摘要
肾脏内部血流量占肾脏总面积大小是医学中评价肾脏损伤等级的重要指标,利用数字图像处理技术处理肾脏重症超声并构建肾脏损伤等级评估模型具有一定现实意义,为此提出了一种基于特征匹配的超声影像评估仿真。在量化肾脏超声血流样本条的同时利用颜色通道值匹配和图像分割方法快速对血流区域成像;再利用滤波算法对成像结果细化,在此基础上结合样本条的物理意义估计血流总量。实验结果表明,本算法能很好地估计肾脏重症超声中的血流总量,不仅为肾脏损伤等级自动化评估提供了新方案,而且降低了现有因医者经验判断方法而造成的损失。与此同时,算法耗时短,可靠性强。
It is an important indicator to measure the level of kidney injury in medicine that the blood flow in the kidney accounts for the total area of the kidney.Using digital image processing technology to process kidney severe ultrasound and construct a model of kidney injury level assessment makes practical significance.For this reason,a simulation research on ultrasound image evaluation based on feature matching is proposed in this paper.In order to extract efficient imaging on blood flow regions,a color channel value matching and image segmentation method was adopted while quantifying renal ultrasound blood flow sample strips.Then,the extracted region was refined according to the filtering algorithm.At the same time,the total blood flow was approximated by combining the corresponding physical value with the sample strips.Experiment results demonstrate that the proposed algorithm can extract and estimate the total blood flow in severe renal ultrasound efficiently,which not only provides a new solution for the automated evaluation of kidney injury levels,but also reduces the loss caused by the existing doctors’ judgment from prior experience.At the same time,it costs less and reliable.
作者
何立风
周广彬
雷涛
杨梅梅
HE Li-feng;ZHOU Guang-bin;LEI Tao;YANG Mei-mei(School of Electronic Information and Artificial Intelligence,Shaanxi Univ,of Science and Technology,Xi’an Shanxi 710021,China;Faculty of Information Science and Technology,Aichi Prefectural Univ.,Aichi 480-1198,Japan;School of Electronic Engineering,Xi’an Univ.of Posts and Telecommunications,Xi'an Shanxi 710000,China)
出处
《计算机仿真》
北大核心
2021年第8期402-406,共5页
Computer Simulation
基金
国家自然科学基金面上项目(61971272)
国家自然科学基金青年基金项目(61603234,61601271)。
关键词
超声影像
特征提取
肾脏损伤
图像分割
滤波
Ultrasound image
Feature extraction
Kidney injury
Image segmentation
Filtering
