Research  |  Core Facilities  |  Patient Studies  |  Tech Transfer  |  Seminars  |  Intranet  |  Careers  |  Search  |  Contact Us  |  Ways To Give                          HOME

More about
Dr. Conchello:

Dr. Conchello's CV in brief

Publications

Genetic Models of Disease Research Program

Computational Microscopy Laboratory

José-Angel Conchello, Ph.D. Associate Member, Genetic Models of Disease Research Program Affiliate Professor, Program in Biomedical Engineering, Chemical Engineering   Department, University of Oklahoma

Research Interests
Investigators interested in the three dimensional (3D) structure of microscopic specimens usually acquire a through-focus series of 2D images, each at a different focal plane through the specimen. These 2D images show sharp focus on structures located within a fraction of a micrometer of the focal plane but also incorporate a large amount of light from unfocused structures outside of the focal plane. This light obscures in-focus detail. We have developed several computational methods to remove out-of-focus light from 2D optical slices and increase fine detail in the resulting 3D stack. To derive these deconvolution algorithms, we begin with a mathematical-physics description of image formation and the recording of these images under the microscope. The goal is to undo image degradation inherent in this process.

The recently-developed structured-illumination microscope provides depth resolution comparable to that of a confocal microscope. I developed and tested an image estimation method that greatly improves the image resolution of this microscope (Conchello, 2005).

Existing deconvolution methods assume that imaging conditions do not vary as a through-focus stack is recorded. However, for many thick biological specimens, this does not hold. An exact formulation for this process, although possible, requires large amounts of computer memory and very long computation times. We recently developed a method that incorporates approximations to the process of image formation and makes computer time and memory requirements practical (Preza and Conchello, 2004).

Current methods also require precise information on imaging conditions, including the refractive index and thickness of all material between the objective and the specimen. We developed a method to estimate these parameters while out-of-focus light is being removed (Markham and Conchello, 1999). We are modifying our method to estimate only the few imaging parameters that are usually unknown. This will greatly reduce computation time and make use of the algorithm more practical.

To make the algorithms available and easy to use, we are closely collaborating with Dr. Mike Dresser to develop OMRFCOSM, a software package for computational microscopy that will include several methods for deconvolution, as well as tools for image correction, visualization and analysis.

Joined OMRF Scientific Staff in 2003.

Mailing Address
Genetic Models of Disease Research Program, MS 48
Oklahoma Medical Research Foundation
825 N.E. 13th Street
Oklahoma City, Oklahoma 73104

Contact Information
Phone: (405) 271-6137
Fax: (405) 271-7312
E-mail: Jose-Conchello@omrf.org