Translator Disclaimer
12 May 2004 Eigen nodule: view-based recognition of lung nodule in chest x-ray CT images using subspace method
Author Affiliations +
We previously proposed a recognition method of lung nodules based on experimentally selected feature values (such as contrast, circularities, etc.) of the suspicious shadows detected by our Quoit filter. In this paper, we propose a new recognition method of lung nodule using each CT value itself in ROI (region of interest) area as a feature value. In the clustering stage, first, the suspicious shadows are classified into some clusters using Principal Component (PC) theories. A set of CT values in each ROI is regarded as a feature vector, and then the eigen vectors and the eigen values are calculated for each cluster by applying Principal Component Analysis (PCA). The eigen vectors (we call them Eigen Images) corresponding to the first 10 largest eigen values, are utilized as base vectors for subspaces of the clusters in the feature space. In the discrimination stage, correlations are measured between the unknown shadow and the subspace which is spanned by the Eigen Images. If the correlation with the abnormal subspace is large, the suspicious shadow is determined to be abnormal. Otherwise, it is determined to be normal. By applying our new method, good results have been acquired.
© (2004) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Yoshihiko Nakamura, Gentaro Fukano, Hotaka Takizawa, Shinji Mizuno, Shinji Yamamoto, Tohru Matsumoto, Yukio Tateno, and Takeshi Iinuma "Eigen nodule: view-based recognition of lung nodule in chest x-ray CT images using subspace method", Proc. SPIE 5370, Medical Imaging 2004: Image Processing, (12 May 2004);

Back to Top