From the Department of Neurosurgery, University of Pittsburgh Medical Center, Pittsburgh, Pa (M.P., A.P., K.H., E.N., S.R., S.C., D.O.O.); Departments of Radiology (D.A., S.W.), Biomedical Informatics (S.W.), and Bioengineering (S.W.), and Intelligent Systems Program (S.W.), University of Pittsburgh, 3240 Craft Pl, Room 322, Pittsburgh, PA 15213; Department of Neurosurgery, University of Washington, Seattle, Wash (J.B., N.T.); Department of Radiology, University of California San Francisco, San Francisco, Calif (E.Y.).
Radiology. 2022 Aug;304(2):385-394. doi: 10.1148/radiol.212181. Epub 2022 Apr 26.
Background After severe traumatic brain injury (sTBI), physicians use long-term prognostication to guide acute clinical care yet struggle to predict outcomes in comatose patients. Purpose To develop and evaluate a prognostic model combining deep learning of head CT scans and clinical information to predict long-term outcomes after sTBI. Materials and Methods This was a retrospective analysis of two prospectively collected databases. The model-building set included 537 patients (mean age, 40 years ± 17 [SD]; 422 men) from one institution from November 2002 to December 2018. Transfer learning and curriculum learning were applied to a convolutional neural network using admission head CT to predict mortality and unfavorable outcomes (Glasgow Outcomes Scale scores 1-3) at 6 months. This was combined with clinical input for a holistic fusion model. The models were evaluated using an independent internal test set and an external cohort of 220 patients with sTBI (mean age, 39 years ± 17; 166 men) from 18 institutions in the Transforming Research and Clinical Knowledge in Traumatic Brain Injury (TRACK-TBI) study from February 2014 to April 2018. The models were compared with the International Mission on Prognosis and Analysis of Clinical Trials in TBI (IMPACT) model and the predictions of three neurosurgeons. Area under the receiver operating characteristic curve (AUC) was used as the main model performance metric. Results The fusion model had higher AUCs than did the IMPACT model in the prediction of mortality (AUC, 0.92 [95% CI: 0.86, 0.97] vs 0.80 [95% CI: 0.71, 0.88]; < .001) and unfavorable outcomes (AUC, 0.88 [95% CI: 0.82, 0.94] vs 0.82 [95% CI: 0.75, 0.90]; = .04) on the internal data set. For external TRACK-TBI testing, there was no evidence of a significant difference in the performance of any models compared with the IMPACT model (AUC, 0.83; 95% CI: 0.77, 0.90) in the prediction of mortality. The Imaging model (AUC, 0.73; 95% CI: 0.66-0.81; = .02) and the fusion model (AUC, 0.68; 95% CI: 0.60, 0.76; = .02) underperformed as compared with the IMPACT model (AUC, 0.83; 95% CI: 0.77, 0.89) in the prediction of unfavorable outcomes. The fusion model outperformed the predictions of the neurosurgeons. Conclusion A deep learning model of head CT and clinical information can be used to predict 6-month outcomes after severe traumatic brain injury. © RSNA, 2022 See also the editorial by Haller in this issue.
背景 在严重创伤性脑损伤（sTBI）后，医生使用长期预后来指导急性临床护理，但在昏迷患者中难以预测结局。目的 开发和评估一种结合深度学习头部 CT 扫描和临床信息的预后模型，以预测 sTBI 后的长期结局。
材料和方法 这是对两个前瞻性数据库进行的回顾性分析。模型构建集包括来自一个机构的 537 名患者（平均年龄，40 岁±17[SD]；422 名男性），时间为 2002 年 11 月至 2018 年 12 月。使用入院头部 CT 对卷积神经网络进行迁移学习和课程学习，以预测 6 个月时的死亡率和不良结局（Glasgow 结局量表评分为 1-3）。这与临床输入相结合，形成整体融合模型。使用内部测试集和来自 Transforming Research and Clinical Knowledge in Traumatic Brain Injury（TRACK-TBI）研究的 18 个机构的 220 名 sTBI 患者（平均年龄，39 岁±17；166 名男性）的外部队列来评估模型。比较模型与国际预后分析和临床试验创伤性脑损伤（IMPACT）模型和三位神经外科医生的预测。接收者操作特征曲线下面积（AUC）被用作主要模型性能指标。
结果 融合模型在预测死亡率（AUC，0.92[95%CI：0.86，0.97]与 0.80[95%CI：0.71，0.88]； <.001）和不良结局（AUC，0.88[95%CI：0.82，0.94]与 0.82[95%CI：0.75，0.90]； =.04）方面的表现均优于 IMPACT 模型。对于外部 TRACK-TBI 测试，与 IMPACT 模型相比，任何模型的性能均未显示出明显差异（死亡率预测的 AUC，0.83；95%CI：0.77，0.90）。与 IMPACT 模型（AUC，0.83；95%CI：0.77，0.89）相比，成像模型（AUC，0.73；95%CI：0.66-0.81； =.02）和融合模型（AUC，0.68；95%CI：0.60，0.76； =.02）在不良结局预测方面表现不佳。融合模型优于神经外科医生的预测。
结论 头部 CT 和临床信息的深度学习模型可用于预测严重创伤性脑损伤后的 6 个月结局。
Zotero 插件