| Summary: | 碩士 === 國立陽明大學 === 生物醫學影像暨放射科學系暨研究所 === 96 === Cerebrovascular diseases are the cause of mortality for a long time. Many of the diagnosis of cerebrovascular diseases have been relied on conventional angiography. The result of diagnosing cerebrovascular diseases from conventional angiography is conceived as the golden standard. However, the procedures of conventional angiography are invasive. Recently, some non-invasively diagnostic tools have been developed, for examples, ultrasound, MR, and CT.
CT angiography (CTA) has been used to detect vessel occlusion and the frontline of stroke imaging. By different post-processing method, the reconstructed CTA data can provide many types of images, such as volume rendering technique (VRT), for the doctor to visualize the nidus from different angles. The common applications are in myocardial infarction and renal artery. However, in head and neck, the density of bone in the base of skull resulted from the contrast agent is similar to that of vessel. This makes it difficult to remove the bone for revealing the intracranial, skull-base, and extracranial vascular structures as compared to the conventional angiography.
In this study, we develop a post-processing method on CTA image. We use the regionally local thresholding and morphological operations to remove the bone and segment the vessel with contrast agent out from CTA image, and create the MIP images to demonstrate the vessels similar to the conventional angiography images. Three-dimensional surface models of vessels are also reconstructed using the marching cubes algorithm and the level set method to better the visualization.
|
|---|
