Date on Master's Thesis/Doctoral Dissertation

5-2015

Document Type

Doctoral Dissertation

Degree Name

Ph. D.

Department

Electrical and Computer Engineering

Degree Program

Electrical Engineering, PhD

Committee Chair

Frieboes, Hermann Bueno

Committee Co-Chair (if applicable)

Welch, Karla

Committee Member

El-Baz, Ayman

Committee Member

Zurada, Jacek

Committee Member

Harnett, Cindy

Subject

Bioinformatics; Data mining; Medicine--Data processing

Abstract

The proposed research applies machine learning techniques to healthcare applications. The core ideas were using intelligent techniques to find automatic methods to analyze healthcare applications. Different classification and feature extraction techniques on various clinical datasets are applied. The datasets include: brain MR images, breathing curves from vessels around tumor cells during in time, breathing curves extracted from patients with successful or rejected lung transplants, and lung cancer patients diagnosed in US from in 2004-2009 extracted from SEER database. The novel idea on brain MR images segmentation is to develop a multi-scale technique to segment blood vessel tissues from similar tissues in the brain. By analyzing the vascularization of the cancer tissue during time and the behavior of vessels (arteries and veins provided in time), a new feature extraction technique developed and classification techniques was used to rank the vascularization of each tumor type. Lung transplantation is a critical surgery for which predicting the acceptance or rejection of the transplant would be very important. A review of classification techniques on the SEER database was developed to analyze the survival rates of lung cancer patients, and the best feature vector that can be used to predict the most similar patients are analyzed.

Share

COinS