3 October 2016 Dictionary learning-based CT detection of pulmonary nodules
Author Affiliations +
Abstract
Segmentation of lung features is one of the most important steps for computer-aided detection (CAD) of pulmonary nodules with computed tomography (CT). However, irregular shapes, complicated anatomical background and poor pulmonary nodule contrast make CAD a very challenging problem. Here, we propose a novel scheme for feature extraction and classification of pulmonary nodules through dictionary learning from training CT images, which does not require accurately segmented pulmonary nodules. Specifically, two classification-oriented dictionaries and one background dictionary are learnt to solve a two-category problem. In terms of the classification-oriented dictionaries, we calculate sparse coefficient matrices to extract intrinsic features for pulmonary nodule classification. The support vector machine (SVM) classifier is then designed to optimize the performance. Our proposed methodology is evaluated with the lung image database consortium and image database resource initiative (LIDC-IDRI) database, and the results demonstrate that the proposed strategy is promising.
© (2016) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Panpan Wu, Kewen Xia, Yanbo Zhang, Xiaohua Qian, Ge Wang, Hengyong Yu, "Dictionary learning-based CT detection of pulmonary nodules", Proc. SPIE 9967, Developments in X-Ray Tomography X, 99671S (3 October 2016); doi: 10.1117/12.2236780; https://doi.org/10.1117/12.2236780
PROCEEDINGS
12 PAGES


SHARE
Back to Top