Ground-based synthetic aperture radar (GBSAR) imaging systems are used extensively in earth observation and remote sensing applications. They are lightweight, cost-efficient, and resolve the main limitations of airborne and spaceborne SAR systems, such as low data collection rate, complicated implantation, and inefficient viewing angle to the imaging scene. This paper evaluates the signal processing results of the developed and implemented GBSAR system at the microwave remote sensing laboratory (MRESL) of the University of Tehran. The radar sensor consists of linear frequency modulated continuous wave (LFMCW) operating in S-band with 330 MHz of bandwidth that is mounted on a 1. 4 m linear rail with a stepper motor. The radar sensor moves every 1 cm and records the backscattered echoes. After the dechirping process and digitizing, the recorded signal is stored in a two-dimensional array, known as SAR 2D raw matrix. Experiments were conducted by illuminating the SAR sensor toward a scene containing corner reflectors. SAR raw data were processed by implementing the back-projection algorithm, and results were evaluated and compared with numerical simulations. Furthermore, the Hann window was employed as a weighting function on the raw signal, and its effect was evaluated on the resulting SAR image. The obtained results are promising and show the effectiveness of the developed GBSAR in the discrimination of targets at two range and cross-range directions.

Background: Reconstruction of positron emission tomography (PET) imagery is essential for accurate diagnosis in nuclear medicine. Thus, several image reconstruction methods were developed to enhance image quality and accuracy. Objectives: We performed PET studies using a new myocardial imaging agent, (5-[18F] fluoropentyl) triphenylphosphonium cation ([18F]FPTP), in left coronary artery (LCA)-occluded rats, and compared the quality of cardiac PET images generated via four reconstruction methods (FBP, OSEM2D, OSEM3D, and 3DRP). Additionally, the infarction size was measured on the polar map of each reconstructed image and compared with defect size measured via 2, 3, 5-triphenyltetrazolium chloride (TTC) staining. Materials and Methods: MicroPET was performed in Sprague-Dawley rats (n = 8) with LCA ligation. Static images were acquired for 30 min after the injection of [18F] FPTP (37 MBq/200  L) via tail vain. MicroPET images were generated using four different reconstruction methods: filtered Backprojection (FBP), two-dimensional or three-dimensional ordered subsets expectation maximization (OSEM2D or OSEM3D), and three-dimensional reprojection (3DRP) algorithm. Image contrast was calculated using the maximum and minimum perfusion values in the polar map. The infarction size was measured on the polar map of each reconstructed image and compared with defect size measured from TTC staining. Results: The location and size of myocardial infarction on PET images correlated closely with that observed with TTC staining. Amongthe four reconstruction methods, OSEM3D provided the best assessment of infarct size (r2 = 0. 994, P< 0. 001) and the highest image contrast, performing significantly better than FBP (P = 0. 005) and 3DRP (P = 0. 005). Conclusion: OSEM3D may provide better image quality and higher contrast than other methods for small animal imaging with the new myocardial imaging agent, [18F] FPTP.

Introduction: Photon attenuation is one of the main causes of quantitative errors and artifacts in SPECT. CT based attenuation map is necessary to correct for this effect accurately. A number of attenuation related artifacts are described. A fast and memory efficient algorithm is proposed for attenuation correction.Methods: In our CT based attenuation correction algorithm, attenuated projections of an arbitrary estimation are calculated. These projections are compared with real projections. The error due to comparison updates the first arbitrary image. This iterative procedure continues till there is no difference between mathematical and real projections. To consider the effect of Poisson noise during reconstruction, EM statistical reconstruction is used. The effect of attenuation is enrolled in both projection and Backprojection steps. To accelerate the reconstruction algorithm, a rotating image in a fix grid with bilinear interpolation is used to simplify ray tracing for creating mathematical projections. OSEM with much faster convergence rate is also used instead of MLEM as iterative algorithm. Different phantom configurations with uniform and non-uniform attenuation map are used to evaluate the accuracy of this algorithm. Projections free from the effect of attenuation were also simulated. Reconstructed image from these attenuation free projections is considered as reference image. Normalized mean square error between reference and corrected image and image contrast were used for quantitative evaluation of this algorithm.Results: contrast of central hot spot reduced to about half of its real contrast due to attenuation. Non-uniformities, like hot skin, bright long related spots and colds bone related spots were also created in non-corrected images due to attenuation. All attenuation related artifacts removed after attenuation correction. NMSE is reduced from 1 in non-corrected images to 0.1 in corrected images. Much better fit in line profile of corrected images and reference images were also observed.Conclusion: By using more suitable models for image formation it is possible to reconstruct images with much higher quantitative and qualitative accuracy which is essential in both diagnostic and image guided therapy procedures.

Introduction: The goal of SPECT is determining activity distribution inside the patient body. To achieve this goal a rotating gamma camera acquires many projections around the patient. These projections are later used by reconstruction algorithms to reconstruct the slices corresponding to activity bio distribution. The primary assumption of these algorithms is that Radon transform is a suitable model for image formation. Therefore Filtered Back Projection (FBP) which used inverse Radon transform is the main method for reconstruction; however this model doesn’t consider the effect of attenuation and non-ideal collimation. Therefore the final reconstructed slices degrade due to the effect of photons attenuation and are blurred due to collimator and detector response (CDR) which is the main cause of resolution degradation. In this study a more suitable model contains the effect of both attenuation and CDR is introduced and used during reconstruction by iterative OSEM algorithm to correct for their degrading effects.Methods: at first a point source of Tc-99m is modeled and placed in different distances from the face of collimator. SIMIND Monte Carlo simulator is used to create projections from this activity source. After finding suitable models, a digital brain phantom (ZUBAL) with a typical Tc99m-ECD distribution is used to validate the ability of our algorithm. The effect of attenuation and CDR were modeled in both projection/Backprojection of the OSEM algorithm.Results: qualitatively, better match between activity phantom and reconstructed images are found in comparison with conventional reconstruction algorithms. Quantitatively, a dramatic increase in amount of activity in different parts of brain image is found. The contrast of gray to white matter is increased more than 30%. SNR is also increased by factor of 2. The FWHM of final reconstructed image of point source of activity is also reduced from 7mm at center of field of view to less than 4mm by suggested algorithm which shows a considerable improvement in image resolution.Conclusion: by using more suitable models for image formation it is possible to reconstruct images with much higher quantitative and qualitative accuracy which is essential in both diagnostic and image guided therapy procedures.