Independent Component Analysis (ICA) is a computational technique for identifying hidden statistically independent sources from multivariate data. In its basic form, ICA decomposes a 2D data matrix ...(e.g. time×voxels) into separate components that have distinct characteristics. In FMRI it is used to identify hidden FMRI signals (such as activations). Since the first application of ICA to Functional Magnetic Resonance Imaging (FMRI) in 1998, this technique has developed into a powerful tool for data exploration in cognitive and clinical neurosciences. In this contribution to the commemorative issue 20years of FMRI I will briefly describe the basic principles behind ICA, discuss the probabilistic extension to ICA and touch on what I think are some of the most notorious loose ends. Further, I will describe some of the most powerful ‘killer’ applications and finally share some thoughts on where I believe the most promising future developments will lie.
Abstract Background Despite many successes, the case-control approach is problematic in biomedical science: It introduces an artificial symmetry whereby all clinical groups (e.g. patients and ...controls) are assumed to be well-defined, when biologically they are often highly heterogeneous. By definition, it also precludes inference over the validity of the diagnostic labels. In response, the NIMH Research Domain Criteria (RDoC) proposes to map relationships between symptom dimensions and broad behavioural and biological domains, cutting across diagnostic categories. To date, however, RDoC has prompted few methods to meaningfully stratify clinical cohorts. Methods We introduce normative modelling for parsing heterogeneity in clinical cohorts while allowing predictions at an individual subject level. This aims to map variation within the cohort and is distinct from, and complementary to, existing approaches that tackle heterogeneity by employing clustering techniques to fractionate cohorts. To demonstrate this approach, we map the relationship between trait impulsivity and reward-related brain activity in a large healthy cohort (N=491). Results We identify participants that are outliers within this distribution and show that the degree of deviation (outlier magnitude) relates to specific attention deficit hyperactivity disorder symptoms (hyperactivity, but not inattention) on the basis of individualized patterns of abnormality. Conclusions Normative modelling provides a natural framework to study disorders at the individual participant level without dichotomizing the cohort. Instead, disease can be considered as an extreme of the normal range or as – possibly idiosyncratic – deviation from normal functioning. It also enables inferences over the degree to which behavioural variables – including diagnostic labels – map onto biology.
Understanding the fundamental organisation of the brain in terms of functional specialisation and integration is one of the principal aims of imaging neuroscience. Many investigations into the ...functional organisation of the brain are predicated on parcellating the brain into patches of assumed piece-wise constant connectivity. There are, however, many brain areas where the assumption of piece-wise constant organisation is violated. Connectivity, and by extension function, often varies continuously across the grey matter according to multiple overlapping modes of change. The organisation is governed by functional heterogeneity (continuous change) as well as functional multiplicity (overlapping modes). Functional heterogeneity and multiplicity have important implications for how we can and should analyse our data and how we ought to interpret the results, both in the classical context of parcellated modes and under models that allow for overlapping modes of continuous change. The goal of this opinion paper is to raise awareness of these issues and highlight recent methodological developments toward accounting for these important fundamental features of brain organisation.
•A lightweight deep learning model, Simple Fully Convolutional Network (SFCN), is presented, achieving state-of-the-art brain age prediction and sex classification performance in UK Biobank MRI brain ...imaging data.•Even with limited number of training subjects (e.g., 50), SFCN performs better than widely-used regression models.•A semi-multimodal ensemble strategy is proposed and achieved first place in the PAC 2019 brain age prediction challenge.•Linear regression can remove brain age prediction bias (even on unlabelled data) while maintaining state-of-the-art performance.
Deep learning has huge potential for accurate disease prediction with neuroimaging data, but the prediction performance is often limited by training-dataset size and computing memory requirements. To address this, we propose a deep convolutional neural network model, Simple Fully Convolutional Network (SFCN), for accurate prediction of brain age using T1-weighted structural MRI data. Compared with other popular deep network architectures, SFCN has fewer parameters, so is more compatible with small dataset size and 3D volume data. The network architecture was combined with several techniques for boosting performance, including data augmentation, pre-training, model regularization, model ensemble and prediction bias correction. We compared our overall SFCN approach with several widely-used machine learning models. It achieved state-of-the-art performance in UK Biobank data (N = 14,503), with mean absolute error (MAE) = 2.14y in brain age prediction and 99.5% in sex classification. SFCN also won (both parts of) the 2019 Predictive Analysis Challenge for brain age prediction, involving 79 competing teams (N = 2,638, MAE = 2.90y). We describe here the details of our approach, and its optimisation and validation. Our approach can easily be generalised to other tasks using different image modalities, and is released on GitHub.
Many sources of fluctuation contribute to the fMRI signal, and this makes identifying the effects that are truly related to the underlying neuronal activity difficult. Independent component analysis ...(ICA) – one of the most widely used techniques for the exploratory analysis of fMRI data – has shown to be a powerful technique in identifying various sources of neuronally-related and artefactual fluctuation in fMRI data (both with the application of external stimuli and with the subject “at rest”). ICA decomposes fMRI data into patterns of activity (a set of spatial maps and their corresponding time series) that are statistically independent and add linearly to explain voxel-wise time series. Given the set of ICA components, if the components representing “signal” (brain activity) can be distinguished form the “noise” components (effects of motion, non-neuronal physiology, scanner artefacts and other nuisance sources), the latter can then be removed from the data, providing an effective cleanup of structured noise. Manual classification of components is labour intensive and requires expertise; hence, a fully automatic noise detection algorithm that can reliably detect various types of noise sources (in both task and resting fMRI) is desirable. In this paper, we introduce FIX (“FMRIB's ICA-based X-noiseifier”), which provides an automatic solution for denoising fMRI data via accurate classification of ICA components. For each ICA component FIX generates a large number of distinct spatial and temporal features, each describing a different aspect of the data (e.g., what proportion of temporal fluctuations are at high frequencies). The set of features is then fed into a multi-level classifier (built around several different classifiers). Once trained through the hand-classification of a sufficient number of training datasets, the classifier can then automatically classify new datasets. The noise components can then be subtracted from (or regressed out of) the original data, to provide automated cleanup. On conventional resting-state fMRI (rfMRI) single-run datasets, FIX achieved about 95% overall accuracy. On high-quality rfMRI data from the Human Connectome Project, FIX achieves over 99% classification accuracy, and as a result is being used in the default rfMRI processing pipeline for generating HCP connectomes. FIX is publicly available as a plugin for FSL.
Psychiatric disorders are increasingly being recognised as having a biological basis, but their diagnosis is made exclusively behaviourally. A promising approach for 'biomarker' discovery has been ...based on pattern recognition methods applied to neuroimaging data, which could yield clinical utility in future. In this review we survey the literature on pattern recognition for making diagnostic predictions in psychiatric disorders, and evaluate progress made in translating such findings towards clinical application. We evaluate studies on many criteria, including data modalities used, the types of features extracted and algorithm applied. We identify problems common to many studies, such as a relatively small sample size and a primary focus on estimating generalisability within a single study. Furthermore, we highlight challenges that are not widely acknowledged in the field including the importance of accommodating disease prevalence, the necessity of more extensive validation using large carefully acquired samples, the need for methodological innovations to improve accuracy and to discriminate between multiple disorders simultaneously. Finally, we identify specific clinical contexts in which pattern recognition can add value in the short to medium term.
Major depressive disorder (MDD) affects multiple large-scale functional networks in the brain, which has initiated a large number of studies on resting-state functional connectivity in depression. We ...review these recent studies using either seed-based correlation or independent component analysis and propose a model that incorporates changes in functional connectivity within current hypotheses of network-dysfunction in MDD. Although findings differ between studies, consistent findings include: (1) increased connectivity within the anterior default mode network, (2) increased connectivity between the salience network and the anterior default mode network, (3) changed connectivity between the anterior and posterior default mode network and (4) decreased connectivity between the posterior default mode network and the central executive network. These findings correspond to the current understanding of depression as a network-based disorder.
The posterior cingulate cortex (PCC) is a central part of the default mode network (DMN) and part of the structural core of the brain. Although the PCC often shows consistent deactivation when ...attention is focused on external events, anatomical studies show that the region is not homogeneous, and electrophysiological recordings in nonhuman primates suggest that it is directly involved in some forms of attention. We report a functional magnetic resonance imaging study of an attentionally demanding task (either a zero- or two-back working memory task). Standard subtraction analysis within the PCC shows a relative deactivation as task difficulty increases. In contrast, a dual-regression functional connectivity analysis reveals a clear dissociation between ventral and dorsal parts of the PCC. As task difficulty increases, the ventral PCC shows reduced integration within the DMN and less anticorrelation with the cognitive control network (CCN) activated by the task. The dorsal PCC shows an opposite pattern, with increased DMN integration and more anticorrelation. At rest, the dorsal PCC also shows functional connectivity with both the DMN and attentional networks. As expected, these results provide evidence that the PCC is involved in supporting internally directed thought, as the region is more highly integrated with the DMN at low task demands. In contrast, the task-dependent increases in connectivity between the dorsal PCC and the CCN are consistent with a role for this region in modulating the dynamic interaction between these two networks controlling the efficient allocation of attention.
We proposed ICA-AROMA as a strategy for the removal of motion-related artifacts from fMRI data (Pruim et al., 2015). ICA-AROMA automatically identifies and subsequently removes data-driven derived ...components that represent motion-related artifacts. Here we present an extensive evaluation of ICA-AROMA by comparing our strategy to a range of alternative strategies for motion-related artifact removal: (i) no secondary motion correction, (ii) extensive nuisance regression utilizing 6 or (iii) 24 realignment parameters, (iv) spike regression (Satterthwaite et al., 2013a), (v) motion scrubbing (Power et al., 2012), (vi) aCompCor (Behzadi et al., 2007; Muschelli et al., 2014), (vii) SOCK (Bhaganagarapu et al., 2013), and (viii) ICA-FIX (Griffanti et al., 2014; Salimi-Khorshidi et al., 2014), without re-training the classifier. Using three different functional connectivity analysis approaches and four different multi-subject resting-state fMRI datasets, we assessed all strategies regarding their potential to remove motion artifacts, ability to preserve signal of interest, and induced loss in temporal degrees of freedom (tDoF). Results demonstrated that ICA-AROMA, spike regression, scrubbing, and ICA-FIX similarly minimized the impact of motion on functional connectivity metrics. However, both ICA-AROMA and ICA-FIX resulted in significantly improved resting-state network reproducibility and decreased loss in tDoF compared to spike regression and scrubbing. In comparison to ICA-FIX, ICA-AROMA yielded improved preservation of signal of interest across all datasets. These results demonstrate that ICA-AROMA is an effective strategy for removing motion-related artifacts from rfMRI data. Our robust and generalizable strategy avoids the need for censoring fMRI data and reduces motion-induced signal variations in fMRI data, while preserving signal of interest and increasing the reproducibility of functional connectivity metrics. In addition, ICA-AROMA preserves the temporal non-artifactual time-series characteristics and limits the loss in tDoF, thereby increasing statistical power at both the subject- and the between-subject analysis level.
Head motion during functional MRI (fMRI) scanning can induce spurious findings and/or harm detection of true effects. Solutions have been proposed, including deleting ('scrubbing') or regressing out ...('spike regression') motion volumes from fMRI time-series. These strategies remove motion-induced signal variations at the cost of destroying the autocorrelation structure of the fMRI time-series and reducing temporal degrees of freedom. ICA-based fMRI denoising strategies overcome these drawbacks but typically require re-training of a classifier, needing manual labeling of derived components (e.g. ICA-FIX; Salimi-Khorshidi et al. (2014)). Here, we propose an ICA-based strategy for Automatic Removal of Motion Artifacts (ICA-AROMA) that uses a small (n=4), but robust set of theoretically motivated temporal and spatial features. Our strategy does not require classifier re-training, retains the data's autocorrelation structure and largely preserves temporal degrees of freedom. We describe ICA-AROMA, its implementation, and initial validation. ICA-AROMA identified motion components with high accuracy and robustness as illustrated by leave-N-out cross-validation. We additionally validated ICA-AROMA in resting-state (100 participants) and task-based fMRI data (118 participants). Our approach removed (motion-related) spurious noise from both rfMRI and task-based fMRI data to larger extent than regression using 24 motion parameters or spike regression. Furthermore, ICA-AROMA increased sensitivity to group-level activation. Our results show that ICA-AROMA effectively reduces motion-induced signal variations in fMRI data, is applicable across datasets without requiring classifier re-training, and preserves the temporal characteristics of the fMRI data.