From the Department of Radiology, Duke University Medical Center, 2301 Erwin Rd, Box 3808, Durham, NC 27710 (C.O.L., E.C, A.L., J.F.); Department of Radiology (J.V.C., F.T., G.C., A.R., Y.L.) and Weill Institute for Neurosciences (Y.W.W.), University of California San Francisco, San Francisco, Calif; Department of Pediatrics, University of Washington, Seattle, Wash (S.J.); Department of Pediatrics, Saint Louis University, St Louis, Mo (A.M.); Mallinckrodt Institute of Radiology, Washington University School of Medicine, St Louis, Mo (R.C.M.); and Children's Hospital Los Angeles, University of Southern California, Los Angeles, Calif (J.L.W.).
Radiol Artif Intell. 2024 Sep;6(5):e240076. doi: 10.1148/ryai.240076.
Purpose To develop a deep learning algorithm to predict 2-year neurodevelopmental outcomes in neonates with hypoxic-ischemic encephalopathy using MRI and basic clinical data. Materials and Methods In this study, MRI data of term neonates with encephalopathy in the High-dose Erythropoietin for Asphyxia and Encephalopathy (HEAL) trial (ClinicalTrials.gov: NCT02811263), who were enrolled from 17 institutions between January 25, 2017, and October 9, 2019, were retrospectively analyzed. The harmonized MRI protocol included T1-weighted, T2-weighted, and diffusion tensor imaging. Deep learning classifiers were trained to predict the primary outcome of the HEAL trial (death or any neurodevelopmental impairment at 2 years) using multisequence MRI and basic clinical variables, including sex and gestational age at birth. Model performance was evaluated on test sets comprising 10% of cases from 15 institutions (in-distribution test set, = 41) and 10% of cases from two institutions (out-of-distribution test set, = 41). Model performance in predicting additional secondary outcomes, including death alone, was also assessed. Results For the 414 neonates (mean gestational age, 39 weeks ± 1.4 [SD]; 232 male, 182 female), in the study cohort, 198 (48%) died or had any neurodevelopmental impairment at 2 years. The deep learning model achieved an area under the receiver operating characteristic curve (AUC) of 0.74 (95% CI: 0.60, 0.86) and 63% accuracy in the in-distribution test set and an AUC of 0.77 (95% CI: 0.63, 0.90) and 78% accuracy in the out-of-distribution test set. Performance was similar or better for predicting secondary outcomes. Conclusion Deep learning analysis of neonatal brain MRI yielded high performance for predicting 2-year neurodevelopmental outcomes. Convolutional Neural Network (CNN), Prognosis, Pediatrics, Brain, Brain Stem Clinical trial registration no. NCT02811263 © RSNA, 2024 See also commentary by Rafful and Reis Teixeira in this issue.
目的 利用 MRI 和基本临床数据开发一种深度学习算法，以预测患有缺氧缺血性脑病的新生儿 2 年神经发育结局。
材料与方法 在这项研究中，回顾性分析了 2017 年 1 月 25 日至 2019 年 10 月 9 日期间，来自 17 家机构的参加高剂量促红细胞生成素治疗窒息和脑病（HEAL）试验（ClinicalTrials.gov：NCT02811263）的足月脑病新生儿的 MRI 数据。协调后的 MRI 方案包括 T1 加权、T2 加权和弥散张量成像。使用多序列 MRI 和基本临床变量（包括性别和出生时的胎龄）训练深度学习分类器，以预测 HEAL 试验的主要结局（2 年时死亡或任何神经发育障碍）。在包含来自 15 家机构的 10%病例的测试集中（分布内测试集， = 41）和来自 2 家机构的 10%病例的测试集中（分布外测试集， = 41）评估模型性能。还评估了预测其他次要结局（包括单独死亡）的模型性能。
结果 在 414 名新生儿（平均胎龄 39 周±1.4[标准差]；232 名男性，182 名女性）中，研究队列中有 198 名（48%）在 2 岁时死亡或存在任何神经发育障碍。深度学习模型在分布内测试集中的 AUC 为 0.74（95%CI：0.60，0.86），准确率为 63%，在分布外测试集中的 AUC 为 0.77（95%CI：0.63，0.90），准确率为 78%。对于预测次要结局，性能相似或更好。
结论 对新生儿脑 MRI 的深度学习分析可实现对 2 年神经发育结局的高预测性能。卷积神经网络（CNN）、预测、儿科学、脑、脑桥临床试验注册号：NCT02811263
©2024 RSNA，保留所有权利。
参见本期 Rafful 和 Reis Teixeira 的评论。
Zotero 插件