Department of Computer Science and Engineering
University of Washington

Biomedical Imaging and Informatics

The Biomedical and Health Informatics Program at the University of Washington is a graduate program in the UW Medical School. Our research group collaborates with and includes students and faculty from this program. Here are some of our collaborative projects.

CT Image Registration, Segmentation, and Retrieval

Jiun-Hung Chen is working on construction of 3D models from image data. His current work concentrates on finding point correspondences between 3D medical entities such as abdominal organs.

Chia-Chi Teng's 2007 dissertation describes a new methodology for obtaining contours of lymph node regions in CT scans of cancer patients. The method first identifies the most similar reference patients using a similarity measure based on landmarks that can be reliably extracted from the images. It then uses a constrained optimization procedure to find the best deformable mapping from the new patient to the reference patient. The mapping is used to map the predrawn contours on the reference patient to the CT scan of the new patient, so that the lymph node regions can be reliably found and the radiation treatment accordingly planned.


3D Skull Shape Analysis and Retrieval

Salvador Ruiz Correa and Jill Lin have worked on automatic classification and retrieval of skull images for craniofacial applications Jill Lin's 2006 dissertation described a new methodology for describing 3D shapes, including skulls and feet, in terms of symbolic shape descriptors that could then be used to classify shapes or to retrieve similar ones.


Ulstrasound Image Analysis


Anatomy Ontologies

The Foundational Model of Anatomy is a large reference ontology for the human body developed by Dr. Cornelius Rosse and his colleagues. We have participated in the development of this model and constructed several query systems that employ graphical user interfaces to access its data.

The Emily graphical query inteface allows the user to query the FMA for relational information about the more than 80,000 anatomical entities represented in the ontology.

The CAIS system represents the 2006 Ph.D. dissertation work of Ravensara Travillian who developed the Structural Difference Method (SDM), a method for comparing pairs of species motivated by graph matching. The CAIS graphical user interface allows users to compare 2 species (human, mouse, rat) with respect to the entities and relationships found in the FMA.