基于MRI和深度学习的桥小脑角区脑膜瘤与听神经瘤分类算法研究

桥小脑角区脑膜瘤与听神经瘤是两种常见的脑部肿瘤，它们的临床表现和影像学表现极为相似，在临床诊断时极易发生误诊．将影像数据与深度学习方法相结合，建立脑膜瘤与听神经瘤的判别模型，可以为两种脑肿瘤的及时准确诊断提供重要手段．本文采集了307名脑肿瘤患者的T₁W-SE序列图像，通过对原始图像进行限制对比度自适应直方图均衡化（Contrast Limited Adaptive Histogram Equalization，CLAHE）等预处理，提升数据集图像质量，再经过建立的三维卷积神经网络（3-Dimensional Convolutional Neural Network，3D CNN）深度学习框架中图像特征的学习，实现对脑膜瘤与听神经瘤的分类．图像增强参数与网络结构参数经过优化后，对脑膜瘤与听神经瘤分类的准确率达到0.918 0，曲线下面积（Area Under Curve，AUC）为0.913 4，实现了对桥小脑角区脑膜瘤与听神经瘤的有效判别．

A Deep Learning Algorithm for Classifying Meningioma and Auditory Neuroma in the Cerebellopontine Angle from Magnetic Resonance Images

Meningioma and auditory neuroma are two types of brain tumors frequently found in the cerebellopontine angle. Misdiagnosis of the two is common due to their similarity in clinical manifestations and imaging manifestations. In this work, the deep learning algorithm was developed to classify the meningioma and auditory neuroma from the magnetic resonance images, assisting the timely and accurate diagnosis to the two brain tumors. T₁-weighted spin-echo (T₁W-SE) images were collected from 307 patients with brain tumors. Contrast limited adaptive histogram equalization (CLAHE) pre-processing was used to improve the quality of the dataset. The image features were learnt in the deep learning framework of 3-dimensional convolutional neural network (3D CNN). By optimizing the image enhancement parameters and the network structure parameters, the accuracy of the meningioma/acoustic neuroma classification model reached 0.918 0, and the area under curve (AUC) was found to be 0.913 4.