An investigation of attenuation correction with attenuation map library in PET imaging

In this work, we propose a novel positron emission tomography (PET) attenuation correction method with attenuation map collection. Our concept is summarized as follows: Firstly, we construct a library containing a large number of attenuation maps for the specific object from different patients. Secondly, by characterizing the statistical information of these maps, we build the statistical template to represent these objects. The statistical template is used to implement attenuation correction for this kind of objects. We evaluate the performance of this method. We generate 30 digital phantoms with 30 attenuation maps to construct the attenuation map library. The phantoms are all consisted of three parts and the linear attenuation coefficients of same part from the 30 phantoms obey Gaussian distribution. To reduce the complexity, the phantoms are with the same geometrical configuration. By averaging the 30 attenuation maps, we build the statistical template. We employ Monte-Carlo simulation to generate the PET data. The results are compared with the conventional PET transmission scan method.