Iterative reconstruction using a Monte Carlo based system transfer matrix for dedicated breast positron emission tomography
UMass Chan Affiliations
Department of RadiologyDocument Type
Journal ArticlePublication Date
2014-08-28
Metadata
Show full item recordAbstract
To maximize sensitivity, it is desirable that ring Positron Emission Tomography (PET) systems dedicated for imaging the breast have a small bore. Unfortunately, due to parallax error this causes substantial degradation in spatial resolution for objects near the periphery of the breast. In this work, a framework for computing and incorporating an accurate system matrix into iterative reconstruction is presented in an effort to reduce spatial resolution degradation towards the periphery of the breast. The GATE Monte Carlo Simulation software was utilized to accurately model the system matrix for a breast PET system. A strategy for increasing the count statistics in the system matrix computation and for reducing the system element storage space was used by calculating only a subset of matrix elements and then estimating the rest of the elements by using the geometric symmetry of the cylindrical scanner. To implement this strategy, polar voxel basis functions were used to represent the object, resulting in a block-circulant system matrix. Simulation studies using a breast PET scanner model with ring geometry demonstrated improved contrast at 45% reduced noise level and 1.5 to 3 times resolution performance improvement when compared to MLEM reconstruction using a simple line-integral model. The GATE based system matrix reconstruction technique promises to improve resolution and noise performance and reduce image distortion at FOV periphery compared to line-integral based system matrix reconstruction.Source
J Appl Phys. 2014 Aug 28;116(8):084903. Link to article on publisher's siteDOI
10.1063/1.4894085Permanent Link to this Item
http://hdl.handle.net/20.500.14038/48180PubMed ID
25371555Related Resources
Link to Article in PubMedae974a485f413a2113503eed53cd6c53
10.1063/1.4894085