Liu Ke, Qin Siyuan, Ning Jinlai, Xin Peijin, Wang Qizheng, Chen Yongye, Zhao Weili, Zhang Enlong, Lang Ning
Department of Radiology, Peking University Third Hospital, Beijing 100191, China.
Department of Informatics, King's College London, London WC2B 4BG, UK.
Cancers (Basel). 2023 May 30;15(11):2974. doi: 10.3390/cancers15112974.
We aim to investigate the feasibility and evaluate the performance of a ResNet-50 convolutional neural network (CNN) based on magnetic resonance imaging (MRI) in predicting primary tumor sites in spinal metastases. Conventional sequences (T1-weighted, T2-weighted, and fat-suppressed T2-weighted sequences) MRIs of spinal metastases patients confirmed by pathology from August 2006 to August 2019 were retrospectively analyzed. Patients were partitioned into non-overlapping sets of 90% for training and 10% for testing. A deep learning model using ResNet-50 CNN was trained to classify primary tumor sites. Top-1 accuracy, precision, sensitivity, area under the curve for the receiver-operating characteristic (AUC-ROC), and F1 score were considered as the evaluation metrics. A total of 295 spinal metastases patients (mean age ± standard deviation, 59.9 years ± 10.9; 154 men) were evaluated. Included metastases originated from lung cancer ( = 142), kidney cancer ( = 50), mammary cancer ( = 41), thyroid cancer ( = 34), and prostate cancer ( = 28). For 5-class classification, AUC-ROC and top-1 accuracy were 0.77 and 52.97%, respectively. Additionally, AUC-ROC for different sequence subsets ranged between 0.70 (for T2-weighted) and 0.74 (for fat-suppressed T2-weighted). Our developed ResNet-50 CNN model for predicting primary tumor sites in spinal metastases at MRI has the potential to help prioritize the examinations and treatments in case of unknown primary for radiologists and oncologists.
我们旨在研究基于磁共振成像（MRI）的ResNet-50卷积神经网络（CNN）在预测脊柱转移瘤原发肿瘤部位方面的可行性并评估其性能。对2006年8月至2019年8月经病理证实的脊柱转移瘤患者的常规序列（T1加权、T2加权和脂肪抑制T2加权序列）MRI进行回顾性分析。将患者分为90%用于训练和10%用于测试的非重叠组。使用ResNet-50 CNN的深度学习模型被训练用于对原发肿瘤部位进行分类。将Top-1准确率、精确率、灵敏度、受试者操作特征曲线下面积（AUC-ROC）和F1分数作为评估指标。共评估了295例脊柱转移瘤患者（平均年龄±标准差，59.9岁±10.9；154例男性）。纳入的转移瘤起源于肺癌（n = 142）、肾癌（n = 50）、乳腺癌（n = 41）、甲状腺癌（n = 34）和前列腺癌（n = 28）。对于5类分类，AUC-ROC和Top-1准确率分别为0.77和52.97%。此外，不同序列子集的AUC-ROC在0.70（T2加权）至0.74（脂肪抑制T2加权）之间。我们开发的用于在MRI上预测脊柱转移瘤原发肿瘤部位的ResNet-50 CNN模型有可能帮助放射科医生和肿瘤学家在原发灶不明的情况下对检查和治疗进行优先排序。
