End-to-end artificial intelligence platform for the management of large vessel occlusions: A preliminary study.
End-to-end artificial intelligence platform for the management of large vessel occlusions: A preliminary study.
复制标题
DOI：
10.1016/j.jstrokecerebrovasdis.2022.106753
复制
发表时间：
2022-09
期刊：
影响因子：
--
通讯作者：
Shujuan Meng;Thi My Linh Tran;Mingzhe Hu;Panpan Wang;T. Yi;Zhusi Zhong;Luoyun Wang;Braden Vogt;Z. Jiao;Arko Barman;U. Çetintemel;Ken Chang;Dat-Thanh Nguyen;Ferdinand K. Hui;I-Yin Pan;Bo Xiao;Li Yang;Hao Zhou;H. Bai
中科院分区：
文献类型：
--
作者：
Shujuan Meng;Thi My Linh Tran;Mingzhe Hu;Panpan Wang;T. Yi;Zhusi Zhong;Luoyun Wang;Braden Vogt;Z. Jiao;Arko Barman;U. Çetintemel;Ken Chang;Dat-Thanh Nguyen;Ferdinand K. Hui;I-Yin Pan;Bo Xiao;Li Yang;Hao Zhou;H. Bai
在这项研究中，我们开发了一种深度学习管道，基于计算机断层血管成像(CTA)图像检测大血管闭塞(LVO)并预测功能结果，以改善LVO患者的管理。方法系列识别器挑选了罗德岛医院2015年至2019年的8650例LVO协议研究，并以确定的薄轴向序列作为数据池。数据被注释为2类：1021个LVO和7629个正常。LVO检测采用的是Inception-V1I3D架构。为了预测结果，323名接受血栓摘除术的患者被选为研究对象。利用三维卷积神经网络(CNN)，以CTA容量和嵌入的治疗前变量作为输入，对30天的MRS结果进行预测。结果在LVO检测模型方面，分析了8,650例患者(中位年龄68岁，四分位数范围(IQR)：58-81；女性3934)的CTA。LVO与NOT的交叉验证AUC为0.74(95%CI：0.72-0.75)。对于MRS分类模型，分析了323例患者(平均年龄75岁，IQR：63-84；女性164例)的CTA。该算法的AUC为0.82(95%CI：0.79-0.84)，灵敏度为89%，特异度为66%。这两个模型随后与医院基础设施集成，其中CTA被实时收集并由该模型处理。结论基于CTA的3D CNN对选择LVO和预测LVO的近期预后是有效的。端到端AI平台可以让用户接收即时的预后预测，方便临床工作流程。
ObjectivesIn this study, we developed a deep learning pipeline that detects large vessel occlusion (LVO) and predicts functional outcome based on computed tomography angiography (CTA) images to improve the management of the LVO patients.MethodsA series identifier picked out 8650 LVO-protocoled studies from 2015 to 2019 at Rhode Island Hospital with an identified thin axial series that served as the data pool. Data were annotated into 2 classes: 1021 LVOs and 7629 normal. The Inception-V1 I3D architecture was applied for LVO detection. For outcome prediction, 323 patients undergoing thrombectomy were selected. A 3D convolution neural network (CNN) was used for outcome prediction (30-day mRS) with CTA volumes and embedded pre-treatment variables as inputs.ResultFor LVO-detection model, CTAs from 8,650 patients (median age 68 years, interquartile range (IQR): 58-81; 3934 females) were analyzed. The cross-validated AUC for LVO vs. not was 0.74 (95% CI: 0.72-0.75). For the mRS classification model, CTAs from 323 patients (median age 75 years, IQR: 63-84; 164 females) were analyzed. The algorithm achieved a test AUC of 0.82 (95% CI: 0.79-0.84), sensitivity of 89%, and specificity 66%. The two models were then integrated with hospital infrastructure where CTA was collected in real-time and processed by the model. If LVO was detected, interventionists were notified and provided with predicted clinical outcome information.Conclusion3D CNNs based on CTA were effective in selecting LVO and predicting LVO mechanical thrombectomy short-term prognosis. End-to-end AI platform allows users to receive immediate prognosis prediction and facilitates clinical workflow.