Vocal Cord Segmentation from CT Images using Machine Learning
Vocal Cord Segmentation from CT Images using Machine Learning
Author : AGUS PRATONDO; BINH P. NGUYEN; CHEE-KONG CHUI; SIM-HENG ONG Published on : ACCAS 2014 (Fukuoka, Japan)
Abstract
Vocal cord segmentation from CT images is difficult because of poor image contrast and unclear boundaries. In many cases, there are regions with very weak signals and ill-defined boundaries, particularly in connective tissues. To overcome this problem, a framework to segment regions of interest using machine learning is proposed. Feature extraction is used to represent the characteristics of the desired regions. The features are then fed to a machine learning process. The proposed framework reduces the complexity and enhances the segmentation accuracy. Experimental results show that vocal cord segmentation using machine learning is able to segment the ill-defined boundaries. We compare the results with level set method. The comparison indicates that vocal cord segmentation via learning machine outperforms the conventional level set method.
