Faculty members working in imaging

Partial differential equations can be used to model digitized images that arise from arial photography or brain scans. Then variational methods and other PDE-based methods may be used to clean up the images and isolate important features such as buildings and tumor boundaries.

He uses numerical methods and wavelets to study inverse problems and enhance medical images. A particular inverse problem he has worked on is the modeling of heat flow of rocket plumes. His work in medical imaging includes the development of new methods for PET image formation and deblurring the discrete gaussian.

In medical imaging, the problem may be to chmod detect in human tissue small masses that may be of cancerous potential via a limited angle tomography technique. To detect land mines, the use of ground penetrating radar emitted from a plane requires the ability to pick out and distinguish the small mines from other ground based objects, such as rocks. By constructing mathematical algorithms to deconvolute the overwhelming array of data received in these situations, images can be created to pinpoint the location of the mines.

He uses probability, potential theory, and wavelets in medical imaging. Applications include a the development of new methods for PET image formation.

He has developed decomposition methods for parallel computing and image deblurring methods. Since 1987 he has worked on automatic identification of the epicardial and endocardial borders of the heart in 2-D echocardiographic images.