期刊文献+

一种基于超像素和改进U-net的多模态脑部肿瘤图像分割方法 被引量:5

Brain MRI Tumor Segmentation Method Based on Superpixel and Mean Shift
下载PDF
导出
摘要 针对多模态脑部肿瘤图像分割难度较大和对脑部肿瘤边缘区域的分割不足等问题,本文将多模态脑部肿瘤图像分割任务分成两部分解决.第一部分是对脑部肿瘤轮廓区域的分割,先用超像素分割算法对图像进行预处理简化图像的表示形式,再提取每个超像素区域的灰度直方图,通过皮尔逊相关系数计算每个超像素区域的相似度,最后用均值漂移算法对剩余的直方图进行迭代运算,完成对脑部肿瘤图像轮廓区域的分割.通过在2D脑部肿瘤图像LGG数据集上的大量实验分析,本文的肿瘤轮廓分割模型可以很好的分割出肿瘤轮廓.第二部分用本文改进的U-net算法对脑部肿瘤图像轮廓区域进行精细的多模态分割.在多模态脑部肿瘤图像数据库Brats2019进行大量的实验,结果表示本文算法能够很好的细分出脑部肿瘤区域. In order to solve the problems such as the difficulty of multimodal brain tumor image segmentation and the insufficient segmentation of brain tumor edge region, this paper divides the multimodal brain tumor image segmentation task into two parts.The first part is the segmentation of brain tumor contour region.Firstly, the image is preprocessed by the super-pixel segmentation algorithm to simplify the representation of the image.Then the gray histogram of each super-pixel region is extracted.The similarity of each super-pixel region is calculated by Pearson correlation coefficient.Finally, the mean shift algorithm is used to iterate the remaining histogram to complete the brain tumor image Segmentation of image contour region.Through a large number of experiments on 2 D brain tumor image LGG data set, the tumor contour segmentation model in this paper can segment the tumor contour well.In the second part, we use the improved u-net algorithm to segment the contour region of brain tumor image.A large number of experiments are carried out in the multimodal brain tumor image database brats2019,and the results show that the algorithm can segment brain tumor regions well.
作者 胡春燕 司明明 陈玮 HU Chun-yan;SI Ming-ming;CHEN Wei(University of Shanghai for Science and Technology,College of Photoelectric Information and Computer Engineering,Shanghai 200093,China)
出处 《小型微型计算机系统》 CSCD 北大核心 2022年第1期91-97,共7页 Journal of Chinese Computer Systems
基金 国家自然科学青年基金项目(61703277)资助。
关键词 脑肿瘤 MRI图像分割 超像素 均值漂移 3D U-net brain tumor MRI image segmentation superpixel mean shift 3D U-net
  • 相关文献

参考文献10

二级参考文献67

共引文献93

同被引文献30

引证文献5

二级引证文献5

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部