Handling Multiplicity in Neuroimaging Through Bayesian Lenses with Multilevel Modeling

Padmala

et al. 2020

Preprint

Self Cite

In this work, we investigate the importance of explicitly accounting for cross-trial variability in neuroimaging data analysis. To attempt to obtain reliable estimates in a task-based experiment, each condition is usually repeated across many trials. The investigator may be interested in (a) condition-level effects, (b) trial-level effects, or (c) the association of trial-level effects with the corresponding behavior data. The typical strategy for condition-level modeling is to create one regressor per condition at the subject level with the underlying assumption that responses do not change across trials. In this methodology of complete pooling, all cross-trial variability is ignored and dismissed as random noise that is swept under the rug of model residuals. Unfortunately, this framework invalidates the generalizability from the confine of specific trials (e.g., particular faces) to the associated stimulus category ("face"). Here we propose an adaptive and computationally tractable framework that meshes well with the current two-level pipeline and explicitly accounts for trial-by-trial variability. The trial-level effects are first estimated per subject through no pooling. To allow generalizing beyond the particular stimulus set employed, the cross-trial variability is modeled at the population level through partial pooling in a multilevel model, which permits accurate effect estimation and characterization. Alternatively, trial-level estimates can be used to investigate, for example, brain-behavior associations or correlations between brain regions. Furthermore, our approach allows appropriate accounting for serial correlation, handling outliers, adapting to data skew, and capturing nonlinear brain-behavior relationships. By applying a Bayesian multilevel model framework at the level of regions of interest to an experimental dataset, we show how multiple testing can be addressed and full results reported without arbitrary dichotomization. Our approach revealed important differences compared to the conventional method at the condition level, including how the latter can distort effect magnitude and precision.Notably, in some cases our approach led to increased statistical sensitivity. In summary, our proposed framework provides an effective strategy to capture trial-by-trial responses that should be of interest to a wide community of experimentalists. and so on. The reliability of a task-based experiment hinges on having a reasonably large number of repetitions associated with a condition. Such repetitions are usually termed "trials", and each trial is considered to be an instantiation of an idealized condition. For example, in an emotion study with three conditions (positive, neutral and negative), the investigator may show 20 different human faces of each emotional valence to the subject in the scanner. From the statistical perspective, the number of trials serves as the sample size for each condition and, per the law of large numbers in probability theory, the average effect estimate for a specific conditio...

Section: Population Analysis Through Trial-level Modelingmentioning

confidence: 99%

mentioning

confidence: 99%

To pool or not to pool: Can we ignore cross-trial variability in FMRI?

Padmala

et al. 2020

Preprint

Self Cite

“…Given the multiplicity of ROIs, investigators commonly perform some correction for multiple comparisons, say via Bonferroni correction. Here, we analyzed the interaction between valence and context by performing a Bayesian multilevel analysis of the ROI data by using the Region-Based Analysis (RBA) program of the AFNI suite (Chen et al, 2019b). In this approach, the data from all ROIs are included in a single multilevel model that evaluates the effects of interest.…”

Section: Region-based Bayesian Analysismentioning

confidence: 99%

“…For summary purposes, posteriors of the effects for every ROI can be plotted separately; but they are not independent and technically are simply marginal distributions (that is, projections along particular variables). For formal details of the approach adopted here, please refer to Chen et al (2019b); for a less technical exposition, see Chen et al (2019a).…”

Section: Region-based Bayesian Analysismentioning

confidence: 99%

Interactions between emotion and action in the brain

Portugal

Alves

Fernandes

et al. 2019

Preprint

Self Cite

A growing literature supports the existence of interactions between emotion and action in the brain, and the central participation of the anterior midcingulate cortex (aMCC) in this regard. In the present functional magnetic resonance imaging study, we sought to investigate the role of self-relevance during such interactions by varying the context in which threating pictures were presented (with guns pointed towards or away from the observer). Participants performed a simple visual detection task following exposure to such stimuli. Except for voxelwise tests, we adopted a Bayesian analysis framework which evaluated evidence for the hypotheses of interest, given the data, in a continuous fashion. Behaviorally, our results demonstrated a valence by context interaction such that there was a tendency of speeding up responses to targets after viewing threat pictures directed towards the participant. In the brain, interaction patterns that paralleled those observed behaviorally were observed most notably in the middle temporal gyrus, supplementary motor area, precentral gyrus, and anterior insula. In these regions, activity was overall greater during threat conditions relative to neutral ones, and this effect was enhanced in the directed towards context. A valence by context interaction was observed in the aMCC too, where we also observed a correlation (across participants) of evoked responses and reaction time data. Taken together, our study revealed the context-sensitive engagement of motor-related areas during emotional perception, thus supporting the idea that emotion and action interact in important ways in the brain. ______________________________________________________________________________________________________

“…Having separate testings for different brain areas in a mass-univariate fashion also places an assumption about independency among brain areas, further constraining its inference. 13…”

Section: Introductionmentioning

confidence: 99%

Explainable Machine Learning Approach to Predict and Explain the Relationship between Task-based fMRI and Individual Differences in Cognition

Pornpattananangkul

Bartonicek

Wang

et al. 2020

Preprint

Predicting individual differences in cognitive processes is crucial, but the ability of task-based fMRI to do so remains dubious, despite decades of costly research. We tested the ability of working-memory fMRI in predicting working-memory, using the Adolescent Brain Cognitive Development (n = 4,350). The conventionally-used mass-univariate approach led to poor out-of-sample prediction (Mean r = .1-.12). However, the multivariate Elastic Net, which draws information across brain regions, enhanced out-of-sample prediction (r = .47) by several folds. The Elastic Net also enabled us to predict cognitive performance from various tasks collected outside of the scanner, highlighting its generalizability. Moreover, using an omics-inspired approach, we combined Elastic Net with permutation, allowing us to statistically infer which brain regions contribute to individual differences while accounting for collinearity. Accordingly, our framework can build an easy-to-interpret predictive fMRI model that transfers knowledge learned from large-scale datasets to smaller samples, akin to polygenic scores in genomics.