人工智能在破裂或先兆破裂颅内多发动脉瘤患者责任动脉瘤判定及治疗中的应用

Application of artificial intelligence for determining responsible aneurysms and developing therapies in patients with multiple intracranial aneurysms presenting with subarachnoid hemorrhage or sentinel headache

目的探究人工智能(AI)在破裂或先兆破裂颅内多发动脉瘤患者的责任动脉瘤判定及治疗策略制定中的价值。方法回顾性分析2019年10月至2021年10月苏州大学附属第一医院神经外科收治并使用AI进行动脉瘤破裂风险评估的20例破裂或先兆破裂颅内多发动脉瘤患者的临床资料。根据有无明确的出血来源及出血倾向性分为出血明确组、出血不明确组和疑似先兆破裂组。根据AI评估结果判断责任动脉瘤的位置,并制定治疗策略。出院时采用改良Rankin量表评分(mRS)评估患者的临床结局。结果20例患者中,出血明确组5例,出血不明确组9例,疑似先兆破裂组6例,共52个动脉瘤。出血明确组和出血不明确组开颅手术中确认的责任动脉瘤均与AI判断结果一致。疑似先兆破裂组患者依据AI的评估结果优先处理较高风险的颅内动脉瘤,出院时mRS均为0~1分。结论对于破裂或先兆破裂颅内多发动脉瘤,AI能够有效、准确地判断责任动脉瘤的位置。AI评估结果能够辅助临床医生制定合理的治疗策略。

Objective To explore the application value of artificial intelligence(AI)for identification of responsible aneurysms and providing therapeutic strategies in patients with multiple intracranial aneurysms(MIAs)who had symptoms of subarachnoid hemorrhage(SAH)or sentinel headache.Methods From October 2019 to October 2021,20 patients with MIAs presenting with SAH or sentinel headache were admitted to the Department of Neurosurgery,the First Affiliated Hospital of Soochow University whose clinical data were retrospectively analyzed in this study.The risk of aneurysm rupture was assessed in all patients by AI.According to the hemorrhage pattern,20 patients were divided into definitive hemorrhage pattern group,non-definitive hemorrhage pattern group,and sentinel headache group.The responsible aneurysm and individualized therapeutic strategies were determined based on the evaluation results of AI.The clinical outcomes of patients were assessed at discharge based on the modified Rankin Scale score(mRS).Results Of the 20 patients,5 were in definitive hemorrhage pattern group,9 in non-definitive hemorrhage pattern group,and 6 in sentinel headache group.All the responsible aneurysms identified by the AI were confirmed during craniotomy in definitive and non-definitive hemorrhage pattern groups.The development of individualized therapeutic strategies in sentinel headache group was based on the evaluation results and all patients were discharged with the mRS of 0-1.Conclusions AI can effectively and accurately identify the responsible aneurysms in patients with MIAs presenting with SAH or sentinel headache.The evaluation results of AI can assist the clinicians in developing reasonable individualized therapeutic strategies.