Meeting Banner
Abstract #3201

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.

Join Here