基于影像信息的人工神经网络的脑胶质瘤分级

目的：探讨利用人工神经网络对脑胶质瘤进行分级的可行性和有效性。方法：回顾性纳入2012至2017年经病理证实的130例脑胶质瘤患者，基于磁共振增强T1加权相二维图像提取的共41个影像学特征。建立人工神经网络并进行特征选择以得到最优化的神经网络模型，随机抽取一半病例的影像学特征数据对神经网络进行训练，并用另一半特征数据对训练完成后的神经网络进行胶质瘤分级效果验证。对神经网络重复进行100次训练和验证，将结果进行平均。结果：经过特征选择后的神经网络模型共选取5个特征作为输入特征，神经网络对脑胶质瘤分级的平均准确率为90.32%，平均敏感度为87.86%，平均特异度为92.49%，受试者操作曲线的曲线下面积为0.9486。结论：人工神经网络作为一种人工智能方法，对脑胶质瘤分级具有较高的准确性，为脑胶质瘤的无创性术前分级提供了可行的辅助手段。

An artificial neural network model for glioma grading using image information

Objective: To explore the feasibility and efficacy of artificial neural network for differentiating high-grade glioma and low-grade glioma using image information.Methods: A total of 130 glioma patients with confirmed pathological diagnosis were selectedretrospectively from 2012 to 2017. Forty one imaging features were extracted from each subjectsbased on 2-dimension magnetic resonance T1 weighted imaging with contrast-enhancement. Anartificial neural network model was created and optimized according to the performance of featureselection. The training dataset was randomly selected half of the whole dataset, and the other halfdataset was used to verify the performance of the neural network for glioma grading. The trainingverificationprocess was repeated for 100 times and the performance was averaged.Results: A total of 5 imaging features were selected as the ultimate input features for the neuralnetwork. The mean accuracy of the neural network for glioma grading was 90.32%, with a meansensitivity at 87.86% and a mean specificity at 92.49%. The area under the curve of receiveroperating characteristic curve was 0.9486.Conclusion: As a technique of artificial intelligence, neural network can reach a relatively highaccuracy for the grading of glioma and provide a non-invasive and promising computer-aideddiagnostic process for the pre-operative grading of glioma.