椭球先验约束的前列腺磁共振图像分割

发布时间：2018-03-12 10:32

  本文选题：磁共振前列腺分割　切入点：图像配准　出处：《南方医科大学学报》2017年03期 　论文类型：期刊论文

【摘要】：目的为了有效的利用图谱的先验信息和待分割图像的灰度信息,提出一种新的椭球先验约束下的前列腺MR图像多图谱分割算法。方法将多图谱分割与椭球形状先验相结合,在多图谱分割过程中引入椭球先验知识,针对椭球先验约束下的前列腺感兴趣区域进行图谱选择,大大避免了前列腺周围组织与器官对图谱选择造成的干扰;其次,在图谱融合过程中加入椭球先验项进行约束,对通过配准技术引入的前列腺图谱形状先验进行校正和补偿,有效避免了由配准误差引起的错误分割的情况。结果对50例前列腺MR图像进行分割实验,实验结果表明该算法对前列腺数据的分割精度均在80%以上,平均精度提高到了88.12%。结论椭球先验约束的前列腺MR图像多图谱分割算法稳定有效,分割结果精确度高。
[Abstract]:Objective to effectively utilize the priori information of the map and the gray level information of the image to be segmented, a new multi-map segmentation algorithm for prostate Mr images with ellipsoidal priori constraints is proposed, which combines multi-spectrum segmentation with a priori ellipsoid shape. The priori knowledge of ellipsoid is introduced in the process of multi-spectrum segmentation to select the region of interest of prostate under the restriction of ellipsoid priori, which greatly avoids the interference of tissues and organs around the prostate to the selection of the map. A priori term of ellipsoid is added in the process of map fusion to correct and compensate the priori of prostate map shape introduced by registration technique. Results 50 cases of prostate Mr images were segmented by experiments. The experimental results show that the accuracy of the algorithm for prostate data segmentation is more than 80%. Conclusion the multi-map segmentation algorithm with ellipsoid priori constraint is stable and effective, and the segmentation accuracy is high.
【作者单位】： 南方医科大学生物医学工程学院;广东省医学图像处理重点实验室;
【基金】：国家自然科学基金(61471188) 广东省科技计划项目(2015B010106008,2015B01013 1011,2014B030301042) 广东省自然科学基金(2014A030313316,2016A030313574)~~
【分类号】：R445.2;R697.3;TP391.41
，

本文编号：1601172