Ph.D. Candidacy Exam - Mouse Brain Image Segmentation with Level Set Method
Date: March 17, 2005
Time: 12:30 PM
Location: LeBow Engineering Center, Room: 240
Automated volume or region of interest (VOIs and ROIs) segmentation is of general importance in medical imaging and a large number of potential computer vision solutions are being explored in the research community. My particular interest is in a mouse neuroinformatics setting. Our laboratory is developing image-based query system to aid neurogeneticists in using our Mouse Brain Library. This large database is of particular value in functional genomics and in particular in a genetic mapping method known as quantitative trait loci mapping. My goal is to develop level set methods to segmentation of the data contained in this resource. Specifically my aims are:
Aim 1: Two and half dimensions region delineation. The MBL consists of images of sections from a large number of brains of mice from an advanced intercross set of strains. For each brain, the database contains images of about 70-80 30um sections- a section every 150 um. For this aim, I will evaluate performance of level set methods in segmenting blocks of sections using the first manually delineated section as seed.
Aim 2: Segmentation of cross animal. The difficulty of the segmentation task will be increased for this aim and instead of automated delineation across the brain of an individual mouse, I will evaluate performance in seeding the level set algorithms with delineations contained in the MBL canonical atlas.
The LeBow Engineering Center is located at the corner of 31st and Market Streets.