Computer-Aided Detection AI Reduces Interreader Variability in Grading Hip Abnormalities With MRI.

Computer-Aided Detection AI Reduces Interreader Variability in Grading Hip Abnormalities With MRI.
复制标题

DOI:
10.1002/jmri.27164
复制
发表时间:
2020-10
影响因子:
4.4
通讯作者:
Majumdar, Sharmila
Majumdar, Sharmila
中科院分区:
医学2区
文献类型:
--
作者:
Tibrewala, Radhika;Ozhinsky, Eugene;Shah, Rutwik;Flament, Io;Crossley, Kay;Srinivasan, Ramya;Souza, Richard;Link, Thomas M.;Pedoia, Valentina;Majumdar, Sharmila

文献摘要

参考文献

被引文献

相似文献

髋部 MRI 的准确解释既耗时又困难,容易出现评审者之间和评审者内部的差异,并且缺乏普遍接受的分级标准来评估形态异常。 1) 开发和评估基于深度学习的模型,用于 MR 图像上髋关节骨关节炎 (OA) 形态异常的二元分类,2) 开发基于人工智能 (AI) 的辅助工具,以了解使用模型预测是否可以提高髋关节分级中读者间的一致性。回顾性研究旨在评估技术发展。从两项研究(242 名来自 LASEM [FORCe] 的患者和 122 名来自 UCSF 的患者)获得的总共 764 个 MRI 体积(364 名患者),分为 65-25-10% 的训练、验证、测试集用于网络训练。 3T MRI、2D T2 FSE、PD SPAIR。使用 MRNet 对软骨病变、骨髓水肿样病变和软骨下囊肿样病变进行自动二元分类,使用网络预测前后读者间一致。受试者工作特征 (ROC) 曲线、曲线下面积 (AUC)、特异性和敏感性以及平衡准确性。对于软骨病变、骨髓水肿样病变和软骨下囊肿样病变,AUC 分别为:0.80(95% 置信区间 [CI] 0.65,0.95)、0.84(95% CI 0.67,1.00)和 0.77(95% CI 0.66,0.85)。放射科医生对二元分类的敏感性和特异性分别为:0.79 (95% CI 0.65, 0.93) 和 0.80 (95% CI 0.59, 1.02)、0.40 (95% CI -0.02, 0.83) 和 0.72 (95% CI 0.59, 0.86)、0.75 (95% CI) 0.45, 1.05) 和 0.88 (95% CI 0.77, 0.98)。使用网络预测和显着图后,读者间平衡准确率从 53%、71% 和 56% 提高到 60%、73% 和 68%。我们已经证明,深度学习方法在髋部 MR 图像的临床分类任务中取得了高性能,并且使用深度学习模型的预测提高了所有病理学中读者间的一致性。 3 1
Accurate interpretation of hip MRI is time-intensive and difficult, prone to inter- and intrareviewer variability, and lacks a universally accepted grading scale to evaluate morphological abnormalities. To 1) develop and evaluate a deep-learning-based model for binary classification of hip osteoarthritis (OA) morphological abnormalities on MR images, and 2) develop an artificial intelligence (AI)-based assist tool to find if using the model predictions improves interreader agreement in hip grading. Retrospective study aimed to evaluate a technical development. A total of 764 MRI volumes (364 patients) obtained from two studies (242 patients from LASEM [FORCe] and 122 patients from UCSF), split into a 65–25–10% train, validation, test set for network training. 3T MRI, 2D T2 FSE, PD SPAIR. Automatic binary classification of cartilage lesions, bone marrow edema-like lesions, and subchondral cyst-like lesions using the MRNet, interreader agreement before and after using network predictions. Receiver operating characteristic (ROC) curve, area under curve (AUC), specificity and sensitivity, and balanced accuracy. For cartilage lesions, bone marrow edema-like lesions and subchondral cyst-like lesions the AUCs were: 0.80 (95% confidence interval [CI] 0.65, 0.95), 0.84 (95% CI 0.67, 1.00), and 0.77 (95% CI 0.66, 0.85), respectively. The sensitivity and specificity of the radiologist for binary classification were: 0.79 (95% CI 0.65, 0.93) and 0.80 (95% CI 0.59, 1.02), 0.40 (95% CI −0.02, 0.83) and 0.72 (95% CI 0.59, 0.86), 0.75 (95% CI 0.45, 1.05) and 0.88 (95% CI 0.77, 0.98). The interreader balanced accuracy increased from 53%, 71% and 56% to 60%, 73% and 68% after using the network predictions and saliency maps. We have shown that a deep-learning approach achieved high performance in clinical classification tasks on hip MR images, and that using the predictions from the deep-learning model improved the interreader agreement in all pathologies. 3 1
DOI: 10.1016/j.joca.2011.10.003
发表时间: 2012-01
影响因子: 7
作者:
Menashe, L.;Hirko, K.;Losina, E.;Kloppenburg, M.;Zhang, W.;Li, L.;Hunter, D. J.
通讯作者: Hunter, D. J.
DOI: 10.1148/radiol.2017170236
发表时间: 2018-04-01
期刊: RADIOLOGY
影响因子: 19.7
作者:
Larson, David B.;Chen, Matthew C.;Langlotz, Curtis P.
通讯作者: Langlotz, Curtis P.
DOI: 10.1016/j.diii.2019.02.007
发表时间: 2019-04-01
影响因子: 5.5
作者:
Roblot, V;Giret, Y.;Fournier, L.
通讯作者: Fournier, L.
DOI: 10.1148/radiol.2018172986
发表时间: 2018-10
期刊: Radiology
影响因子: 19.7
作者:
Liu F;Zhou Z;Samsonov A;Blankenbaker D;Larison W;Kanarek A;Lian K;Kambhampati S;Kijowski R
通讯作者: Kijowski R
DOI: 10.1007/s12065-015-0128-8
发表时间: 2015-09
影响因子: 2.6
作者:
Urbanowicz RJ;Moore JH
通讯作者: Moore JH