Wang David, Klatzky Roberta, Amesur Nikhil, Stetten George
Carnegie Mellon University, Pittsburgh, PA, USA.
Med Image Comput Comput Assist Interv. 2006;9(Pt 1):654-61. doi: 10.1007/11866565_80.
We have derived and evaluated parameters from ultrasound images of the neck to permit a computer to automatically characterize and differentiate between the carotid artery and jugular vein at image acquisition time during vascular interventions, given manually placed seed points. Our goal is to prevent inadvertent damage to the carotid artery when targeting the jugular vein for catheterization. We used a portable 10 MHz ultrasound system to acquire cross sectional B-mode ultrasound images of these great vessels at 10 fps. An expert user identified the vessels in the first frame by touching the vessels on the screen with his fingertip, and the computer automatically tracked the vessels and calculated a best-fit ellipse for each vessel in each subsequent frame. Vessel location and radii were further analyzed to produce parameters that proved useful for differentiating between the carotid artery and jugular vein. These parameters include relative location of the vessels, distension of the vessel walls, and consistent phase difference between the arterial and venous pulsations as determined by temporal Fourier analysis.
我们从颈部超声图像中推导并评估了参数，以便在血管介入过程中图像采集时，给定手动放置的种子点，计算机能够自动识别颈动脉和颈静脉并进行区分。我们的目标是在将颈静脉作为导管插入目标时，防止意外损伤颈动脉。我们使用便携式10兆赫超声系统以每秒10帧的速度获取这些大血管的横截面B型超声图像。一位专业用户通过用指尖触摸屏幕上的血管在第一帧中识别出血管，然后计算机自动跟踪血管并为后续每一帧中的每个血管计算最佳拟合椭圆。进一步分析血管位置和半径以生成有助于区分颈动脉和颈静脉的参数。这些参数包括血管的相对位置、血管壁的扩张以及通过时间傅里叶分析确定的动脉和静脉搏动之间的一致相位差。
