My thesis work is about the correction of Partial Volume Effect (PVE). PVE is existing in all imaging system however the phenomenon rather affects in the emission tomography, which decreases the spatial resolution of the reconstructed image.
At the beginning I researched about this phenomenon and I tried to understand the relevant algorithms from the articles. I evaluated them according to the effectiveness and field of use. That was helpful to choose the algorithms which can be the most appropriate for the realization. In this semester my goals was to make some of those algorithms in MATLAB environment. I used its Image Processing Toolbox, thus the first stage of my work was to learn the basic rules and working mechanisms of this program. Then I made 4 algorithms. However I had to also build phantom images for their functions and testing of the algorithms. The exact design and values of the images were really important also as much as the algorithms. So the phantom design was also a detailed work. In course of the semester I built algorithms which are called as Recovery Coefficient [RC], Geometric Transfer Matrix [GTM], Deconvolution and Multiresolution approach. The different algorithms have different assumptions and have different functions however under the comparison the algorithms yield more or less same correction results if we don’t count with few little differences. The first 3 algorithms yield intensity corrected values, however the last algorithm yield real PVE corrected image which has higher spatial resolution than the original one.
In summary the algorithms yield valuable correction results on their own test image, however the results were not so distinct on a common image either. In the future I would like to continue this work. I would like to build up further algorithm which can give me wider approach in this field. Moreover I would like to generate more realistic test images for the more reliable result of their evaluation.