基于多参数配准模型的脑核磁影像分割算法

发布时间：2018-04-15 06:12

  本文选题：图像分割 + 图像配准　； 参考：《电子学报》2017年09期

【摘要】：配准技术在基于多图谱的分割方法中能有效地将医学图谱的先验知识融入分割过程,再结合以高效的标记融合算法,最终实现精确地自动分割.针对图谱配准的较大误差及其对标记融合的重要影响,本文建立了一种新的概率图模型框架并以此提出了基于多参数配准模型的分割算法,将此方法与高效的标记融合算法相结合,可以提高目标图像中特定组织区域的分割精度,更使其在少量图谱分割的情形下具有重要应用.首先,使用多种配准参数对所有目标图像进行配准;然后,分别采用不同的算法对配准图像进行灰度融合和标记融合,实现训练图像的重构过程;最后,利用高效的标记融合算法对重构后的图像进行融合得到最终精确的分割结果.实验结果表明该方法均优于本文其他分割算法,能够有效提升脑部组织分割精度.
[Abstract]:Registration technology can effectively integrate the priori knowledge of medical atlas into the segmentation process in the multi-atlas based segmentation method, and then combine the efficient label fusion algorithm to achieve accurate automatic segmentation.In view of the large error of map registration and its important influence on marker fusion, a new probabilistic graph model framework is established and a segmentation algorithm based on multi-parameter registration model is proposed.The combination of this method and the efficient label fusion algorithm can improve the segmentation accuracy of the specific tissue areas in the target image, and make it more important in the case of a small number of map segmentation.First, we use a variety of registration parameters to register all the target images. Then, we use different algorithms to carry out gray level fusion and label fusion to realize the reconstruction process of the training image.An efficient label fusion algorithm is used to fuse the reconstructed images to obtain accurate segmentation results.The experimental results show that the proposed method is superior to other algorithms and can effectively improve the accuracy of brain tissue segmentation.
【作者单位】： 上海电力学院自动化工程学院;国网浙江省电力公司金华供电公司;
【基金】：国家自然科学基金(No.61203224) 上海市教育委员会创新项目(No.13YZ101)
【分类号】：R318;TP391.41
，

本文编号：1752863