摘要
目的 分析人工智能(artificial intelligence,AI)与医师判读头颈部血管CTA征象的一致性,评价AI对头颈部血管病变的诊断能力。方法 选取2023年1—11月武汉科技大学附属老年病医院医学影像信息系统(picture archiving and communication system,PACS)系统中进行头颈部计算机断层血管造影(computed tomography angiography,CTA)和头颈部数字减影血管造影(digital subtraction angiography,DSA)检查的106例脑血管疾病患者为研究对象,分别由2名放射科医师与AI自动分析软件对CTA检查进行阅片和诊断。以DSA检查结果为标准,比较判读与高年资医师判读结果的差异。结果在CTA结果判断中,AI对血管狭窄诊断的准确率为87.21%(75/86),低于医师判断的96.51%(83/86),差异有统计学意义(χ^(2)=4.976,P=0.026)。医师判读与AI自动判读诊断动脉瘤的准确率比较,差异无统计学意义(P>0.05)。在具体CTA征象判断方面,医师判断与AI判读对血管狭窄程度、斑块范围、斑块形态判断率比较,差异无统计学意义(P均>0.05)。医师判断与AI判读动脉瘤大小、形态、瘤颈情况以及动脉瘤有无子囊比较,差异无统计学意义(P均>0.05)。结论 AI判读头颈部血管CTA征象的能力接近医师水平,但对动脉斑块的鉴别能力较低,可用于辅助放射科医师的阅片。
Objective To analyze the consistency of artificial intelligence(AI) and physicians' interpretation of vascular CTA signs in head and neck,and evaluate the diagnostic ability of AI for vascular diseases in head and neck.Methods A total of 106 patients with cerebrovascular diseases who underwent head and neck computed tomography angiography(CTA) and digital subtraction angiography(DSA) examinations in the picture archiving and communication system(PACS) of Geriatric Hospital Affiliated to Wuhan University of Science and Technology from January to November 2023 were selected as the research subjects,the CTA examinations were reviewed and diagnosed by two radiologists and AI automatic analysis software respectively.The results of DSA examination were used as the standard to compare the differences between AI and senior physicians.ResultsI n the judgment of CTA results,the accuracy of AI in the diagnosis of vascular stenosis was 87.21%(75/86),which was lower than 96.51%(83/86) judged by physicians,and the difference was statistically significan(χ~2=4.976,P=0.026).There was no significant difference in the accuracy of aneurysm diagnosis between interpretation and AI automatic interpretation(P>0.05).In the judgment of specific CTA signs,there was no significant difference in the degree of vascular stenosis,plaque range and plaque morphology between the physician's judgment and AI interpretation(all P>0.05).There was no significant difference in the size,shape,neck of the aneurysm and the presence or absence of the ascus of the aneurysm between the judgment of the physician and the AI(all P>0.05).Conclusion AI's ability to interpret CTA signs of head and neck vessels is close to that of physicians,but its ability to distinguish arterial plaque is low,and it can be used to assist radiologists in reading radiographs.
作者
谢镜
吴磊
XIE Jing;WU Lei(School of Public Health,School of Medicine,Wuhan University of Science and Technology,Wuhan 430065,Hubei,China;Department of Interventional Medicine,Geriatric Hospital Affiliated to Wuhan University of Science and Technology,Wuhan 430071,Hubei,China)
出处
《世界复合医学(中英文)》
2024年第10期90-93,共4页
World Journal of Complex Medicine
关键词
人工智能
计算机断层血管造影
血管狭窄
动脉瘤
Artificial intelligence
Computed tomography angiography
Vascular stenosis
Aneurysm
