Partial Volume Effect Correction using Anisotropic Backward Diffusion
Budapest University of Technology and Economics
Hungarian Association for Image Processing and Pattern Recognition (KEPAF), 2013
@article{marta2013partial,
title={Partial Volume Effect Correction using Anisotropic Backward Diffusion},
author={Marta, Zsolt and Szirmay-Kalos, Laszlo},
year={2013}
}
This paper proposes an algorithm for correcting Partial Volume Effect in Positron Emission Tomography (PET) images, using registered Computed Tomography (CT) data to enhance the blurred PET image. The algorithm is based on a forward-and-backward anisotropic heat equation solver, deblurring the PET image along CT gradients. A forward diffusion force is also utilized to stabilize the process where necessary. The algorithm retains average original PET intensity avoiding the introduction of negative PET values, a crucial property in clinical uses. We implemented the algorithm for GPU and tested it with real measured PET/CT data.
February 14, 2013 by hgpu
