Convolutional Forward Modeling for Actual Slice Profile Estimation
Xiaoguang Lu1, Peter Speier2, and Ti-chiun Chang3
1Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, United States, 2Siemens Healthcare, Erlangen, Germany, 3Siemens Corporate Technology, Princeton, NJ, United States
Resolving
slice thickness for better MR reconstruction is desirable, where actual slice
profile plays a crucial role. Conventional blind deconvolution formulation
includes both original signals and slice profile as unknowns, which is an
ill-posed problem with high complexity. We propose a convolutional forward
model (CFM), leveraging additional orthogonal stack(s) with an added
convolution process in the formulation to fit actual forward imaging process
accurately, resulting in a significantly simplified slice profile estimation
problem. The actual slice profile is calculated through a data-driven approach.
Experimental results demonstrate that the proposed method is robust to handle
various challenges.
This abstract and the presentation materials are available to members only;
a login is required.
