Loading ...
Sorry, an error occurred while loading the content.

Wednesday, 4/8, Mert Sabuncu - Image-driven population analysis

Expand Messages
  • Anastasia Yendiki
    Wed 4/8 at noon Seminar room 2204 149 13th St., Charlestown Navy Yard Mert Sabuncu, Ph.D. Postdoctoral Associate Computer Science and Artificial Intelligence
    Message 1 of 1 , Apr 1 10:30 AM
    View Source
    • 0 Attachment
      Wed 4/8 at noon
      Seminar room 2204
      149 13th St., Charlestown Navy Yard

      Mert Sabuncu, Ph.D.
      Postdoctoral Associate
      Computer Science and Artificial Intelligence Lab, MIT

      Title:
      Image-driven Population Analysis through Mixture Modeling

      Abstract:
      In this talk, I will present *iCluster*, a fast and efficient algorithm
      that clusters a set of images while co-registering them using a
      parametrized, nonlinear transformation model. The output of the algorithm
      is a small number of template images that represent different modes in a
      population and an image-driven partitioning of the population. This is in
      contrast with traditional, hypothesis-driven computational anatomy
      approaches that assume a single template to construct an atlas and
      partition the population based on clinical/demographic/genetic data. We
      derive the algorithm using a generative model of an image population as a
      mixture of deformable template images. I will present experimental results
      that demonstrate the utility of iCluster in several contexts. In one
      experiment, I will motivate a multiple atlas strategy to develop automatic
      segmentation tools for a pool of subjects that consists of healthy
      controls and schizophrenia patients. Next, I will show how we employed
      iCluster to partition a data set of 416 whole brain MR volumes of subjects
      aged 18 through 96 years into three sub-populations, which mainly
      correspond to age groups. The templates reveal significant structural
      differences across these age groups that confirm previous findings in
      aging research. In another experiment, we ran iCluster on a group of
      dementia patients and healthy controls. The algorithm discovered two modes
      that correspond to a healthy majority and a sub-population of patients
      with dementia. These experiments suggest that the proposed image-driven
      clustering strategy can be used to discover sub-populations associated
      with interesting structural or functional "modes."
    Your message has been successfully submitted and would be delivered to recipients shortly.