Mixed Fuzzy C-Mean Clustering and its Application to CT Image Segmentation in PET/CT Imaging

• Main Author: 李曉飛
• Other Authors: Ching-Han Hsu
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/58522363647655396391

碩士 === 國立清華大學 === 原子科學系 === 93 === A PET/CT (Positron Emission Tomography/ Computer Tomography) has unique capability of acquiring accurately aligned functional and anatomical images for human body, and supplies the excellent worth in clinical diagnosis. However, PET images are not able to provide correct quantitative analysis due to the attenuation of photons. Many researches have applied computed tomography (CT) data as X-ray based attenuation correction for positron emission tomography (PET) imaging. In this study, we present an automatic segmented method of CT images in whole body scan to improve attenuation correction in PET imaging. A mixed fuzzy C-means (FCM) clustering which combines the use of the intensity attribute of the homogeneous objects with the standard deviation attribute of the inhomogeneous objects is introduced. The experimental results indicate that this method not only enhances the anatomical localization of bone and air in CT images, but also reduces the influence from the contrast agent. Besides, it reduces the bias due to transmission data, and promotes the practical utility in clinical diagnosis.