Abstract
We present a practical “how-to” guide to help determine whether single-subject fMRI independent components (ICs) characterise structured noise or not. Manual identification of signal and noise after ICA decomposition is required for efficient data denoising: to train supervised algorithms, to check the results of unsupervised ones or to manually clean the data. In this paper we describe the main spatial and temporal features of ICs and provide general guidelines on how to evaluate these. Examples of signal and noise components are provided from a wide range of datasets (3T data, including examples from the UK Biobank and the Human Connectome Project, and 7T data), together with practical guidelines for their identification. Finally, we discuss how the data quality, data type and preprocessing can influence the characteristics of the ICs and present examples of particularly challenging datasets.
Highlights
Spatial Independent Component Analysis (ICA) has proven to be a powerful tool for blind source separation of fMRI data (Beckmann and Smith, 2004; Hyvarinen, 1999; McKeown et al, 1998) into 3D spatial maps and 1D time courses
FMRI data contain both structured and stochastic noise, in the context of ICA-based denoising the term noise refers to just the structured noise, given that ICA decomposition aims at un-mixing the data into non-Gaussian sources
The decision process we suggest to classify a component as signal is: 1) spatial maps first and foremost need to be plausible, i.e. be located in grey matter (GM), away from the main veins, white matter (WM) or cerebrospinal fluid (CSF); 2) time series should be without sudden, abrupt changes; 3) power spectra should largely show power at low frequency
Summary
Spatial Independent Component Analysis (ICA) has proven to be a powerful tool for blind source separation of fMRI data (Beckmann and Smith, 2004; Hyvarinen, 1999; McKeown et al, 1998) into 3D spatial maps and 1D time courses. A method that uses the temporal information without task-related information on single-echo data has been proposed by Thomas and colleagues (Thomas et al, 2002) This approach identifies (random or structured) noise components using an unsupervised algorithm that examines the Fourier decomposition of the time series. A rule of thumb is that a component should be kept in the data, unless it is clearly artefactual (“innocent until proven guilty”) This highlights the value in identifying general rules and features for signal and noise components, more broadly than enforcing that single-subject RSNs have a strong spatial match to known (typically group-level-derived) RSNs. The two main noise categories are noise related to the subject (motion/physiological effects) or related to the acquisition (MR physics artefacts). We discuss some of the factors influencing the ICs and present examples of challenging applications
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.