人工智能在基于CT灌注原始图像重组头颅CT血管成像的应用

Application value of artificial intelligence on reconstruction of cerebral CT angiography based on CT perfusion datasets

目的:探讨利用人工智能软件从CTP原始图像中重组的CTA的临床应用价值。方法:收集临床疑似缺血性脑血管病患者60例,所有患者均进行头颅CT平扫、CTA与CTP检查,将CT灌注原始图像导入脑灌注智能分析系统获得重组CTA(即CTPA)。以常规CTA扫描结果为参照,CTA与CTPA组各分析540个节段血管,采用Kappa检验比较两者图像质量以及脑动脉狭窄显示的一致性。结果:CTPA图像显示脑动脉强化程度、脑动脉细节显示能力、图像噪声及静脉污染等与CTA相似。CTA和CTPA对脑内动脉显示具有很好的一致性,Kw值均大于0.75。共540个节段血管中,仅15个节段的血管存在CTPA与CTA显示不一致的结果。结论:利用AI软件,能够方便、快捷地迅速从CTP灌注原始图像中重组出血管图像,所得CTPA与常规CTA比较,有较好的一致性,能准确显示动脉狭窄和闭塞。

Objective:To explore the application value of reconstructed CT angiography(CTA)derived from CTP datasets by using artificial intelligence(Al)software.Methods:Sixty patients with clinically suspected ischemic cerebrovascular disease were enrolled.All cases were examined by cerebral CT,CT angiography(CTA)and CT perfusion(CTP).The CT perfusion data of all patients were imported into the brain perfusion intelligent analysis system to obtain angiography(hereafter called CTPA).We analyzed 540 segmental vessels in CTA and CTPA groups,reference as conventional CTA.The Kappa test was used to compare the consistency of image quality and cerebral artery stenosis between two groups.Results:CTPA group was similar to CTA group in terms of cerebral artery enhancement,arteriole detail display ability,image noise and venous pollution.Kw val-ues were>0.75 for the consistency test between CTPA group and CTA group on the display of cerebral artery stenosis.Only 15 segments of the 540 segmental vessels were inconsistent between CTPA and CTA.Conclusion:By using AI software,the cerebral angiography can be reconstructed conveniently and quickly from CTP datasets.The CTPA can accurately display arte-rial stenosis and occlusion compared with the CTA.