全科医学临床与教育2024，Vol.22Issue(3) ：208-211,后插1.DOI:10.13558/j.cnki.issn1672-3686.2024.003.005

磁共振薄层扫描结合人工智能脑结构分割技术分析海马体积辅助诊断脑小血管病认知功能障碍

Hippocampus volume by thin layer magnetic resonance scan combined with artificial intelligence brain structure segmentation technology for diagnosis of cognitive dysfunction in cerebral small vessel disease

王含春 汪群芳 罗长国
全科医学临床与教育2024，Vol.22Issue(3) ：208-211,后插1.DOI:10.13558/j.cnki.issn1672-3686.2024.003.005
✕
  • 维普
  • 万方数据

磁共振薄层扫描结合人工智能脑结构分割技术分析海马体积辅助诊断脑小血管病认知功能障碍

Hippocampus volume by thin layer magnetic resonance scan combined with artificial intelligence brain structure segmentation technology for diagnosis of cognitive dysfunction in cerebral small vessel disease

王含春 1汪群芳 2罗长国1
扫码查看

作者信息

  • 1. 321000 浙江金华,金华市第二医院放射科
  • 2. 321000 浙江金华,金华市第二医院老年科
  • 折叠

摘要

目的 分析磁共振薄层扫描结合人工智能脑结构分割技术分析海马体积辅助诊断脑小血管病认知功能障碍的应用价值.方法 选择确诊为脑小血管病患者84例,入院采用简易智力状态检查量表(MMSE)分为认知功能障碍组39例和正常组45例.采用1.43T磁共振薄层扫描结合人工智能脑结构分割技术分析内侧颞叶区和海马的体积绝对值及百分比.结果 认知功能障碍组年龄大于正常组(t=8.63,P＜0.05),内侧颞叶区和海马的体积绝对值及百分比明显低于正常组(t分别=5.86、5.00、6.03、9.63,P均＜0.05),而内侧颞叶萎缩视觉(MTA)评分明显高于正常组(t=-4.75,P＜0.05).相关性分析显示,内侧颞叶区和海马的体积绝对值及百分比与MTA评分呈负相关(r分别=-0.46、-0.50、-0.60、-0.63,P均＜0.05),与 MMSE评分呈正相关(r分别=0.41、0.49、0.57、0.60,P均＜0.05).受试者工作特征曲线(ROC)显示,海马体积百分比预测认知功能障碍的曲线下面积为0.88,95％CI 0.82～0.90,最佳临界值为0.31％,即海马体积百分比＜0.31％诊断认知功能障碍的灵敏度为80.53％,特异度为85.62％.结论 磁共振薄层扫描结合人工智能脑结构分割技术能够精准定位脑功能亚区,通过准确测量海马体积能够辅助诊断脑小血管病的认知功能障碍,海马体积百分比＜0.31％有较好的诊断性能.

Abstract

Objective To analyze the value of hippocampus volume by thin layer magnetic resonance imaging com-bined with artificial intelligence brain structure segmentation technology in the diagnosis of cognitive impairment for cere-bral small vessel disease.Methods A tatal of 84 patients with confirmed cerebral small blood vessel disease were se-lected and divided into cognitive impairment group(n=39)and normal group(n=45)according to mini mental state ex-amination(MMSE)on admission.The absolute volume and percentage of the medial temporal lobe and hippocampus were analyzed using 1.43T magnetic resonance thin layer scan combined with artificial intelligence brain structure seg-mentation technology.Results The age of impairment group was older than normal group(t=8.63,P＜0.05).The abso-lute volume and percentage of the medial temporal lobe and hippocampus in cognitive impairment group were significant-ly lower than normal group(t=5.86,5.00,6.03,9.63,P＜0.05),but the medial temporal lobe atrophy(MTA)score was significantly higher than the normal group(t=-4.75,P＜0.05).Correlation anaylsis showed that the absolute volume and percentage of the medial temporal lobe and hippocampus were negatively correlated with the MTA score(r=-0.46,-0.50,-0.60,-0.63,P＜0.05)and positively correlated with the MMSE score(r=0.41,0.49,0.57,0.60,P＜0.05).Receiver op-erating characteristic curve(ROC)showed that area under curve of hippocampal volume percentage for predicting cogni-tive dysfunction was 0.88(95％CI 0.82-0.90,P＜0.05),with cut-off value of 0.31％,the sensitivity and specificity for diagnosing cognitive dysfunction by hippocampal volume percentage＜0.31％were 80.53％and 85.62％,respectively.Conclusion Thin layer MRI scan combined with artificial intelligence brain structure segmentation technology can accurately locate brain functional subregions,and accurate measurement of hippocam-pal volume can assist in the diagnosis of cognitive dysfunction for cerebral small vessel disease.Achieving cut-off value of hippocampal volume percentage＜0.31％has good diagnostic performance.

关键词

磁共振/人工智能脑结构分割技术/海马/脑小血管病/认知功能障碍/内侧颞叶萎缩视觉

Key words

magnetic resonance imaging/artificial intelligence brain structure segmentation technology/hippocam-pus/cerebral small vessel disease/cognitive dysfunction/medial temporal lobe atrophy

引用本文复制引用

基金项目

金华市科技局项目(2023-4-152)

出版年

2024
全科医学临床与教育
浙江大学

全科医学临床与教育

影响因子：0.63
ISSN：1672-3686
参考文献量21
段落导航相关论文