From the Department of Radiology and Research Institute of Radiology (J.Y., Y.A., S.Y.O., S.M.L., J.B.S.) and Department of Convergence Medicine (K.C., N.K.), University of Ulsan College of Medicine, Asan Medical Center, 88 Olympic-ro 43-gil, Songpa-gu, Seoul 138-736, Korea.
Radiology. 2023 Oct;309(1):e230606. doi: 10.1148/radiol.230606.
Background Most artificial intelligence algorithms that interpret chest radiographs are restricted to an image from a single time point. However, in clinical practice, multiple radiographs are used for longitudinal follow-up, especially in intensive care units (ICUs). Purpose To develop and validate a deep learning algorithm using thoracic cage registration and subtraction to triage pairs of chest radiographs showing no change by using longitudinal follow-up data. Materials and Methods A deep learning algorithm was retrospectively developed using baseline and follow-up chest radiographs in adults from January 2011 to December 2018 at a tertiary referral hospital. Two thoracic radiologists reviewed randomly selected pairs of "change" and "no change" images to establish the ground truth, including normal or abnormal status. Algorithm performance was evaluated using area under the receiver operating characteristic curve (AUC) analysis in a validation set and temporally separated internal test sets (January 2019 to August 2021) from the emergency department (ED) and ICU. Threshold calibration for the test sets was conducted, and performance with 40% and 60% triage thresholds was assessed. Results This study included 3 304 996 chest radiographs in 329 036 patients (mean age, 59 years ± 14 [SD]; 170 433 male patients). The training set included 550 779 pairs of radiographs. The validation set included 1620 pairs (810 no change, 810 change). The test sets included 533 pairs (ED; 265 no change, 268 change) and 600 pairs (ICU; 310 no change, 290 change). The algorithm had AUCs of 0.77 (validation), 0.80 (ED), and 0.80 (ICU). With a 40% triage threshold, specificity was 88.4% (237 of 268 pairs) and 90.0% (261 of 290 pairs) in the ED and ICU, respectively. With a 60% triage threshold, specificity was 79.9% (214 of 268 pairs) and 79.3% (230 of 290 pairs) in the ED and ICU, respectively. For urgent findings (consolidation, pleural effusion, pneumothorax), specificity was 78.6%-100% (ED) and 85.5%-93.9% (ICU) with a 40% triage threshold. Conclusion The deep learning algorithm could triage pairs of chest radiographs showing no change while detecting urgent interval changes during longitudinal follow-up. © RSNA, 2023 See also the editorial by Czum in this issue.
背景 大多数解释胸部 X 光片的人工智能算法都仅限于单个时间点的图像。然而，在临床实践中，特别是在重症监护病房（ICU）中，会使用多张 X 光片进行纵向随访。目的 开发并验证一种深度学习算法，该算法使用胸腔注册和减影技术对无变化的胸部 X 光片对进行分类，这些 X 光片对来自于纵向随访数据。材料与方法 该深度学习算法是使用 2011 年 1 月至 2018 年 12 月在一家三级转诊医院的成人的基线和随访胸部 X 光片进行回顾性开发的。两名胸部放射科医生随机选择“有变化”和“无变化”的图像对进行审查，以确定正常或异常状态，作为真实数据。在验证集和时间上分离的内部测试集中（2019 年 1 月至 2021 年 8 月），使用接收者操作特征曲线（AUC）分析评估算法性能，这些数据集来自急诊部（ED）和 ICU。对测试集进行了阈值校准，并评估了 40%和 60%分诊阈值的性能。结果 本研究纳入了 329036 名患者的 3304996 张胸部 X 光片（平均年龄为 59 岁±14[标准差]；男性患者 170433 名）。训练集包括 550779 对 X 光片。验证集包括 1620 对（810 对无变化，810 对有变化）。测试集包括 533 对（ED；265 对无变化，268 对有变化）和 600 对（ICU；310 对无变化，290 对有变化）。该算法在验证集、ED 和 ICU 中的 AUC 分别为 0.77、0.80 和 0.80。在 40%的分诊阈值下，ED 和 ICU 中无变化的特异性分别为 88.4%（237/268 对）和 90.0%（261/290 对）。在 60%的分诊阈值下，ED 和 ICU 中无变化的特异性分别为 79.9%（214/268 对）和 79.3%（230/290 对）。对于紧急发现（实变、胸腔积液、气胸），在 40%的分诊阈值下，ED 和 ICU 的特异性分别为 78.6%-100%和 85.5%-93.9%。结论 深度学习算法可以对显示无变化的胸部 X 光片对进行分类，同时在纵向随访期间检测到紧急的间隔变化。
