The main goal of this research is to develop an open and extensible software environment for medical image segmentation. Image segmentation has many applications in medical imaging, including tumor localization, and radiation therapy planning; however its current use falls far short of its potential. One reason for this is the lack of support for image segmentation in commercially available general-purpose software tools. Our proposed research will overcome this limitation by providing a software environment will include a wide variety of state-of-the-art segmentation algorithms, and it will be extensible so that users can add new algorithms easily. Our software environment will further segmentation algorithm research by providing the basic infrastructure for image segmentation algorithm development, such as tools for image display, region-of-interest selection, manual segmentation, and algorithm evaluation, as well as the data structures for segmentation. In Phase I of this research, we have developed a prototype of such a software environment using recent advances in software engineering. In Phase II we will further develop the prototype to produce a full-featured software system. We will design and add new segmentation algorithms and expand the algorithm evaluation capabilities. We will also apply the software to three important medical problems where reliable image segmentation is the key requirement.
We envision several families of products from this research: (l) Software applications for solving specific medical problems, targeted to clinical specialists. (2) General-purpose segmentation program for clinical generalists, to be combined with a develop infrastructure kit for use by software vendors and algorithm researchers. (3) Software components for use by software vendors to add segmentation capabilities to their image processing software. (4) Add in Software Modules
