[论文解读] Unsupervised Brain Abnormality Detection Using High Fidelity Image Reconstruction Networks.

Kazuma Kobayashi, Ryuichiro Hataya+9 more|arXiv (Cornell University)|May 26, 2020

Radiomics and Machine Learning in Medical Imaging参考文献 66被引用 6

一句话总结

该论文提出了一种基于具有结构相似性约束的内省变分自编码器的高保真图像重建网络，用于无监督的体素级脑部MRI异常检测。该模型在正常MRI扫描上进行训练，通过将查询图像映射到学习到的潜在空间来重建输入图像，并通过测量输入图像与重建图像之间多层级特征差异来识别异常——在对比增强型病灶（如转移性脑肿瘤）上表现出色。

ABSTRACT

Recent advances in deep learning have facilitated near-expert medical image analysis. Supervised learning is the mainstay of current approaches, though its success requires the use of large, fully labeled datasets. However, in real-world medical practice, previously unseen disease phenotypes are encountered that have not been defined a priori in finite-size datasets. Unsupervised learning, a hypothesis-free learning framework, may play a complementary role to supervised learning. Here, we demonstrate a novel framework for voxel-wise abnormality detection in brain magnetic resonance imaging (MRI), which exploits an image reconstruction network based on an introspective variational autoencoder trained with a structural similarity constraint. The proposed network learns a latent representation for normal anatomical variation using a series of images that do not include annotated abnormalities. After training, the network can map unseen query images to positions in the latent space, and latent variables sampled from those positions can be mapped back to the image space to yield normal-looking replicas of the input images. Finally, the network considers abnormality scores, which are designed to reflect differences at several image feature levels, in order to locate image regions that may contain abnormalities. The proposed method is evaluated on a comprehensively annotated dataset spanning clinically significant structural abnormalities of the brain parenchyma in a population having undergone radiotherapy for brain metastasis, demonstrating that it is particularly effective for contrast-enhanced lesions, i.e., metastatic brain tumors and extracranial metastatic tumors.

研究动机与目标

• 解决在有限标注数据集中未见过的脑部异常检测挑战。
• 开发一种无需假设的无监督方法，用于识别脑部MRI中的异常区域，而无需依赖标注的病理信息。
• 利用高保真图像重建，从未标注数据中学习正常解剖变异。
• 通过量化重建图像中多层级特征差异，实现体素级异常评分。

提出的方法

• 在一组正常脑部MRI扫描上训练一个带有内省注意力机制的变分自编码器，以学习正常解剖变异的解耦潜在表征。
• 在训练过程中引入结构相似性（SSIM）约束，以保留细微图像细节并提高重建保真度。
• 训练完成后，将输入MRI扫描编码到潜在空间，并从该潜在空间采样以解码生成外观正常的输入重建图像。
• 异常评分通过计算原始图像与重建图像在特征图上的多尺度差异获得，突出显示结构偏离的区域。
• 该方法采用分层特征级比较策略，以检测仅靠像素级重建损失无法捕捉的细微异常。
• 该框架以端到端方式运行，支持在未见扫描中实现无监督的体素级异常区域定位。

实验结果

研究问题

• RQ1 无监督深度学习模型是否能够在不依赖标注病理数据的情况下检测MRI扫描中的脑部异常？
• RQ2 带有结构相似性损失的变分自编码器在重建正常脑部解剖结构并识别提示病理的偏离方面效果如何？
• RQ3 多层级特征差异评分在多大程度上能够精确定位脑部MRI中的异常区域，尤其是对比增强型病灶？
• RQ4 该方法在临床相关的、此前未见过的脑部异常（如转移性肿瘤）上的表现如何？
• RQ5 该模型能否泛化到训练集中未包含的多样化、未见过的脑实质异常表型？

主要发现

• 所提出方法在全面标注的数据集上对对比增强型病灶（包括转移性和颅外转移性肿瘤）的检测表现出高敏感性。
• 引入结构相似性损失显著提升了重建质量，并相比标准L2损失更好地检测到细微结构异常。
• 基于多层级特征差异的异常评分在定位病理区域方面优于像素级重建误差。
• 该模型对训练数据中未出现的先前未见疾病表型表现出鲁棒性，证实了其无监督泛化能力。
• 该框架成功实现了体素级异常定位，实现了脑部MRI中可疑病理区域的精确空间映射。