From Google Health, 1600 Amphitheatre Pkwy, Mountain View, CA 94043 (S.K., J.Y., S.J., R.P., Z.N., C.C., N.B., S.M.M., T.H., A.P.K., G.S.C., L.P., K.C., P.H.C.C., Y.L., K.E., D.T., S.S., S.P.); Advanced Clinical, Deerfield, Ill (C.L.); Apollo Radiology International, Hyderabad, India (S.R.K.); TB Department, Center of Infectious Disease Research in Zambia, Lusaka, Zambia (M.M.); Sibanye Stillwater, Weltevreden Park, Roodepoort, South Africa (J.M.); and Clickmedix, Gaithersburg, Md (T.S.).
Radiology. 2023 Jan;306(1):124-137. doi: 10.1148/radiol.212213. Epub 2022 Sep 6.
Background The World Health Organization (WHO) recommends chest radiography to facilitate tuberculosis (TB) screening. However, chest radiograph interpretation expertise remains limited in many regions. Purpose To develop a deep learning system (DLS) to detect active pulmonary TB on chest radiographs and compare its performance to that of radiologists. Materials and Methods A DLS was trained and tested using retrospective chest radiographs (acquired between 1996 and 2020) from 10 countries. To improve generalization, large-scale chest radiograph pretraining, attention pooling, and semisupervised learning ("noisy-student") were incorporated. The DLS was evaluated in a four-country test set (China, India, the United States, and Zambia) and in a mining population in South Africa, with positive TB confirmed with microbiological tests or nucleic acid amplification testing (NAAT). The performance of the DLS was compared with that of 14 radiologists. The authors studied the efficacy of the DLS compared with that of nine radiologists using the Obuchowski-Rockette-Hillis procedure. Given WHO targets of 90% sensitivity and 70% specificity, the operating point of the DLS (0.45) was prespecified to favor sensitivity. Results A total of 165 754 images in 22 284 subjects (mean age, 45 years; 21% female) were used for model development and testing. In the four-country test set (1236 subjects, 17% with active TB), the receiver operating characteristic (ROC) curve of the DLS was higher than those for all nine India-based radiologists, with an area under the ROC curve of 0.89 (95% CI: 0.87, 0.91). Compared with these radiologists, at the prespecified operating point, the DLS sensitivity was higher (88% vs 75%, < .001) and specificity was noninferior (79% vs 84%, = .004). Trends were similar within other patient subgroups, in the South Africa data set, and across various TB-specific chest radiograph findings. In simulations, the use of the DLS to identify likely TB-positive chest radiographs for NAAT confirmation reduced the cost by 40%-80% per TB-positive patient detected. Conclusion A deep learning method was found to be noninferior to radiologists for the determination of active tuberculosis on digital chest radiographs. © RSNA, 2022 See also the editorial by van Ginneken in this issue.
背景 世界卫生组织（WHO）建议进行胸部 X 光检查，以促进结核病（TB）筛查。然而，许多地区仍然缺乏胸部 X 光片解读方面的专业知识。目的 开发一种深度学习系统（DLS），以检测胸部 X 光片中的活动性肺结核，并比较其与放射科医生的表现。材料和方法 使用来自 10 个国家的回顾性胸部 X 光片（1996 年至 2020 年采集）对 DLS 进行了训练和测试。为了提高泛化能力，采用了大规模胸部 X 光预训练、注意力池化和半监督学习（“嘈杂学生”）。在四个国家的测试集（中国、印度、美国和赞比亚）和南非的采矿人群中对 DLS 进行了评估，通过微生物学检测或核酸扩增检测（NAAT）证实阳性的 TB。比较了 DLS 与 14 名放射科医生的表现。作者使用 Obuchowski-Rockette-Hillis 程序研究了 DLS 与 9 名放射科医生的功效。考虑到世卫组织 90%敏感性和 70%特异性的目标，将 DLS 的工作点（0.45）预设为有利于敏感性。结果 共使用 22284 名受试者的 165754 张图像（平均年龄 45 岁；21%为女性）进行模型开发和测试。在四个国家的测试集（1236 名受试者，17%患有活动性 TB）中，DLS 的受试者工作特征（ROC）曲线高于所有 9 名印度籍放射科医生，ROC 曲线下面积为 0.89（95%CI：0.87，0.91）。与这些放射科医生相比，在预设的工作点，DLS 的敏感性更高（88%比 75%，<0.001），特异性非劣效（79%比 84%，=0.004）。在其他患者亚组、南非数据集和各种特定于 TB 的胸部 X 光片发现中，趋势相似。在模拟中，使用 DLS 来识别可能的 TB 阳性胸部 X 光片以进行 NAAT 确认，每检测到一个 TB 阳性患者可降低 40%-80%的成本。结论 一种深度学习方法被发现与放射科医生在数字胸部 X 光片中确定活动性肺结核的能力相当。
