A novel segmentation model for medical images with intensity inhomogeneity based on adaptive perturbation
A novel segmentation model for medical images with intensity inhomogeneity based on adaptive perturbation
复制标题
DOI：
10.1007/s11042-018-6735-5
复制
发表时间：
2018-10
影响因子：
3.6
通讯作者：
Haiping Yu;Fazhi He;Yiteng Pan
中科院分区：
文献类型：
--
作者：
Haiping Yu;Fazhi He;Yiteng Pan
在医学领域，由于低对比度、复杂噪声和灰度不均匀性等原因，对医学图像进行精确分割一直是一个难题。为了克服这些障碍，本文提出了一种新的基于边缘的活动轮廓模型（ACM）的医学图像分割。具体地说，提出了一种精确的正则化方法来保持水平集函数具有符号距离性质，从而保证了演化曲线的稳定性和数值计算的准确性。更重要的是，自适应扰动集成到框架的边缘为基础的ACM。扰动技术可以平衡曲线演化的稳定性和分割的准确性，是分割灰度不均匀医学图像的关键。人工和真实的医学图像的实验表明，所提出的分割模型优于国家的最先进的方法在鲁棒性噪声和分割精度。
In medical field, it remains challenging to accurately segment medical images due to low contrast, complex noises and intensity inhomogeneity. To overcome these obstacles, this paper provides a novel edge-based active contour model (ACM) for medical image segmentation. Specifically, an accurate regularization approach is presented to maintain the level set function with a signed distance property, which guarantees the stability of the evolution curve and the accuracy of the numerical computation. More significantly, an adaptive perturbation is integrated into the framework of the edge-based ACM. The perturbation technique can balance the stability of curve evolution and the accuracy of segmentation, which is key for segmenting medical images with intensity inhomogeneity. A number of experiments on both artificial and real medical images demonstrate that the proposed segmentation model outperforms state-of-the-art methods in terms of robustness to noise and segmentation accuracy.