Benchmarking functional connectome-based predictive models for resting-state fMRI

Abstract

Functional connectomes reveal biomarkers of individual psychological or clinical traits. However, there is great variability in the analytic pipelines typically used to derive them from rest-fMRI cohorts. Here, we consider a specific type of studies, using predictive models on the edge weights of functional connectomes, for which we highlight the best modeling choices. We systematically study the prediction performances of models in 6 different cohorts and a total of 2 000 individuals, encompassing neuro-degenerative (Alzheimer's, Post-traumatic stress disorder), neuro-psychiatric (Schizophrenia, Autism), drug impact (Cannabis use) clinical settings and psychological trait (fluid intelligence). The typical prediction procedure from rest-fMRI consists of three main steps: defining brain regions, representing the interactions, and supervised learning. For each step we benchmark typical choices: 8 different ways of defining regions-either pre-defined or generated from the rest-fMRI data-3 measures to build functional connectomes from the extracted time-series, and 10 classification models to compare functional interactions across subjects. Our benchmarks summarize more than 240 different pipelines and outline modeling choices that show consistent prediction performances in spite of variations in the populations and sites. We find that regions defined from functional data work best; that it is beneficial to capture between-region interactions with tangent-based parametrization of covariances, a midway between correlations and partial correlation; and that simple linear predictors such as a logistic regression give the best predictions. Our work is a step forward to establishing reproducible imaging-based biomarkers for clinical settings.

Bootstrapping promotes the RSFC-behavior associations: an application of individual cognitive traits prediction

“…To have direct comparisons among different parcellations, the baseline CPM, SVR and LASSO methods were used to predict the cognitive functions with FC matrix under different parcellations. The results were illustrated in Figure 7, and the main tendency was that the prediction accuracy increased with dimension except LASSO, and grown slowly from 100 to 300 independent components, and similar trend was also revealed in other works Dadi et al, 2019). To investigate the bootstrapping methods on high-dimensional data, the analysis was based on 300 independent components.…”

“…SVR displayed the best performance among three models. Compared with LASSO, SVR performed slightly better and the similar trends were also revealed by Cui et al (2018) and Dadi et al (2019) on RSFC predictions. However, efficiency was a crucial problem for high-dimensional data while SVR was tremendously expensive in computation for model fitting (Cui et al, 2018;Shen et al, 2017).…”

“…Group ICA has been found to be a better way to define the functional brain nodes than the atlas-based method for its better prediction accuracy (Dadi et al, 2019), hence, group ICA parcellation was adopted to calculate the corresponding FC in the study.…”

“…For the regression tasks, correlation analysis between RSFC and the target phonotypic measure (e.g. cognitive trait) is the most common approach to select features (Dadi et al, 2019;Gabrieli et al, 2015). However, correlation analysis may not be able to catch the reliable correlational relationship since rs-fMRI data exhibits low signal-to-noise ratio (SNR) and correlation analysis is sensitive to outliers (R. Wilcox, 2004;R.…”

Nonreplication of functional connectivity differences in autism spectrum disorder across multiple sites and denoising strategies

“…One potential source of variability across rs‐fMRI studies has been the methods used for data preprocessing. The blood oxygenation‐level dependent (BOLD) signal, while sensitive to changes related to brain activity, is also highly vulnerable to head motion and physiological noise, which can spuriously influence measures of functional connectivity and ultimately affect conclusions from functional connectivity studies (Dadi et al, ; Power et al, ; Power, Barnes, Snyder, Schlaggar, & Petersen, ; Satterthwaite et al, ; Van Dijk, Sabuncu, & Buckner, ; Yan et al, ). Ideally, effective data preprocessing methods would minimize the influence of such nuisance signals and improve reproducibility.…”