自动ASPECTS评分法在急性缺血性卒中早期影像评估中的应用价值  

The Value of Automated ASPECTS Scoring in Imaging Assessment of Early Ischemic Changes in Acute Ischemic Stroke

作　　者：荆利娜[1] 高培毅[1,2] 杜万良 沈宓[1] 秦海强[3] 克德娜[1] 马国锋[1] 周剑 JING Li-Na;GAO Pei-Yi;DU Wan-Liang;SHEN Mi;QIN Hai-Qiang;KE De-Na;MA Guo-Feng;ZHOU Jian(Department of Radiology,Beijing Tiantan Hospital,Capital Medical University,Beijing 100070,China;Beijing Neurosurgical Institute,Beijing 100070,China;Department of Neurology,Beijing Tiantan Hospital,Capital Medical University,Beijing 100070,China)

机构地区：[1]首都医科大学附属北京天坛医院放射科,北京100070 [2]北京市神经外科研究所 [3]首都医科大学附属北京天坛医院神经内科

出　　处：《中国卒中杂志》2021年第5期463-469,共7页Chinese Journal of Stroke

基　　金：“十三五”国家重点研发计划项目(2016YFC1301604);国家自然科学基金资助项目(81870907)。

摘　　要：目的评价自动ASPECTS评分法在急性缺血性卒中早期影像评估中的实际临床应用价值。方法回顾性分析2020年6-10月在北京天坛医院连续就诊的急性缺血性卒中病例,收集影像归档和通信系统上由RAPID软件评估的基于CT的自动ASPECTS评分,以DWI高信号作为梗死核心标准获得DWI ASPECTS评分,将自动ASPECTS与DWI ASPECTS评分法进行组内相关系数与Cohen’s kappa分析,并计算自动ASPECTS评分法的敏感度、特异度及准确度。结果分析大脑中动脉供血区急性缺血性卒中病例72例,基于分值的自动ASPECTS与DWI ASPECTS的组内相关系数为0.746(P<0.001)。在ASPECTS≥6分与ASPECTS<6分的二分类分析时,自动ASPECTS与DWI ASPECTS评分系统一致性较高(kappa=0.742)。基于区域分布的自动ASPECTS评分法的敏感度、特异度及准确度分别为53.64%、85.99%及73.33%,其中皮层区分别为47.61%、98.47%、78.94%;脑深部区分别为62.68%、67.27%、64.93%。结论基于分值的自动ASPECTS评分法可以评估急性缺血性卒中受累范围的大小。基于区域分布的自动ASPECTS评分法的敏感度、特异度及准确度在每个区域存在差异,因此最终诊断需要结合患者临床症状体征综合判断。Objective To assess the application value of automatic ASPECTS scoring in early imaging assessment of acute ischemic stroke(AIS)in the real clinical setting.Methods This retrospective analysis enrolled consecutive AIS patients admitted to Beijing Tiantan Hospital between 2020 June and 2020 October.The data of automated ASPECTS scores extracted by RAPID software were collected,and the DWI ASPECTS scores were obtained based on the hyperintensities(as infarct core standard)on DWI.The automated ASPECTS and DWI ASPECTS were compared by using interclass correlation coefficient(ICC)and Cohen's kappa analysis,then the sensitivity,specificity and accuracy of automated ASPECTS scoring were calculated.Results Seventy-two cases with AIS in middle cerebral artery territory were included.The ICC between score-based automated ASPECTS and DWI ASPECTS was 0.746(P<0.001).In the dichotomization analysis of ASPECTS≥6 and ASPECTS<6,the two scoring tools showed good agreement(kappa=0.742).The total sensitivity,specificity and accuracy of region-based automatic ASPECTS scoring were 53.64%,85.99%and 73.33%,respectively;those were 47.61%,98.47%and 78.94%in the cortical region,respectively,and those were 62.68%,67.27%,64.93%in the deep brain region,respectively.Conclusions The score-based automated ASPECTS scoring can reflect the size of infarct region in AIS.The sensitivity,specificity and accuracy of region-based automatic ASPECTS scoring varied in different brain regions,thereby the involved region in AIS need to be identified by combing clinical symptoms and signs.

关 键 词：Alberta卒中项目早期CT评分 计算机断层成像 缺血性卒中 机器学习 

分 类 号：R743.3[医药卫生—神经病学与精神病学]