Font Size: a A A

Research On The Imaging Method For Task-specific Panel-based PET Scanner

Posted on:2015-05-06Degree:DoctorType:Dissertation
Country:ChinaCandidate:L WanFull Text:PDF
GTID:1224330428466095Subject:Biomedical engineering
Abstract/Summary:PDF Full Text Request
Positron emission tomography (PET) is a non-invasive medical imaging technique, which provides three-dimensional images reflecting the functional and biological processes in the body. Diseases are biological processes, and PET is sensitive to these processes. Therefore, PET has become crucial for clinical applications, such as the early detection and diagnosis, treatment planning and monitoring in fields of oncology, cardiology and neurology.The traditional PET consists of hundreds of detector blocks, which are arranged to for-m ring geometry. In contrast, panel-PET system adopts an opposing pair of large-area panel detectors. This configuration has an inherent large solid-angle which leads to a high sensitiv-ity. Furthermore, its open and adjustable geometry allows other operation being performed simultaneously during a PET scanning. Thus, it is suitable for certain novel applications, such as PET-guided radiation therapy, biopsy and surgery. Its compact structure also re-duces the manufacture complexity and makes it convenient in applications like portable or bedside imaging. However, appropriate tomographic reconstruction methods for panel-PET had yet to be developed. The major obstacle is the inherit data incompleteness problem (in-cluding missing view and data truncation), which causes severe distortions and artifacts in reconstructed images and inaccuracies in quantification.In this dissertation study, we investigated the tomographic image reconstruction meth-ods for panel-PET system. First, the iterative algorithm was applied to image reconstruction for the system with smaller missing view (<30degrees). A small animal PET adopting a pair of large-area detectors in compact scanner geometry was built, and its performance prop-erties were evaluated. No noticeable image artifacts were observed over the entire imaging volume even though the data are theoretically incomplete.18F-FDG rat images with good visual quality were achieved. More importantly, the results showed that a central sensitivity of28.2%can be obtained with an energy window (EW) of250-750keV and a coincidence time window (CW) of10ns. According to the results of literature search, it is the highest- sensitivity PET system. The parallex error effects were corrected by applying model-based image reconstruction. Therefore, an isotropic and uniform spatial resolution of1.2mm full width at half maximum (FWHM) was obtained over the entire imaging field of view.Secondly, time of flight (TOF) information was utilized as a constraint in image recon-struction for panel PET system with larger missing view (>90degrees). A panel TOF-PET scanner was designed and its imaging performance was evaluated. Monte Carlo simula-tion results showed that the distortions and artifacts were effectively reduced with enhanced timing resolution. The results also demonstrated that the proposed panel-PET system had the feasibility of detecting lesions not less than1cm in diameter, with a300ps FWHM timing resolution. A noticeable improvement in detectablity of small lesions was observed. The improved timing resolution can also relax the requirements of the accuracy of system response matrix and the necessity of normalization. Therefore, the panel TOF-PET system has more flexibility in design and applications.The last but not the least, we developed a novel TOF-based image reconstruction algorithm for ROI imaging, where data truncations exist in most projection views. Monte Carlo simulation showed only trivial differences between the results of ROI reconstruction and traditional global reconstruction. Furthermore, the ROI reconstruction only uses the coincidence events originated inside the ROI, thus the event counts for reconstruction are greatly reduced. The case study indicated that the ROI reconstruction only requires~10%event counts of the global reconstruction. Since the computational time for image reconstruction is almost proportional to the number of events, the proposed algorithm will be practically valuable for certain applications requiring fast reconstruction, such as real-time PET-guided tumor-tracking radiation therapy.
Keywords/Search Tags:panel-PET, missing view, data truncation, iterative reconstruction, time offlight, region of interest
PDF Full Text Request
Related items