基于多尺度梯度矢量场GAC模型的MR医学图像分割

MR Image Segmentation Based on GAC Model with Multiscale Gradient Vector Field

医学图像分割是图像分割技术的一个重要应用领域,GAC(测地线活动轮廓)模型是基于PDE(偏微分方程)方法中一种常用的图像分割模型,使用这种模型时,如何选择合适的平滑尺度是影响分割效果的重要因素之一。提出了一种基于多尺度梯度矢量场GAC模型图像对象轮廓提取的MR图像分割方法,用多尺度梯度矢量取代GAC模型中单一尺度下平滑图像的梯度矢量,提高了GAC模型的收敛速度,有效地改善了局部极小值问题。实验结果验证了该方法的有效性。

PDE （partial differential equation） based GAC （geodesic active contour） model is useful for medical image segmentation. To improve the local minimum problem of GAC model, usually we need smooth the image with a proper smoothness scale before we acquire the gradients of the image. But it is difficult to choose the smoothness scale as we usually do in acquiring the gradients of the approximating image by smoothening it with a single scale. In order to overcome this drawback, multi-scale gradient vector field is used instead of single-scaled gradient vector of images in GAC model. The multi-scale gradient vector field, which can be obtained by updating the gradient vector for each position of the image from lower to higher levels resolution, is still smooth enough in the whole image and accurate for the main edges of the image. The experimental results show that this improved GAC model is effective for MRI（ magnetic resonance imaging） segmentation.