中国组织工程研究 ›› 2021, Vol. 25 ›› Issue (24): 3796-3803.doi: 10.12307/2021.082

• 数字化骨科 digital orthopedics • 上一篇    下一篇

基于代谢组学腰椎间盘退变的计算机辅助诊断

江丽红1,吴晓锋1,欧阳林2,3,罗爱芳2,3,黄  丽2,3   

  1. 1闽南师范大学数学与统计学院,福建省漳州市   363000;2厦门大学医学院医学影像研究所,福建省漳州市   363000;3解放军第九〇九医院医学影像科,福建省漳州市   363000
  • 收稿日期:2020-07-27 修回日期:2020-07-29 接受日期:2020-08-25 出版日期:2021-08-28 发布日期:2021-03-05
  • 通讯作者: 吴晓锋,教授,博士生导师,闽南师范大学数学与统计学院,福建省漳州市 363000 欧阳林,博士,主任医师,厦门大学医学院医学影像研究所,福建省漳州市 363000;解放军第九〇九医院医学影像科,福建省漳州市 363000
  • 作者简介:江丽红,女,1996年生,广东省广州市人,硕士,主要从事机器学习与健康医疗大数据研究。
  • 基金资助:
    福建省科技计划项目(2019Y31010067),项目负责人:欧阳林;第九〇九医院青年苗圃基金(18Y021),课题负责人:罗爱芳

Computer aided diagnosis of lumbar disc degeneration based on metabolomics

Jiang Lihong1, Wu Xiaofeng1, Ouyang Lin2, 3, Luo Aifang2, 3, Huang Li2, 3   

  1. 1School of Mathematics and Statistics, Minnan Normal University, Zhangzhou 363000, Fujian Province, China; 2Institute of Medical Imaging, School of Medicine, Xiamen University, Zhangzhou 363000, Fujian Province, China; 3Department of Medical imaging, PLA 909th Hospital, Zhangzhou 363000, Fujian Province, China
  • Received:2020-07-27 Revised:2020-07-29 Accepted:2020-08-25 Online:2021-08-28 Published:2021-03-05
  • Contact: Wu Xiaofeng, Professor, Doctoral supervisor, School of Mathematics and Statistics, Minnan Normal University, Zhangzhou 363000, Fujian Province, China Ouyang Lin, MD, Chief physician, Institute of Medical Imaging, School of Medicine, Xiamen University, Zhangzhou 363000, Fujian Province, China; Department of Medical imaging, PLA 909th Hospital, Zhangzhou 363000, Fujian Province, China
  • About author:Jiang Lihong, Master, School of Mathematics and Statistics, Minnan Normal University, Zhangzhou 363000, Fujian Province, China
  • Supported by:
    the Science and Technology Project of Fujian Province, No. 2019Y31010067 (to OL); the Fund of the 909th Hospital Youth Nursery, No. 18Y021 (to LAF) 

摘要:

文题释义:
MRI:是利用原子核在强磁场内发生共振产生的信号绘制出人体内部立体图像的技术,可对人体某个部位进行各种角度和各种平面成像,临床医生可根据腰椎间盘的磁共振成像评价腰椎间盘退变等级。MRI定量技术可定量检测腰椎间盘的多种生化代谢物,针对目前临床对腰椎间盘退变等级诊断有一定困难,文章尝试基于MRI检测的生化代谢指标对腰椎间盘退变进行诊断。
机器学习:是一种实现人工智能的方法,包括逻辑回归、支持向量机、神经网络、决策树、朴素贝叶斯等多种算法,可在大量样本中利用算法发现数据规律并训练分类器,在得到新的样本时利用训练好的分类器对样本进行分类和预测。实验利用收集到的腰椎间盘MRI代谢数据训练了3种机器学习分类器,为腰椎间盘退变等级诊断提供了新方法。

背景:腰椎间盘退变诊断对预防腰椎疾病意义重大,但目前对其诊断主要依赖于影像医师的主观评价,易因个人经验不足产生误判。
目的:建立自动识别腰椎间盘退变等级的计算机辅助诊断方法,为影像医师提供参考。
方法:采用Spearman相关分析验证腰椎间盘的MRI代谢指标与腰椎间盘退变的Pfirrmann等级相关性,并建立可用于腰椎间盘退变智能诊断的Softmax回归、神经网络和支持向量机等多种分类器。
结果与结论:相关性分析结果表明,椎间盘相邻上下位椎体脂肪分数FF值和T2*值等3种生化代谢指标都与腰椎间盘退变显著相关,softmax回归、神经网络和支持向量机3种诊断模型的分类准确率分别为0.477,0.515和0.523,kappa系数分别为0.311,0.300和0.330。实际分析结果表明,采用MRI代谢指标建立腰椎间盘退变智能辅助诊断是可行的,为腰椎间盘退变诊断提供了一种可期的途径。
https://orcid.org/0000-0002-7568-9741 (江丽红) 

中国组织工程研究杂志出版内容重点:人工关节;骨植入物;脊柱;骨折;内固定;数字化骨科;组织工程

关键词: 骨, 腰椎间盘退变, MRI, Pfirrmann等级, Spearman相关分析, softmax回归, 神经网络, 支持向量机, 代谢指标

Abstract: BACKGROUND: The diagnosis of lumbar disc degeneration is of great significance for the prevention of lumbar disease, and the diagnosis of lumbar disc degeneration mainly relies on the subjective evaluation of the imaging physician, which is likely to misjudge because of insufficient experience. 
OBJECTIVE: To propose a computer-aided diagnosis technique for classification on the lumbar disc degeneration, and to provide reference for imaging doctors. 
METHODS: Spearman correlation analysis is used to verify the correlation between magnetic resonance imaging metabolic indices of lumbar intervertebral disc and the Pfirrmann grades of lumbar disc degeneration. Several classifiers for the intelligent diagnosis of lumbar disc degeneration are developed by means of machine learning strategies such as the Softmax regression, the neural network and the support vector machine. 
RESULTS AND CONCLUSSION: The result of correlation analysis showed that three metabolic indices such as fat fraction (FF) of adjacent upper and lower vertebral bodies of degenerative disc, T2* values were significantly correlated with lumbar disc degeneration. The classification accuracy of the softmax regression, the neural network and the support vector machine respectively was 0.477, 0.515 and 0.523, and kappa’s coefficient of these three diagnostic models was 0.311, 0.300 and 0.330, respectively. The actual analysis indicates that it is feasible to establish a computer-aided intelligent diagnosis of lumbar disc degeneration by using the MRI metabolic indices, showing a promising approach for the diagnosis of lumbar disc degeneration.

Key words: bone, lumbar disc degeneration, MRI, Pfirrmann level, Spearman correlation analysis, softmax regression, neural network, support vector machine, metabolic indices

中图分类号: