[论文解读] Deep Learning for identifying radiogenomic associations in breast cancer

Zhe Zhu, Ehab A. AlBadawy+4 more|arXiv (Cornell University)|Nov 29, 2017

Radiomics and Machine Learning in Medical Imaging参考文献 23被引用 12

一句话总结

本研究调查了深度学习是否能够通过动态对比增强磁共振成像（DCE-MRI）扫描识别乳腺癌中的放射基因组关联，方法是将肿瘤分子亚型（Luminal A 与 非-Luminal A）进行分类。在270例患者扫描中，采用三种深度学习方法——从零开始训练、迁移学习和现成特征——最佳性能（AUC = 0.65）通过使用预训练网络提取的特征实现，表明深度学习可在乳腺癌放射基因组研究中发挥支持作用。

ABSTRACT

Purpose: To determine whether deep learning models can distinguish between breast cancer molecular subtypes based on dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). Materials and methods: In this institutional review board-approved single-center study, we analyzed DCE-MR images of 270 patients at our institution. Lesions of interest were identified by radiologists. The task was to automatically determine whether the tumor is of the Luminal A subtype or of another subtype based on the MR image patches representing the tumor. Three different deep learning approaches were used to classify the tumor according to their molecular subtypes: learning from scratch where only tumor patches were used for training, transfer learning where networks pre-trained on natural images were fine-tuned using tumor patches, and off-the-shelf deep features where the features extracted by neural networks trained on natural images were used for classification with a support vector machine. Network architectures utilized in our experiments were GoogleNet, VGG, and CIFAR. We used 10-fold crossvalidation method for validation and area under the receiver operating characteristic (AUC) as the measure of performance. Results: The best AUC performance for distinguishing molecular subtypes was 0.65 (95% CI:[0.57,0.71]) and was achieved by the off-the-shelf deep features approach. The highest AUC performance for training from scratch was 0.58 (95% CI:[0.51,0.64]) and the best AUC performance for transfer learning was 0.60 (95% CI:[0.52,0.65]) respectively. For the off-the-shelf approach, the features extracted from the fully connected layer performed the best. Conclusion: Deep learning may play a role in discovering radiogenomic associations in breast cancer.

研究动机与目标

• 确定深度学习模型是否能够仅使用DCE-MRI图像区分Luminal A与非-Luminal A乳腺癌亚型。
• 评估三种深度学习策略的性能：从零开始训练、迁移学习和现成特征提取。
• 评估预训练神经网络从医学影像数据中提取放射基因组特征的实用性。
• 使用10折交叉验证和AUC作为主要指标对模型进行验证。

提出的方法

• 分析了270名患者的DCE-MRI扫描，由放射科医生提取并标注肿瘤区域。
• 应用了三种深度学习方法：在肿瘤区域上从零开始训练网络、微调ImageNet预训练网络（GoogleNet、VGG、CIFAR），以及使用预训练网络提取的特征。
• 对于现成特征方法，将预训练网络全连接层的特征用作支持向量机分类器的输入。
• 通过10折交叉验证评估模型，以确保鲁棒性和泛化能力。
• 以受试者工作特征曲线下面积（AUC）作为主要性能指标。
• 网络架构包括GoogleNet、VGG和CIFAR，均针对医学图像分类进行了调整。

实验结果

研究问题

• RQ1 深度学习模型是否能够仅使用DCE-MRI图像准确分类乳腺癌分子亚型？
• RQ2 从零开始训练、迁移学习和现成特征提取之间的性能表现有何差异？
• RQ3 哪种深度学习架构和特征层能实现最佳的放射基因组分类性能？
• RQ4 DCE-MRI中是否存在可被预训练深度神经网络捕捉到的显著放射基因组信号？

主要发现

• 现成特征方法在区分Luminal A与非-Luminal A亚型中取得了最高的AUC（0.65，95%置信区间：[0.57, 0.71]）。
• 现成特征方法中表现最佳的特征来自预训练网络的全连接层。
• 从零开始训练的AUC为0.58（95%置信区间：[0.51, 0.64]），表明在无预训练的情况下性能有限。
• 迁移学习的AUC为0.60（95%置信区间：[0.52, 0.65]），优于从零开始训练，但低于现成特征方法。
• 结果表明，在此情境下，预训练的深度特征比微调或随机初始化的网络在放射基因组分类中更为有效。
• 总体而言，深度学习在利用非侵入性影像识别乳腺癌放射基因组关联方面展现出潜力。