人工智能辅助CTA联合Graeb评分预测颅内动脉瘤破裂出血患者预后的价值  

The Application Value of Artificial Intelligence in Predicting the Prognosis of Patients with Intracranial Aneurysm Rupture and Bleeding Using CTA Combined with Graeb Score

作　　者：王海霞 邰兆琴 许亚春 WANG Haixia;TAI Zhaoqin;XU Yachun(Department of Imaging,Haian People's Hospital,Haian,Jiangsu 226600,China)

机构地区：[1]江苏省海安市人民医院影像科,江苏海安226600

出　　处：《转化医学杂志》2024年第9期1402-1407,共6页Translational Medicine Journal

基　　金：南通市卫生健康委员会科研课题(MS2022100)。

摘　　要：目的 探讨人工智能(AI)辅助CT血管成像(CTA)联合Graeb评分预测颅内动脉瘤(IAN)破裂出血患者预后的价值。方法回顾性选取2021年1月至2024年1月江苏省海安市人民医院实施头颈部CTA及Graeb评分的IAN破裂出血患者142例为研究对象。收集患者一般资料和CTA影像学资料、Graeb评分,并采用AI诊断系统辅助分析。治疗结束后随访3个月,根据IAN破裂再出血及死亡情况,将患者分为预后不良组和预后良好组。计数资料采用χ^(2)检验,计量资料采用独立样本t检验;IAN破裂出血患者预后方面的预测价值采用受试者工作特征(ROC)曲线对AI辅助分析下CTA联合Graeb评分分析;IAN破裂出血患者预后的影响因素采用多因素Logistic回归分析。结果 103例患者预后良好,纳入预后良好组;39例患者术后出现动脉瘤再破裂或死亡,纳入预后不良组。2组年龄、男女比例、糖尿病史、高血压史、IAN的部位及数目比较差异无统计学意义(P>0.05);预后不良组有饮酒史、吸烟史、世界神经外科联盟(WFNS)分级介于Ⅳ~Ⅴ级、Hunt-Hess分级Ⅳ~Ⅴ级的患者比例明显高于预后良好组(P<0.05);预后不良组AR值、SR值、入射夹角、Graeb评分均高于预后良好组(P<0.05)。ROC曲线分析显示,SR值、AR值、入射夹角、Graeb评分对IAN破裂出血的预后均有一定的预测价值,四者联合预测的整体效能更好,曲线下面积(AUC)为0.925,灵敏度和特异度分别为87.2%、92.2%。多因素Logistic回归分析结果显示,有吸烟史、饮酒史、WFNS分级和Hunt-Hess分级为Ⅳ~Ⅴ级以及较高的SR值、入射夹角、AR值和Graeb评分均为IAN破裂出血患者预后不良的危险因素。结论 AI辅助CTA联合Graeb评分能有效地用于IAN破裂出血患者预后的预测。其中AR值、SR值、入射夹角、Graeb评分对IAN破裂出血患者的预后具有一定的预测价值,四者联合预测的整体效能更好。有吸烟史、饮酒史、WFNS分级和Hunt-HObjective To explore the application value of artificial intelligence(AI)in CT angiography(CTA)combined with Graeb score in predicting the prognosis of patients with intracranial aneurysm rupture and bleeding.Methods A total of 142 patients with intracranial aneurysm rupture and bleeding who underwent head and neck CTA and Graeb scoring at Haian People's Hospital from January 2021 to January 2024 were retrospectively selected as the study subjects.Patient information and CTA findings were collected and processed using an AI diagnostic system.After three months of follow-up,patients were divided into a poor prognosis group and a good prognosis group based on aneurysm rerupture and death.χ^(2) test and t-test were used for data analysis.ROC curve analysis was applied to assess the predictive value of CTA combined with Graeb score under AI assistance.Logistic regression analysis was used to explore factors affecting prognosis.Results Thirty-nine patients were included in the poor prognosis group with aneurysm rerupture or death,while the remaining 103 patients were included in the good prognosis group.There were no significant differences in age,male-to-female ratio,history of diabetes,history of hypertension,or the number and location of aneurysms between the two groups(P>0.05).However,the poor prognosis group had higher proportions of smoking and drinking histories,WFNS grade IV–V,and Hunt-Hess grade IV–V compared to the good prognosis group(P<0.05).The AR value,SR value,incident angle,and Graeb score were significantly higher in the poor prognosis group(P<0.05).ROC curve analysis showed that AR value,SR value,incident angle,and Graeb score had predictive value for prognosis,with their combination achieving the best overall performance(AUC=0.925,sensitivity=87.2%,specificity=92.2%).Logistic regression analysis identified smoking history,drinking history,WFNS grade IV–V,Hunt-Hess grade IV–V,and higher AR value,SR value,incident angle,and Graeb score as risk factors for poor prognosis.Conclusion AI-assisted

关 键 词：人工智能 CT血管成像 颅内动脉瘤 动脉瘤 破裂 Graeb评分 预后 

分 类 号：R651.1[医药卫生—外科学]