The goal of the first part is to develop an application which helps doctors
to do a diagnosis of the cervical spine. The application takes a set of
X-Ray images (lateral, frontal, peg view, two obliques) and does some image
processing stuff on them to clear the edges of the bones and finally shows
the alignment of the bones to each other.
We have to discover if we can use available image processing libraries, e.g. Image Vision Library from Silicon Graphics, Inc.
At first we have to find out which edge detection works best to get binaery images which contains most of needed information. After removing isolated pixels we do an edge following. With the result it should be possible to detect angulations or displacements.
After we discovered that the ImageVision Library can not satisfy us, we go on now
with an own graphics library written in a previous project. ImageVision Library
was good to play with, to get the knowledge how it has to be but it is not
The steps of the image processing in the application will be nearly the same. We have to implement a sharpen, edge detection and noise removal operator. After that we have to find a way to implement some sort of heuristic functions which finds the alignments.
In the past weeks we got heavy support from the doctors. They descriped in a way that computer people can understand what they need in the program. We focus on detecting the alignment lines in lateral projections:
Here is a page which shows you some of the X-Ray images we are working with. The page loades about 560kB of pictures!
[startpage] [project part 1] [project part 2] [week 1] [week 2] [week 3] [week 4] [week 5] [week 6] [week 7] [week 8] [week 9] [week 10] [week 11] [References]