From the Department of Radiology, Research Institute of Radiological Science and Center for Clinical Imaging Data Science (I.S., H.K., S.S.A., B.S., S.-K.L.), Yonsei University College of Medicine, Seoul, Korea.
From the Department of Radiology, Research Institute of Radiological Science and Center for Clinical Imaging Data Science (I.S., H.K., S.S.A., B.S., S.-K.L.), Yonsei University College of Medicine, Seoul, Korea
AJNR Am J Neuroradiol. 2021 May;42(5):838-844. doi: 10.3174/ajnr.A7003. Epub 2021 Mar 18.
Differentiating glioblastoma from solitary brain metastasis preoperatively using conventional MR images is challenging. Deep learning models have shown promise in performing classification tasks. The diagnostic performance of a deep learning-based model in discriminating glioblastoma from solitary brain metastasis using preoperative conventional MR images was evaluated.
Records of 598 patients with histologically confirmed glioblastoma or solitary brain metastasis at our institution between February 2006 and December 2017 were retrospectively reviewed. Preoperative contrast-enhanced T1WI and T2WI were preprocessed and roughly segmented with rectangular regions of interest. A deep neural network was trained and validated using MR images from 498 patients. The MR images of the remaining 100 were used as an internal test set. An additional 143 patients from another tertiary hospital were used as an external test set. The classifications of ResNet-50 and 2 neuroradiologists were compared for their accuracy, precision, recall, F1 score, and area under the curve.
The areas under the curve of ResNet-50 were 0.889 and 0.835 in the internal and external test sets, respectively. The area under the curve of neuroradiologists 1 and 2 were 0.889 and 0.768 in the internal test set and 0.857 and 0.708 in the external test set, respectively.
A deep learning-based model may be a supportive tool for preoperative discrimination between glioblastoma and solitary brain metastasis using conventional MR images.
术前使用常规磁共振成像区分胶质母细胞瘤和单发脑转移瘤具有挑战性。深度学习模型在执行分类任务方面显示出了良好的效果。本研究旨在评估一种基于深度学习的模型在使用术前常规磁共振成像区分胶质母细胞瘤和单发脑转移瘤方面的诊断性能。
回顾性分析了 2006 年 2 月至 2017 年 12 月在我院经组织学证实的 598 例胶质母细胞瘤或单发脑转移瘤患者的病例记录。对术前增强 T1WI 和 T2WI 进行预处理，并使用矩形感兴趣区进行大致分割。使用来自 498 例患者的 MR 图像对深度神经网络进行训练和验证。其余 100 例患者的 MR 图像作为内部测试集。另外 143 例患者来自另一家三级医院，作为外部测试集。比较了 ResNet-50 和 2 位神经放射科医生的分类准确性、精密度、召回率、F1 评分和曲线下面积。
ResNet-50 在内部和外部测试集中的曲线下面积分别为 0.889 和 0.835。神经放射科医生 1 和 2 的曲线下面积在内部测试集中分别为 0.889 和 0.768，在外部测试集中分别为 0.857 和 0.708。
基于深度学习的模型可能是使用常规磁共振成像术前区分胶质母细胞瘤和单发脑转移瘤的一种辅助工具。
