Toward a unified framework for interpreting machine-learning models in neuroimaging.
Toward a unified framework for interpreting machine-learning models in neuroimaging.
复制标题
DOI：
10.1038/s41596-019-0289-5
复制
发表时间：
2020-04
期刊：
影响因子：
14.8
通讯作者：
Woo, Choong-Wan
中科院分区：
文献类型：
--
作者：
Kohoutova, Lada;Heo, Juyeon;Cha, Sungmin;Lee, Sungwoo;Moon, Taesup;Wager, Tor D.;Woo, Choong-Wan
机器学习是创建将大脑功能与行为联系起来的计算模型的强大工具，它在神经科学中的应用越来越广泛。然而，这些模型很复杂，通常很难解释，因此很难评估它们的神经科学有效性和对理解大脑的贡献。为了使基于神经成像的机器学习模型具有可解释性，它们应该（i）对人类来说是可理解的，（ii）提供关于在特定的大脑通路或区域中表示哪些心理或行为结构的有用信息，以及（iii）证明它们是基于相关的神经生物学信号，而不是伪影或混淆。在本协议中，我们引入了一个统一的框架，包括模型，功能和生物学水平的评估，以提供互补的结果，支持理解模型如何以及为什么工作。虽然该框架可以应用于不同类型的模型和数据，但该协议为功能性MRI数据集和基于多变量模式的预测模型提供了实用工具和选定分析方法的示例。该协议的用户应该熟悉MATLAB或Python中的基本编程。该协议将有助于构建更可解释的基于神经成像的机器学习模型，有助于对大脑机制和大脑健康的累积理解。虽然这里提供的分析构成了一组有限的测试，需要几个小时到几天才能完成，这取决于数据的大小和可用的计算资源，我们设想注释和解释模型的过程是一个开放式的过程，涉及多个研究和实验室的合作努力。
Machine learning is a powerful tool for creating computational models relating brain function to behavior, and its use is becoming widespread in neuroscience. However, these models are complex and often hard to interpret, making it difficult to evaluate their neuroscientific validity and contribution to understanding the brain. For neuroimaging-based machine-learning models to be interpretable, they should (i) be comprehensible to humans, (ii) provide useful information about what mental or behavioral constructs are represented in particular brain pathways or regions, and (iii) demonstrate that they are based on relevant neurobiological signal, not artifacts or confounds. In this protocol, we introduce a unified framework that consists of model-, feature- and biology-level assessments to provide complementary results that support the understanding of how and why a model works. Although the framework can be applied to different types of models and data, this protocol provides practical tools and examples of selected analysis methods for a functional MRI dataset and multivariate pattern-based predictive models. A user of the protocol should be familiar with basic programming in MATLAB or Python. This protocol will help build more interpretable neuroimaging-based machine-learning models, contributing to the cumulative understanding of brain mechanisms and brain health. Although the analyses provided here constitute a limited set of tests and take a few hours to days to complete, depending on the size of data and available computational resources, we envision the process of annotating and interpreting models as an open-ended process, involving collaborative efforts across multiple studies and laboratories.
登录
查看更多内容
影响因子：
4.3
作者：
Baldassarre L;Pontil M;Mourão-Miranda J
通讯作者：
Mourão-Miranda J
影响因子：
5.3
作者：
Baliki, Marwan N.;Mansour, Ali;Apkarian, A. Vania
通讯作者：
Apkarian, A. Vania
影响因子：
4.7
作者：
Casey BJ;Cannonier T;Conley MI;Cohen AO;Barch DM;Heitzeg MM;Soules ME;Teslovich T;Dellarco DV;Garavan H;Orr CA;Wager TD;Banich MT;Speer NK;Sutherland MT;Riedel MC;Dick AS;Bjork JM;Thomas KM;Chaarani B;Mejia MH;Hagler DJ Jr;Daniela Cornejo M;Sicat CS;Harms MP;Dosenbach NUF;Rosenberg M;Earl E;Bartsch H;Watts R;Polimeni JR;Kuperman JM;Fair DA;Dale AM;ABCD Imaging Acquisition Workgroup
通讯作者：
ABCD Imaging Acquisition Workgroup
影响因子：
5.7
作者：
Brodersen, Kay H.;Wiech, Katja;Lomakina, Ekaterina I.;Lin, Chia-shu;Buhmann, Joachim M.;Bingel, Ulrike;Ploner, Markus;Stephan, Klaas Enno;Tracey, Irene
通讯作者：
Tracey, Irene
影响因子：
3.3
作者：
Carrasquillo Y;Gereau RW 4th
通讯作者：
Gereau RW 4th