Alumni Directory

Eric Foo
PhD, Summer 2008

Home Dept: Mechanical Engineering

PhD Dissertation title: A Framework for Tumor Segmentation and Interactive Immersive Visualization of Medical Image Data for Surgical Planning

PhD Advisor: Eliot Winer

Area of PhD research: Medical imaging technology has changed the way doctors and surgeons diagnose and operate in the last 100 years. Since the discovery of the X-ray, doctors and surgeonshave successfully utilized the ability to "see" inside a patient to further improve patient care. The advancement of computed tomography and magnetic resonance technology with faster and higher resolution scans has improved the analysis and visualization of patient data. As technology continues to accommodate the need for higher quality medical imaging, there is a continuing need for additional research in efficient ways of extracting crucial information from these vast amounts of data. This dissertation presents the framework for analyzing and visualizing digital medical images. Two new segmentation methods have been developed: a probability based segmentation algorithm, and a segmentation algorithm that uses a fuzzy rule based system to generate "similarity" values for segmentation. A visualization software application has also been developed to effectively view and manipulate digital medical images on a desktop computer as well as in an immersive environment. For the probabilistic segmentation algorithm, image data are first enhanced by manually setting the appropriate window center and width, and if needed a sharpening or noise removal filter is applied. To initialize the segmentation process, a user places a seed point within the object of interest and defines a search region for segmentation. Based on the pixels' spatial and intensity properties, a probabilistic selection criterion is used to extract pixels with a high probability of belonging to the object. To facilitate the segmentation of multiple slices, an automatic seed selection algorithm was developed to keep the seeds in the object as its shape and/or location changes between consecutive slices. The second segmentation method, a new segmentation method using a fuzzy rule based system to segment tumors in a three-dimensional CT data was also developed. To initialize the segmentation process, the user selects a region of interest (ROI) within the tumor in the first image of the CT study set. Using the ROI's spatial and intensity properties, fuzzy inputs are generated for use in the fuzzy rules inference system. Using a set of predefined fuzzy rules, the system generates a defuzzified output for every pixel in terms of similarity to the object. Pixels with the highest similarity values are selected as tumor. This process is automatically repeated for every subsequent slice in the CT set without further user input, as the segmented region from the previous slice is used as the ROI for the current slice. This creates a propagation of information from the previous slices, used to segment the current slice. The membership functions used during the fuzzification and defuzzification processes are adaptive to the changes in the size and pixel intensities of the current ROI. The proposed method is highly customizable to suit different needs of a user, requiring information from only a single two-dimensional image. Segmentation results from both algorithms showed success in segmenting the tumor from seven of the ten CT datasets with less than 10% false positive errors and five test cases with less than 10% false negative errors. The consistency of the segmentation results statistics also showed a high repeatability factor, with low values of inter- and intra-user variability for both methods.

Employment upon graduating:
academicPost Doctoral Student
Iowa State University
Ames, IA

<---- Back to search results