined pinhole and source geometries with multiple rays to account for collimator penetration. For validation we compared our simulations to analytical models, Monte Carlo simulations from literature, and experiments with 99 mTc sources using a variety of pinhole geometries including knife and keel edges. We find that shape and magnitude of the simulated PSFs are in very good agreement with analytical and experimental results. Importantly, our simulations show that the absolute detection efficiency of pinhole systems can be computed with an accuracy error of less than 10% using the ray-tracing approach. In contrast to Monte Carlo simulations ray-tracing simulations are computational very efficient and therefore very fast. In conclusion, we developed a ray-tracing method that calculates PSFs and detection efficiencies for different pinhole and source geometries quickly and reliably