Abstract

Single-photon emission computed tomography (SPECT) is a biomedical imaging modality for visualisation and quantification of biochemical and physiological processes in living tissues. The resolution of SPECT is quite limited and can be improved by the use of high resolution anatomical information so as to limit smoothing to within anatomically defined boundaries.

In this project we work on SPECT reconstruction using anatomical information from Magnetic Resonance Imaging (MRI) as a priori knowledge about the activity distribution in the brain. The trade-off between enforcing the anatomical information and preserving emission contrast in regions that do not correspond to anatomically defined boundaries is one of the main concerns for such studies.

Spatially variant dIffusion PDE ’s are able to preserve important edges while filter image from noise. We try various approaches to embed available anatomical data into this variational framework.