报告题目：Image Reconstruction for Quantitation in Brain PET , SPECT and CT

报告人:    Professor  Andrew Todd-Pokropek

         (University College London )

报告时间：5月25日（周一）15:30

报告地点：广B106

报告摘要:

Quantitation of brain uptake of a tracer, important in the analysis of diseases such as AD and MS, requires accurate tomographic reconstruction, attenuation correction, partial volume correction and image fusion. It is an essential component of CT, SPECT and PET imaging. Two classes of reconstruction algorithms exist: analytic such as Filtered Back Projection (FBP), and iterative such as Maximum Likelihood Expectation Maximisation (MLEM) and its many variants. All such algorithms have parameters which need to be tuned, for example number of iterations, internal corrections, and measures of success such as quadratic error, resolution etc.  In particular with iterative algorithms better results can be obtained, but convergence is an issue, and many such algorithms seem to start by converging towards a good solution and then diverging, and so stopping criteria need to be investigated. Simple optimization can be considered as a least squares problem, but non-linear methods many in fact produce better solutions. The best quantitative results will also depend on acquisition parameters, such as number of events included / dose. Note also time of flight (TOF) acquisition in PET. Similar algorithmic issues exist with respect to image fusion (including with MR). Image quality v quantitative accuracy is clinically important. A number of case studies are presented illustrating various methodologies and limitations. These issues are important in trying to obtain the best clinical results. The field for developing tomographic reconstruction and correction algorithms is currently very competitive.

报告人简介：

Professor (Emeritus) Andrew Todd-Pokropek was Head of the Department of Medical Physics and Bioengineering’ University College London. His initial research theme was instrumentation in nuclear medicine, including the design and construction of the first SPECT system (for brains) in Europe. In 1992 he became Full Professor, University College London, Head of the Group Medical Image Processing in Medical Physics and responsible for physics applications in the Dept. of Radiology of the Hospital for Sick Children GOS. Research themes included the extraction of quantitative information from medical images, in particular  in nuclear medicine CT and MRI, registration and fusion of multi-modality images, n-D image and signal processing – including visualization. He wrote the Portable Image Processing(PIP) package, which was chosen by the IAEA for its project "upgrading of analogue gamma cameras" distributed in more than 150 copies to >40 notably developing countries, marketed by spin-off company ‘Leapfrog Technology’ Ltd. In 1999- 2008 2000- 2007 was a Co-applicant of major EPSRC/MRC grant to create Interdisciplinary Research Collaboration (IRC) on medical image and signal processing.  Current research includes advances in tomographic reconstruction and the optimizing of image quality in medical images