| 李坤成,邓小元,刘树良.人工神经网络在MRI 脑结构测量诊断Alzheimer 病的进一步应用研究[J].中国医学影像技术,2000,16(12):1029~1031 |
| 人工神经网络在MRI 脑结构测量诊断Alzheimer 病的进一步应用研究 |
| A Further Study of the Application of Artificial Neural Network to Diagnosisof Alzheimer’s Disease with MR Imaging |
| 投稿时间：2000-08-12 |
| DOI： |
| 中文关键词: 人工神经网络 Alzheimer 病 磁共振成像 |
| 英文关键词:Artificial neural network Alzheimer’s disease Magnetic resonance imaging |
| 基金项目: |
|
| 摘要点击次数: 1094 |
| 全文下载次数: 67 |
| 中文摘要: |
| 目的评价人工神经网络在MRI 脑结构测量诊断Alzheimer 病(AD) 的应用价值。方法应用110T
MRI 仪(Siemens Magnetom Impact) 对28 名可能AD 患者(年龄6813 ±717 岁) 和28 名正常人(年龄6816 ±715 岁) 进
行头颅扫描,获取3D GRE 脉冲序列T1WI ,然后在重建图像上对经部分颅内容积标准化处理、大脑的5 个感兴趣区
域(包括杏仁核、海马、内嗅皮层、颞叶和侧脑室颞角) 共10 个指标(分左右) 进行了体积测量研究。使用自编的反馈
式人工神经网络软件与传统统计学处理软件(SPSS) ,同时对测量数据进行分析处理。结果对上述5 个测量指标
的数据分析处理,人工神经网络可将AD 与正常人鉴别开来,并对新个体作出正确的诊断,其诊断的敏感度为97 %、
特异度100 %、准确度达98.5 %; 而应用SPSS 软件进行判别分析时,其诊断的敏感度、特异度和准确度分别为
90.9 %、97 %和93.9 %。结论应用人工神经网络结合MRI 脑结构体积测量是诊断AD 的一种实用而可靠的手段。 |
| 英文摘要: |
| Purpose To evaluate the value of artificial neural network in MR imaging measurement study of brain of Alzheimer’s
disease. Methods Twenty-eight patients with probable AD (age 68. 3 ±7.7year) and 28 normal controls matched with age (age68.6
±7.5year) were studied and the 1. 0T MRI scanner (Siemens Magnetom IMPACT) with 3D GRE pulse sequence was used. Volumetric data of the bilateral amygdalae ,hippocampi ,entorhinal cortices ,temporal lobes and temporal horns of lateral ventricles were obtained by outlining those regions on the serial reformatted images ,and were normalized by partial cranial volume ,then the SPSS software and back propagation network software made by ourselves were used to process and analyze the measured data ,and compared the
results. Results Above-mentioned five indices as discriminant factors ,the artificial neural network could completely differentiate AD
from normal controls ,and new cases were correctly diagnosed ,the sensitivity was 97 % ,specifisity was 100 % ,accuracy was 98. 5 %.
But the sensitivity ,specificity and accuracy were 90. 9 % ,97 %and 93. 9 %respectively in traditional discrimination function analysis
method. This shows that using MRI to measure the brain structures and combining with artificial neural network are possible to make
diagnosis for AD patients. Conclusion Artificial neural network combining with MRI measurement data is probable to become a useful and reliable clinical tool to diagnose AD patients. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
