Autoencoder based self-supervised test-time adaptation for medical image analysis.
Autoencoder based self-supervised test-time adaptation for medical image analysis.
复制标题
DOI：
10.1016/j.media.2021.102136
复制
发表时间：
2021-08
影响因子：
10.9
通讯作者：
Prince JL
中科院分区：
文献类型：
--
作者：
He Y;Carass A;Zuo L;Dewey BE;Prince JL
深度神经网络已成功应用于医学图像分析任务，如分割和合成。然而，即使网络是在来自源域的大型数据集上训练的，它在看不见的测试域上的性能也不能保证。在临床实践中部署深度学习时，与网络训练数据不同的数据的性能下降是一个主要问题（称为域转移）。现有的工作集中在用来自测试域的数据重新训练模型，或者将测试域的数据与网络训练数据协调。一种常见的做法是将精心训练的模型分发给多个用户（例如，临床中心），然后每个用户使用该模型来处理他们自己的数据，这可能具有域转移（例如，改变成像参数和机器）。然而，缺乏源训练数据的可用性和训练新模型的成本通常阻止使用已知方法来解决用户特定的域偏移。在这里，我们问我们是否可以设计一个模型，一旦分发给用户，可以快速适应每个新的网站，而无需昂贵的重新训练或访问源训练数据？在本文中，我们提出了一个可以在推理过程中根据单个测试主题进行自适应的模型。该模型由三个部分组成，它们都是神经网络：任务模型（T），执行图像分析任务，如分割;一组自动编码器（AE）;和一组适配器（As）。任务模型和自动编码器是在源数据集上训练的，并且在计算上可能是昂贵的。在部署阶段，适配器被训练来变换测试图像及其特征，以最小化由自动编码器的重建损失测量的域偏移。在测试阶段，只有适配器被优化，因此具有计算效率。该方法在视网膜光学相干断层扫描（OCT）图像分割和磁共振成像（MRI）T1加权到T2加权图像合成上进行了验证。我们的方法，其短的优化时间的适配器（10次迭代在一个单一的测试主题）和它的额外所需的磁盘空间的自动编码器（约15 MB），可以实现显着的性能改善。我们的代码可在https://github.com/YufanHe/self-domain-adapted-network上公开获取。
Deep neural networks have been successfully applied to medical image analysis tasks like segmentation and synthesis. However, even if a network is trained on a large dataset from the source domain, its performance on unseen test domains is not guaranteed. The performance drop on data obtained differently from the network’s training data is a major problem (known as domain shift) in deploying deep learning in clinical practice. Existing work focuses on retraining the model with data from the test domain, or harmonizing the test domain’s data to the network training data. A common practice is to distribute a carefully-trained model to multiple users (e.g., clinical centers), and then each user uses the model to process their own data, which may have a domain shift (e.g., varying imaging parameters and machines). However, the lack of availability of the source training data and the cost of training a new model often prevents the use of known methods to solve user-specific domain shifts. Here, we ask whether we can design a model that, once distributed to users, can quickly adapt itself to each new site without expensive retraining or access to the source training data? In this paper, we propose a model that can adapt based on a single test subject during inference. The model consists of three parts, which are all neural networks: a task model (T) which performs the image analysis task like segmentation; a set of autoencoders (AEs); and a set of adaptors (As). The task model and autoencoders are trained on the source dataset and can be computationally expensive. In the deployment stage, the adaptors are trained to transform the test image and its features to minimize the domain shift as measured by the autoencoders’ reconstruction loss. Only the adaptors are optimized during the testing stage with a single test subject thus is computationally efficient. The method was validated on both retinal optical coherence tomography (OCT) image segmentation and magnetic resonance imaging (MRI) T1-weighted to T2-weighted image synthesis. Our method, with its short optimization time for the adaptors (10 iterations on a single test subject) and its additional required disk space for the autoencoders (around 15 MB), can achieve significant performance improvement. Our code is publicly available at: https://github.com/YufanHe/self-domain-adapted-network.
登录
查看更多内容
影响因子：
1.2
作者：
He, Yufan;Carass, Aaron;Prince, Jerry L.
通讯作者：
Prince, Jerry L.
使用完全卷积回归网络的视网膜OCT的结构化层表面分割。
影响因子：
10.9
作者：
He Y;Carass A;Liu Y;Jedynak BM;Solomon SD;Saidha S;Calabresi PA;Prince JL
通讯作者：
Prince JL
影响因子：
10.6
作者：
Dar, Salman U. H.;Yurt, Mahmut;Cukur, Tolga
通讯作者：
Cukur, Tolga
影响因子：
5.7
作者：
Carass A;Cuzzocreo JL;Han S;Hernandez-Castillo CR;Rasser PE;Ganz M;Beliveau V;Dolz J;Ben Ayed I;Desrosiers C;Thyreau B;Romero JE;Coupé P;Manjón JV;Fonov VS;Collins DL;Ying SH;Onyike CU;Crocetti D;Landman BA;Mostofsky SH;Thompson PM;Prince JL
通讯作者：
Prince JL