http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Automatic Detection of GGO in CT Lung Images by Using Statistical Features and Neural Networks
Hyoungseop Kim,Yoshifumi Katsumata,Joo Kooi Tan,Seiji Ishikawa 대한전자공학회 2009 ITC-CSCC :International Technical Conference on Ci Vol.2009 No.7
In this paper, we described an algorithm of automatic detection of GGO candidate region to develop a CAD system from lung CT images by use of statistical features which is obtained density and shape features. In this algorithm, first, image pre-processing techniques such as segmentation of lung areas, binarization technique are introduced. In the second step, statistical features based on density features which are obtained mean, standard deviation, skewness, and kurtosis. Also two shape features which are obtained spiral scanning filter, and Gabor filter are introduced. In our clustering step, GGO area can be detect by using artificial neural networks. The proposed technique applied to 31 lung CT image sets. From this database, classification rates of a true positive rate of 84.2%, false positive rate of 57% and number of false positive 1.07/slice under the receiver operating characteristic analysis were achieved. The aim of this study is segmentation of lungs region and detection of abnormal area for the GGO by using thoracic MDCT image sets. This study also tried to decrease the amount of false positive rates and increase the amount of true positive rates so that the accuracy of performance.