Many medical conditions can be diagnosed based on non-invasive imaging. The images include x-ray, CT, MRI and ultrasound. Once those images are created, the biomedical image processing starts. All projects involve the common need to extract information from the image that will allow physicians and researchers to better understand a problem.
The biomedical image processing projects are all collaborative with other researchers or end users who have found open problems in need of a solution.
Sample projects include:
Registration of 2D and 3D images – a time series (movie) of a joint in motion can be combined with a static 3D image of the joint to provide information on the positions of the bones in that joint during dynamic or loaded activities
Ultrasound image segmentation – ultrasound is an inexpensive medical imaging process. If the spine can be imaged in 3D with ultrasound, then spinal compression can be better understood and a tool developed to do monitoring of patients or works for back injuries
3D segmentation – The positions of key muscles and ligaments join bones can be determined from the Visual Human data. Rather than segmenting each image manually, every 3rd or Nth image can be used. We are quantifying how sparsely the images can be segmented to build 3D models and still provide a good 3D model through interpolation versus smoothness of the object contours
Find out more about Dr. Barney Smith’s research.
Find out more about the Signal Processing Lab.