Digital Image Analysis with Medical Images

Tyson Lipscomb HON 396 09/05/07 Project Abstract

For my project I will be researching and implementing a computer program that will address the problem of image registration and comparison in medical applications. The problem that will be addressed is the creation of an automated or semi-automated program that will analyze a set of medical images, such as x-rays or MRIs, and show the regions in which change has occurred and analyze any changes in size of areas that the user defines as being areas of interest. Once complete, the program will be able to accept a series of images, arranged those images into aligned layers, and accept input from the user that will select areas that require analysis. After these initial steps are complete the program will analyze the images and the regions of interest and create a new layer showing the areas that have changed, display the rate of change if the original input images are dated and determine an approximate percentage of change between each image.

Java will be used to write this program as it is the computer language with which I am most familiar. I anticipate using the Java Advanced Imaging (JAI) API for much of the image-specific programming required in this project as it has more advanced features dealing with image manipulation and analysis than the default Java API. The first stage of programming will consist of developing a GUI-based system for selecting input images and manually aligning them. This will consist of programming various image manipulations such as partial transparency, rotation and scaling. If time allows I will implement a feature that will automatically align the input images for the user. The next part of the programming phase will be writing the algorithms that will analyze the input images and display the changes between them both graphically and mathematically. This process will then be further refined to accept input from the user specifying what areas, if any, should be ignored and what areas are of special interest to the user. I expect the image comparison and analysis programming will be the most difficult and time consuming and the GUI development to be the least intensive.