摘要
骨密度是判断骨骼强度的一个重要指标,事关骨头的健康程度.通过对云南省某三甲医院200例骨密度检测报告进行分析,对比观察骨头图像发现骨头黑色像素占比与骨密度值成正相关,并通过数据验证了此结论.分别求得灰度值(0~255)与骨密度相关系数,选取正相关性最强的灰度区间0~7与负相关最强的灰度区间183~218作为预测骨密度值的相关因素,基于BP神经网络构建预测函数,并比较其拟合结果,选取拟合相关系数大的区间作为预测区间并选定此区间的预测函数.然后,基于信息熵,构建了骨密度值与骨质状况的关系模型,获得了判定骨质状况的骨密度区间,并用大量的样本验证了此模型的科学性.最后对20组样本进行骨密度值及骨质状况的预测,得到成功预测骨质状况的概率为90%.
Bone mineral density, an important indicator of bone strength, is related to the health of bones. This paper analyzes the bone mineral density test report of 200 cases in a top three hospital in Yunnan Province. By comparing the bone images, it was found that the proportion of bone black pixels is positively correlated with the bone density value, and the conclusion is verified by data. The correlation coefficient between gray value(0-255) and bone density was obtained, and the gray region 0-7 with the strong positive correlation and the gray interval 183-218 with the strong correlation were selected as the correlation factors for predicting the bone density value. Based on BP neural network, the prediction function was constructed, and the fitting result was compared. The interval with large correlation coefficient was selected as the prediction interval and the prediction function of this interval was selected. Then, based on the information entropy, the relationship model between bone density and bone status was constructed. The bone density interval for determining the bone condition was obtained. The scientificity of the model was verified by a large number of samples. Finally, 20 groups of samples were predicted for bone mineral density and bone status, and the probability of successfully predicting bone status was 90%.
作者
王钦
徐建新
刘超
WANG Qin;XU Jianxin;LIU Chao(Institute of Quality Development,Kunming University of Science and Technology,Kunming 650093,China;The First People's Hospital of Yunnan Province,Kunming 650032,China)
出处
《昆明理工大学学报(自然科学版)》
CAS
北大核心
2019年第6期96-104,共9页
Journal of Kunming University of Science and Technology(Natural Science)
基金
云南省万人计划项目(KKRD201908105)
关键词
骨密度
图像灰度
神经网络
关系模型
图像熵
bone mineral density
image gray level
neural network
relationship model
image entropy