Evaluation of the Robustness of the Brain Parenchymal Fraction for Brain Atrophy Measurements
Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
In certain diseases, like Multiple Sklerosis and Alzheimer's disease, the progression of the disease can be measured by whole brain atrophy. A difficulty with this is that all people have very different scull sizes, thus also very different brain sizes. This makes it almost impossible to establish "normal values" for brain size. The spread is very large and the method is not practical to use for individual patients. A method with less spread in healthy persons is to use the ratio of the Brain Parenchymal Fraction (BPF). The use of Brain Parenchymal Fraction has increased steadily since it was first introduced in 1999.
BPF = BPV/ICV
This study was performed to increase the knowledge of what is normal and to evaluate the robustness of the BPF as a measurement for brain atrophy. Among other things, the change in the BPF when calculated from incomplete volumes (parts of the scull missing in the set of MR images) was evaluated.
The results show that when parts are missing from the top (superior) of the scull the resulting BPF is strictly higher than the correct PBF and when parts are missing from the lower (inferior) part of teh scull the resulting BPF is stritly lower than teh correct value.
Two different methods where tried to compensate for missing parts. The first method was to find a variable factor to compensate with, the size of this factor was depending on how much of the scull that was missing. The second method was to interpolate the ICV and BPV curves and from the new interpolated curves, calculate a new BPF.
The method of compensating incomplete volumes using a factor calculated as a function of the intercranial volume of the first/last available slice turned out to be the better.
Place, publisher, year, edition, pages
2013. , 53 p.
Brain Parenchymal Fraction, BPF, Brain atrophy, MS, Multiple Sklerosis
Medical Image Processing
IdentifiersURN: urn:nbn:se:liu:diva-105801ISRN: LiTH-IMT/MI30-A-EX-13/513-SEOAI: oai:DiVA.org:liu-105801DiVA: diva2:710651
Subject / course
2013-11-07, Linköping, 15:00 (Swedish)
Andersson, MatsWarntjes, Marcel