Computer Visual System for 3D Flow Analysis of Aortic MR Images

• Main Authors: Chia-Ho Su, 蘇家禾
• Other Authors: Yung-Nien Sun
• Format: Others
• Language: zh-TW
• Published: 2001
• Online Access: http://ndltd.ncl.edu.tw/handle/59055421102108753313

碩士 === 國立成功大學 === 資訊工程研究所 === 89 === In recent years, the applications of functional MRI (magnetic resonance images) in medical diagnose have been rapidly developed. Besides high-speed and high-resolution imaging, magnetic resonance imaging also supplies quantitative assessment of biological functional distribution. Especially in the analytic diagnosis of cardiovascular system, magnetic resonance imaging provides both the anatomy and Phase-Contrast furnish images which enriched flow information. In Phase-Contrast MR Imaging, velocity information can be acquired across an imaging plane in an imaging period, and all the three orthogonal components of velocity vector are readily provided. So magnetic resonance imaging can assist us in understanding the status and course of blood flow during a cardiac cycle that is helpful in cardiac diagnoses. During a cardiac cycle, we have to observe the motion of heart and the blood flow of vessel from a series of MR images. Thus if we segment the vessel area manually, it will be very time consuming. In this paper, we first use the active contour model to perform the automatic detection of vessel contours in a series of MR images. However, in the traditional active contour model, the initial contour must be very close to the actual object boundary, otherwise it won’t be easily attracted by object boundary. To overcome this limitation, we introduce a novel external force — Electric Gradient Field (EGF). EGF uses the principle, that positive and negative charges mutually attract, to increase the capture range of object boundary. It can be effectively applied to vessel wall tracking. After we finish the tracking of vessel wall, we then use these vessel contours to construct the three-dimensional vessel model by using the generalized cylinder model. Finally we employ the 3D vessel model and incorporate the MR velocity images to visualize the blood flow. Depending on the types of flow fields, we provide appropriate methods for blood flow visualization, including the 3D profile, vector map, and 3D streamline. The three methods were experimentally effective in visualizing various types of blood flow fields.