| Popis: |
During the past project period, we proposed to collaborate closely with DOE’s Thomas Jefferson National Accelerator Facility (Jefferson Lab or JLab) to design a compact, ultra-high-resolution, high-sensitivity gamma camera for quantifying brain-tumor distributions of I-131. We also proposed to continue our on-going research in developing and evaluating pinhole collimation for quantitative ultra-high-resolution imaging of I-131-labeled MAbs. We have made excellent progress in accomplishing much of the research related to pinhole collimation. Many of the most significant results have been presented in peer-reviewed journal articles and conference proceedings. We have also made good progress in collaborating with JLab's Detector Group in developing a compact, ultra-high-resolution, gamma camera. A prototype I-131 imager was delivered to Duke on May 28, 2003. Our research results are summarized in the following sections. A. JLAB-DUKE DEDICATED BRAIN-TUMOR IMAGING SYSTEM A.1. Determination of Optimal Collimator Design During the current project period a prototype I-131 dedicated brain imager has been designed and built. Computer simulations and analysis of alternate designs were performed at Duke to determine an optimal collimator design. Collimator response was characterized by spatial resolution and sensitivity. Both geometric (non-penetrative) and penetrative sensitivities were considered in selecting an optimal collimator design. Based on these simulation results, two collimator designs were selected and more » built by external vendors. Initial imaging results were obtained using these collimators. B. INITIAL DEVELOPMENT OF SPECT RECONSTRUCTION SOFTWARE FOR JLAB-DUKE CAMERA B.1. Modeling Thick Septa and Collimator Holes: Geometrical-Phantom Study A geometrical phantom was designed to illuminate spatial resolution effects. The phantom includes a uniformly attenuating medium that consists of all voxels within an elliptical cylinder that is centered on the axis of rotation, infinitely long, and with minor and major diameters of 15.0 and 22.0cm. Computer-simulated projections of the phantom were created. We observed that thick septa create a periodic and strong variation in sensitivity across the surface of the gamma camera, as evident in these projections. C. PINHOLE POINT-RESPONSE FUNCTION (PRF) AND ROOT-MEAN-SQUARE (RMS) NOISE During the previous project period, we developed an accurate analytic expression to determine the sensitivity of pinhole collimation that included the effects of penetration. During the current project period, we have developed an accurate model of the point-response function (PRF) of pinhole collimators. D. COMPLETE SAMPLING: THEORETICAL AND COMPUTATIONAL DEVELOPMENTS During this project period, we have investigated the complete-sampling conditions for orbits of pinhole collimators and have published these data. We have made progress in both complete-sampling theory and also in computational methods. Pinhole collimation has similar complete sampling properties to cone-beam collimation. Complete sampling of an object cannot be obtained from the circular rotation of the aperture about the object. We have investigated helical-orbit pinhole SPECT scans as a method of obtaining completely sampled data. Helical orbits were evaluated because they offer the potential of a small ROR for high sensitivity and resolution combined with complete sampling. We compared reconstructions from simulated circular-orbit and simulated helical-orbit projection data and observed a marked improvement in image quality when helical-orbits are used for the data acquisition. E. PINHOLE CALIBRATION STUDIES We have begun a study of the effects of mechanical and electronic shifts on reconstruction that suggests that even very small shifts (a fraction of a millimeter) can introduce substantial artifacts in the reconstruction. We have acquired experimental calibration data using point sources. We fitted the centroids simultaneously to the expected mechanical and electronic components determined by an analytically derived equation. The mechanical and electronic shifts were determined using a maximum likelihood fit. The data show good agreement with expectation. The axial-shift results for the pinhole collimator demonstrated an approximately sinusoidal characteristic; that is, it is angularly dependent. To test the effect mechanical shift on reconstructed image quality, experimental data were acquired using a micro cold-rod phantom. The SPECT image that used the correct estimated mechanical shift was markedly improved compared with the other uncorrected SPECT images. « less |
