Brain atrophy occurs as a symptom of many diseases. The software package, Statistical Parametric Mapping (SPM) is one of the most respected and commonly used tools in the neuroimaging community for quantifying the amount of grey matter (GM) in the brain based on magnetic resonance (MR) images. One aspect of quantifying GM volume is to identify, or segment, regions of the brain image corresponding to grey matter. A recent trend in the field of image segmentation is to model an image as a graph composed of vertices and edges, and then to "cut" the graph into subgraphs corresponding to different segments. In this thesis, we incorporate image segmentation algorithms based on graph-cuts into a GM volume estimation system, and then we compare the GM volume estimates with those achieved via SPM. To aid in this comparison, we use 20 T1-weighted normal brain MR images simulated using BrainWeb . We obtained results verifying the graph-cuts technique better approximated the GM volumes by halving the error resulting from SPM preprocessing.
Library of Congress Subject Headings
Brain--Imaging--Mathematical models; Neuroanatomy--Mathematical models
Department, Program, or Center
School of Mathematical Sciences (COS)
Zanca, Ashley, "A Graph theoretic approach to quantifying grey matter volume in neuroimaging" (2013). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus