一种基于主动轮廓模型的MRI医学图像序列边缘提取算法

A new edge extraction algorithm of active contour model for magnetic resonance imaging

为了提取核共振成像图像序列的边缘,提出了一种改进主动轮廓模型的边缘提取算法。通过调整原始公式的一些参数使得该模型不但能精确地提取图像中的凸形物体的边缘,而且能够接近边缘的凹陷处;引入自适应改变大小的外部约束能量来增大外能的吸引范围,使控制点能够不依赖于初始轮廓而快速地收敛到目标的真实轮廓;结合匹配技术,提高边缘提取结果层间传递的精度。实验结果表明该算法仅需少量用户交互就能快速准确地从医学图像序列中提取出感兴趣的物体边缘。

In order to extract the edge of the Magnetic Resonance Imaging（MRI） series, a new algorithm based on an improved active contour model was proposed. It modified some terms of the initial formula, which made the snake model not only precisely extract the edge of the protruding object in images but also approach to the concave part of the edge. It introduced exterior sanction energy, which could adaptively change its value to enlarge suction, so that control points could quickly eonstringe to the real contour without relying on the initial contour. Combined with pattern matching, this new method could increase the precision of transferring the result of edge extraction from one slice to another. Experiments show that this algorithm can obtain the edge of the desired object from a series of medical images faster and more accurately than before, and with only little user interaction.