Abstract
Periventricular lesions are associated with hyperintensity in Magnetic Resonance brain images of Acquired Immunodeficiency Syndrome patients. The objective of this research has been the development of a user-friendly software to quantify such lesions for the purpose of studying ways to prevent or reverse this disease. Basically this work has led to a tool for the medical community to quantify periventricular lesions which are currently assessed in a qualitative manner. Provisions are made for a radiologist to input lesion seedpoints interactively. A comprehensive feature vector is generated and then used as the homogeneity criterion in a region growing algorithm. Two different stopping criteria, average Mahalanobis distance and neural network, are evaluated as part of the region growing algorithm. The results obtained indicate that the perpendicular region growing with the neural network stopping criterion provides the closest match to the manual segmentation of lesions. A Graphical-User-Interface is designed in order to have a clinically usable software environment.
Madhavan, Sridhar (1996). Quantification of periventricular lesions from MR brain images. Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -1996 -THESIS -M3392.