|
Sarang Joshi Assistant Professor of Radiation Oncology and Biomedical Engineering University of North Carolina at Chapel Hill
|
 |
|
|
 Medial
Based Deformable Geometry: We have been developing a Bayesian multi-scale
three dimensional deformable template approachbased on a medial representation
for the segmentation and shape characterizationof anatomical objects in
medical imagery. The modeling approach taken for building exemplary templates
and associated transformations is basedon a multi-scale medial representation.
The transformations defined in this framework are parameterized directly
in terms of natural shape operations, such as thickening, twisting and
bending.
 Tumor Detection and Segmentation:
We have developing automatic detection and segmentation of various types
of Brian Tumors imaged via multimodal imaging techniques such as T1, T2
and Gadlinium contarct enhanced MRI. The algorithms are based on using
Global and Local structural variation beetween normal subjects and tumor
patients such as asymetry and structular Grey and White matter deviations
from normal atlases.
|
|
|
Diffeomorphic
Deformable Atlas Mapping: We have been studying the variability of
human anatomy by generating smooth maps from a single template to a familyof
targets. In our approach the template represents the typical structure;variation
is accommodated by the definition of probabilistic diffeomorphictransformations
applied to the templates.
Shape of Brain Sub-Structures:
The
shape and volume of the hippocampus has been associated with various studies
of neuropsychiatric illnesses.We have been building analytical methods
for characterizing the shape and variation of such brain structurs in variusdeseasessuch
as Schizophernia, Alzhimers and Epilepsy.
Anatomical
ShapeVariation During Radiation Treatment:Although the prostate, bladder
and the rectum are nearly stable in shape there can be clinically appreciable
deformation of these structures during the course of external beam radiation
treatment for prostate cancer. Using deformable atlas mapping we have been
developing automated procedures for quantifying anatomical changes that
take place during treatment of the bladder rectum and the prostate in patients,
as imaged via intra-treatment CT images.