Deep learning-assisted identification and quantification of aneurysmal subarachnoid hemorrhage in non-contrast CT scans: Development and external validation of Hybrid 2D/3D UNet
Ping Hu,
Haizhu Zhou,
Tengfeng Yan,
Hongping Miu,
Feng Xiao,
Xinyi Zhu,
Lei Shu,
Shuang Yang,
Ruiyun Jin,
Wenlei Dou,
Baoyu Ren,
Lizhen Zhu,
Wanrong Liu,
Yihan Zhang,
Kaisheng Zeng,
Minhua Ye,
Shigang Lv,
Miaojing Wu,
Gang Deng,
Rong Hu,
Renya Zhan,
Qianxue Chen,
Dong Zhang,
Xingen Zhu
Affiliations
Ping Hu
Department of Neurosurgery, The Second Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006, China; Jiangxi Key Laboratory of Neurological Tumors and Cerebrovascular Diseases, Nanchang, Jiangxi 330006, China; Jiangxi Health Commission Key Laboratory of Neurological Medicine, Nanchang, Jiangxi 330006, China; Institute of Neuroscience, Nanchang University, Nanchang, Jiangxi 330006, China
Haizhu Zhou
School of Physics and Technology, Wuhan University, Wuhan, Hubei 430060, China
Tengfeng Yan
Department of Neurosurgery, The Second Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006, China; Jiangxi Key Laboratory of Neurological Tumors and Cerebrovascular Diseases, Nanchang, Jiangxi 330006, China; Jiangxi Health Commission Key Laboratory of Neurological Medicine, Nanchang, Jiangxi 330006, China; Institute of Neuroscience, Nanchang University, Nanchang, Jiangxi 330006, China
Hongping Miu
Department of Neurosurgery, Southwest Hospital, Third Military Medical University (Army Medical University), Chongqing 400038, China
Feng Xiao
Department of Neurosurgery, First Affiliated Hospital, School of Medicine, Zhejiang University, Hangzhou, Zhejiang 310003, China,
Xinyi Zhu
Department of Neurosurgery, Renmin Hospital of Wuhan University, Wuhan, Hubei 430060, China
Lei Shu
Department of Neurosurgery, The Second Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006, China; Jiangxi Key Laboratory of Neurological Tumors and Cerebrovascular Diseases, Nanchang, Jiangxi 330006, China; Jiangxi Health Commission Key Laboratory of Neurological Medicine, Nanchang, Jiangxi 330006, China; Institute of Neuroscience, Nanchang University, Nanchang, Jiangxi 330006, China
Shuang Yang
School of Physics and Technology, Wuhan University, Wuhan, Hubei 430060, China
Ruiyun Jin
Department of Neurosurgery, The Second Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006, China
Wenlei Dou
Department of Neurosurgery, The Second Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006, China
Baoyu Ren
Department of Neurosurgery, The Second Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006, China
Lizhen Zhu
Department of Neurosurgery, The Second Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006, China
Wanrong Liu
Department of Neurosurgery, The Second Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006, China
Yihan Zhang
Department of Neurosurgery, The Second Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006, China
Kaisheng Zeng
Department of Neurosurgery, The Second Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006, China
Minhua Ye
Department of Neurosurgery, The Second Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006, China
Shigang Lv
Department of Neurosurgery, The Second Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006, China
Miaojing Wu
Department of Neurosurgery, The Second Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006, China
Gang Deng
Department of Neurosurgery, Renmin Hospital of Wuhan University, Wuhan, Hubei 430060, China
Rong Hu
Department of Neurosurgery, Southwest Hospital, Third Military Medical University (Army Medical University), Chongqing 400038, China
Renya Zhan
Department of Neurosurgery, First Affiliated Hospital, School of Medicine, Zhejiang University, Hangzhou, Zhejiang 310003, China,
Qianxue Chen
Department of Neurosurgery, Renmin Hospital of Wuhan University, Wuhan, Hubei 430060, China
Dong Zhang
School of Physics and Technology, Wuhan University, Wuhan, Hubei 430060, China
Xingen Zhu
Department of Neurosurgery, The Second Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006, China; Jiangxi Key Laboratory of Neurological Tumors and Cerebrovascular Diseases, Nanchang, Jiangxi 330006, China; Jiangxi Health Commission Key Laboratory of Neurological Medicine, Nanchang, Jiangxi 330006, China; Institute of Neuroscience, Nanchang University, Nanchang, Jiangxi 330006, China; Corresponding authors.
Accurate stroke assessment and consequent favorable clinical outcomes rely on the early identification and quantification of aneurysmal subarachnoid hemorrhage (aSAH) in non-contrast computed tomography (NCCT) images. However, hemorrhagic lesions can be complex and difficult to distinguish manually. To solve these problems, here we propose a novel Hybrid 2D/3D UNet deep-learning framework for automatic aSAH identification and quantification in NCCT images. We evaluated 1824 consecutive patients admitted with aSAH to four hospitals in China between June 2018 and May 2022. Accuracy and precision, Dice scores and intersection over union (IoU), and interclass correlation coefficients (ICC) were calculated to assess model performance, segmentation performance, and correlations between automatic and manual segmentation, respectively. A total of 1355 patients with aSAH were enrolled: 931, 101, 179, and 144 in four datasets, of whom 326 were scanned with Siemens, 640 with Philips, and 389 with GE Medical Systems scanners. Our proposed deep-learning method accurately identified (accuracies 0.993–0.999) and segmented (Dice scores 0.550–0.897) hemorrhage in both the internal and external datasets, even combinations of hemorrhage subtypes. We further developed a convenient AI-assisted platform based on our algorithm to assist clinical workflows, whose performance was comparable to manual measurements by experienced neurosurgeons (ICCs 0.815–0.957) but with greater efficiency and reduced cost. While this tool has not yet been prospectively tested in clinical practice, our innovative hybrid network algorithm and platform can accurately identify and quantify aSAH, paving the way for fast and cheap NCCT interpretation and a reliable AI-based approach to expedite clinical decision-making for aSAH patients.
