2011 Poster Sessions : Automatic Segmentation of Laryngeal Cartilages Using Support Vector Machines

Student Name : Reeve Ingle
Advisor : Dwight Nishimura
Research Areas: Information Systems
Abstract:
Magnetic Resonance Imaging (MRI) is critical in the staging of laryngeal cancer, one of the most common types of head and neck caner. However, the presence and extent of cartilage invasion is difficult to assess, and accurate staging of the tumor is necessary to properly treat laryngeal caner and avoid unnecessary procedures such as total laryngectomy. Accurate, automated delineation of the laryngeal cartilages from the surrounding soft tissue could significantly aid the clinician during diagnosis.

In this work, we investigate the use of supervised and unsupervised learning algorithms to automatically segment high-resolution magnetic resonance (MR) images of the larynx. We implement a multi-contrast, multi-dimensional technique, which has proven useful for segmenting articular cartilage in the knee. This approach is hindered by a lack of automatic intensity correction to compensate for the coil sensitivity profile when a dedicated array is used. We propose a custom intensity correction algorithm, and we compare the performance of support vector machines and k-means clustering to automatically segment the cartilages from high-resolution MR images of the larynx.

Bio:
Reeve Ingle is a PhD candidate in the Department of Electrical Engineering at Stanford University, where he works with Professor Dwight Nishimura in the Magnetic Resonance Systems Research Lab (MRSRL). His current research interests include signal processing, image processing, optimization, and machine learning applications in Magnetic Resonance Imaging (MRI). He received the B.S. in Electrical Engineering from Georgia Tech in 2008 and the M.S. in Electrical Engineering from Stanford in 2010, and he has prior co-op and internship experience at NASA (2005 – 2008) and the Department of Defense (2007).