NeuroCrypt: Machine Learning Over Encrypted Distributed Neuroimaging Data.
NeuroCrypt: Machine Learning Over Encrypted Distributed Neuroimaging Data.
复制标题
DOI：
10.1007/s12021-021-09525-8
复制
发表时间：
2022-01
期刊：
影响因子：
3
通讯作者：
Plis SM
中科院分区：
文献类型：
--
作者：
Senanayake N;Podschwadt R;Takabi D;Calhoun VD;Plis SM
神经影像领域可以从建立机器学习模型中受益匪浅，以发现和预测疾病，并发现新颖的生物标志物，但是由于隐私或监管问题，在各个组织和研究中心收集的许多数据都无法共享（尤其是对于临床方面）数据或罕见疾病）。此外，在多个大型研究中汇总数据会导致大量重复的技术债务，并且所需的资源对于单个站点而言是不可能的。对跨组织分发的数据进行培训可能会导致模型比仅在任何组织的数据上培训的模型都要好得多。尽管有分散共享的方法，但这些方法通常并不能提供只能提供的最高样本隐私保证的保证。此外，这种方法通常集中在概率解决方案上。在本文中，我们提出了一种方法，该方法通过以安全和确定性的方式进行联合分析来利用数据集的潜力扩散在许多数据收集组织之间，仅当仅共享和操纵加密的数据时。该方法基于安全的多方计算，该计算指的是密码协议，该协议可以在分布式输入上启用函数的分布式计算，而无需揭示有关输入的其他信息。它使多个组织能够在其联合数据上训练机器学习模型，并将经过训练的模型应用于加密数据，而无需向其他方揭示其敏感数据。在我们提出的方法中，组织（或站点）安全地合作建立了机器学习模型，因为它本来可以对所有组织的汇总数据进行培训。重要的是，该方法不需要一个受信任的方（即聚合者），每个贡献站点在过程中起着同等的作用，并且任何网站都无法学习任何其他网站的个人数据。我们使用不同的机器学习算法（包括逻辑回归和卷积神经网络模型）在人类结构和功能磁共振成像数据集（包括逻辑回归和卷积神经网络模型）中进行了一系列经验评估，证明了所提出的方法的有效性。
The field of neuroimaging can greatly benefit from building machine learning models to detect and predict diseases, and discover novel biomarkers, but much of the data collected at various organizations and research centers is unable to be shared due to privacy or regulatory concerns (especially for clinical data or rare disorders). In addition, aggregating data across multiple large studies results in a huge amount of duplicated technical debt and the resources required can be challenging or impossible for an individual site to build. Training on the data distributed across organizations can result in models that generalize much better than models trained on data from any of organizations alone. While there are approaches for decentralized sharing, these often do not provide the highest possible guarantees of sample privacy that only cryptography can provide. In addition, such approaches are often focused on probabilistic solutions. In this paper, we propose an approach that leverages the potential of datasets spread among a number of data collecting organizations by performing joint analyses in a secure and deterministic manner when only encrypted data is shared and manipulated. The approach is based on secure multiparty computation which refers to cryptographic protocols that enable distributed computation of a function over distributed inputs without revealing additional information about the inputs. It enables multiple organizations to train machine learning models on their joint data and apply the trained models to encrypted data without revealing their sensitive data to the other parties. In our proposed approach, organizations (or sites) securely collaborate to build a machine learning model as it would have been trained on the aggregated data of all the organizations combined. Importantly, the approach does not require a trusted party (i.e. aggregator), each contributing site plays an equal role in the process, and no site can learn individual data of any other site. We demonstrate effectiveness of the proposed approach, in a range of empirical evaluations using different machine learning algorithms including logistic regression and convolutional neural network models on human structural and functional magnetic resonance imaging datasets.
登录
查看更多内容
神经影像学中脑部疾病的单受试者预测：前景和陷阱
影响因子：
5.7
作者：
Arbabshirani MR;Plis S;Sui J;Calhoun VD
通讯作者：
Calhoun VD
影响因子：
3
作者：
Aine CJ;Bockholt HJ;Bustillo JR;Cañive JM;Caprihan A;Gasparovic C;Hanlon FM;Houck JM;Jung RE;Lauriello J;Liu J;Mayer AR;Perrone-Bizzozero NI;Posse S;Stephen JM;Turner JA;Clark VP;Calhoun VD
通讯作者：
Calhoun VD
影响因子：
5.7
作者：
Fischl, Bruce
通讯作者：
Fischl, Bruce
影响因子：
4.8
作者：
Fu Z;Sui J;Turner JA;Du Y;Assaf M;Pearlson GD;Calhoun VD
通讯作者：
Calhoun VD
影响因子：
8.7
作者：
Gupta, Otkrist;Raskar, Ramesh
通讯作者：
Raskar, Ramesh