Parts | Days | Selection | Search | Updates | Downloads | Help

ST: Fachverband Strahlen- und Medizinphysik

ST 1: Biomedical Imaging I

ST 1.8: Talk

Monday, March 9, 2015, 15:45–16:00, BZ.08.02 (HS 3)

Improved spatial resolution of X-ray phase-contrast computed tomography via iterative image deconvolution — •Fabio De Marco, Lorenz Birnbacher, Marian Willner, Mathias Marschner, Julia Herzen, and Franz Pfeiffer — Lehrstuhl für Biomedizinische Physik, Physik-Department & Institut für Medizintechnik, Technische Universität München

In a grating-based phase-contrast computed tomography (PC-CT) setup, differential phase data is retrieved by laterally scanning one grating of the Talbot-Lau interferometer. From the resulting set of images, a projection for each of three modalities (attenuation, differential phase and darkfield) is calculated.

The resolution of the stepping images (and by extension, the tomographic reconstructions), is limited mainly by the source size and the detector response. Their impact can be described as a convolution of the undisturbed image with the system's point-spread function (PSF). Having measured the shape of this PSF, a deconvolution can be performed to approximate the undisturbed image.

We examined the ability of several deconvolution algorithms to counteract these resolution-limiting effects. The algorithms were applied to the stepping images of CT scans of biological soft-tissue samples. The resolution of tomographic reconstructions is improved significantly, especially in the phase-contrast modality. The boundaries between tissue types are sharpened, which increases the number of recognizable features. However, an increase of noise is observed for all employed algorithms. The impact on quantitative values is also discussed.