Multiple levels of contextual influence on action-based timing behavior and cortical activation.

Procedures used to elicit both behavioral and neurophysiological data to address a particular cognitive question can impact the nature of the data collected. We used functional near-infrared spectroscopy (fNIRS) to assess performance of a modified finger tapping task in which participants performed synchronized or syncopated tapping relative to a metronomic tone. Both versions of the tapping task included a pacing phase (tapping with the tone) followed by a continuation phase (tapping without the tone). Both behavioral and brain-based findings revealed two distinct timing mechanisms underlying the two forms of tapping. Here we investigate the impact of an additional-and extremely subtle-manipulation of the study's experimental design. We measured responses in 23 healthy adults as they performed the two versions of the finger-tapping tasks either blocked by tapping type or alternating from one to the other type during the course of the experiment. As in our previous study, behavioral tapping indices and cortical hemodynamics were monitored, allowing us to compare results across the two study designs. Consistent with previous findings, results reflected distinct, context-dependent parameters of the tapping. Moreover, our results demonstrated a significant impact of study design on rhythmic entrainment in the presence/absence of auditory stimuli. Tapping accuracy and hemodynamic responsivity collectively indicate that the block design context is preferable for studying action-based timing behavior.

Correction of global physiology in resting-state functional near-infrared spectroscopy.

Significance: Resting-state functional connectivity (RSFC) analyses of functional near-infrared spectroscopy (fNIRS) data reveal cortical connections and networks across the brain. Motion artifacts and systemic physiology in evoked fNIRS signals present unique analytical challenges, and methods that control for systemic physiological noise have been explored. Whether these same methods require modification when applied to resting-state fNIRS (RS-fNIRS) data remains unclear. Aim: We systematically examined the sensitivity and specificity of several RSFC analysis pipelines to identify the best methods for correcting global systemic physiological signals in RS-fNIRS data. Approach: Using numerically simulated RS-fNIRS data, we compared the rates of true and false positives for several connectivity analysis pipelines. Their performance was scored using receiver operating characteristic analysis. Pipelines included partial correlation and multivariate Granger causality, with and without short-separation measurements, and a modified multivariate causality model that included a non-traditional zeroth-lag cross term. We also examined the effects of pre-whitening and robust statistical estimators on performance. Results: Consistent with previous work on bivariate correlation models, our results demonstrate that robust statistics and pre-whitening are effective methods to correct for motion artifacts and autocorrelation in the fNIRS time series. Moreover, we found that pre-filtering using principal components extracted from short-separation fNIRS channels as part of a partial correlation model was most effective in reducing spurious correlations due to shared systemic physiology when the two signals of interest fluctuated synchronously. However, when there was a temporal lag between the signals, a multivariate Granger causality test incorporating the short-separation channels was better. Since it is unknown if such a lag exists in experimental data, we propose a modified version of Granger causality that includes the non-traditional zeroth-lag term as a compromising solution. Conclusions: A combination of pre-whitening, robust statistical methods, and partial correlation in the processing pipeline to reduce autocorrelation, motion artifacts, and global physiology are suggested for obtaining statistically valid connectivity metrics with RS-fNIRS. Further studies should validate the effectiveness of these methods using human data.

MALINI (Machine Learning in NeuroImaging): A MATLAB toolbox for aiding clinical diagnostics using resting-state fMRI data.

Resting-state functional Magnetic Resonance Imaging (rs-fMRI) has been extensively used for diagnostic classification because it does not require task compliance and is easier to pool data from multiple imaging sites, thereby increasing the sample size. A MATLAB-based toolbox called Machine Learning in NeuroImaging (MALINI) for feature extraction and disease classification is presented. The MALINI toolbox extracts functional and effective connectivity features from preprocessed rs-fMRI data and performs classification between healthy and disease groups using any of 18 popular and widely used machine learning algorithms that are based on diverse principles. A consensus classifier combining the power of multiple classifiers is also presented. The utility of the toolbox is illustrated by accompanying data consisting of resting-state functional connectivity features from healthy controls and subjects with various brain-based disorders: autism spectrum disorder from autism brain imaging data exchange (ABIDE), Alzheimer's disease and mild cognitive impairment from Alzheimer's disease neuroimaging initiative (ADNI), attention deficit hyperactivity disorder from ADHD-200, and post-traumatic stress disorder and post-concussion syndrome acquired in-house. Results of classification performed on the above datasets can be obtained from the main article titled "Supervised machine learning for diagnostic classification from large-scale neuroimaging datasets" [1]. The data was divided into homogeneous and heterogeneous splits, such that 80% could be used for training, model building and cross-validation, while the remaining 20% of the data could be used as a hold-out independent test data for replication of the classification performance, to ensure the robustness of the classifiers to population variance in image acquisition site and age of the sample.

Supervised machine learning for diagnostic classification from large-scale neuroimaging datasets.

There are growing concerns about the generalizability of machine learning classifiers in neuroimaging. In order to evaluate this aspect across relatively large heterogeneous populations, we investigated four disorders: Autism spectrum disorder (N = 988), Attention deficit hyperactivity disorder (N = 930), Post-traumatic stress disorder (N = 87) and Alzheimer's disease (N = 132). We applied 18 different machine learning classifiers (based on diverse principles) wherein the training/validation and the hold-out test data belonged to samples with the same diagnosis but differing in either the age range or the acquisition site. Our results indicate that overfitting can be a huge problem in heterogeneous datasets, especially with fewer samples, leading to inflated measures of accuracy that fail to generalize well to the general clinical population. Further, different classifiers tended to perform well on different datasets. In order to address this, we propose a consensus-classifier by combining the predictive power of all 18 classifiers. The consensus-classifier was less sensitive to unmatched training/validation and holdout test data. Finally, we combined feature importance scores obtained from all classifiers to infer the discriminative ability of connectivity features. The functional connectivity patterns thus identified were robust to the classification algorithm used, age and acquisition site differences, and had diagnostic predictive ability in addition to univariate statistically significant group differences between the groups. A MATLAB toolbox called Machine Learning in NeuroImaging (MALINI), which implements all the 18 different classifiers along with the consensus classifier is available from Lanka et al. (2019) The toolbox can also be found at the following URL: https://github.com/pradlanka/malini .

Combining Prospective Acquisition CorrEction (PACE) with retrospective correction to reduce motion artifacts in resting state fMRI data.

BACKGROUND: Head movement in the scanner causes spurious signal changes in the blood-oxygen-level-dependent (BOLD) signal, confounding resting state functional connectivity (RSFC) estimates obtained from functional magnetic resonance imaging (fMRI). We examined the effectiveness of Prospective Acquisition CorrEction (PACE) in reducing motion artifacts in BOLD data. METHODS: Using PACE-corrected RS-fMRI data obtained from 44 subjects and subdividing them into low- and high-motion cohorts, we investigated voxel-wise motion-BOLD relationships, the distance-dependent functional connectivity artifact and the correlation between head motion and connectivity metrics such as posterior cingulate seed-based connectivity and network degree centrality. RESULTS: Our results indicate that, when PACE is used in combination with standard retrospective motion correction strategies, it provides two principal advantages over conventional echo-planar imaging (EPI) RS-fMRI data: (a) PACE was effective in eliminating significant negative motion-BOLD relationships, shown to be associated with signal dropouts caused by head motion, and (b) Censoring with a lower threshold (framewise displacement >0.5 mm) and a smaller window around the motion corrupted time point provided qualitatively equivalent reductions in the motion artifact with PACE when compared to a more conservative threshold of 0.2 mm required with conventional EPI data. CONCLUSIONS: PACE when used in conjunction with retrospective motion correction methods including nuisance signal and motion parameter regression, and censoring, did prove effective in almost eliminating head motion artifacts, even with a lower censoring threshold. Use of a lower censoring threshold could provide substantial savings in data that would otherwise be lost to censoring. Three-dimensional PACE has negligible overhead in terms of scan time, sequence modifications or additional and hence presents an attractive option for head motion correction in high-throughput resting-state BOLD imaging.

Effective connectivity within a triple network brain system discriminates schizophrenia spectrum disorders from psychotic bipolar disorder at the single-subject level.

OBJECTIVE: Schizophrenia spectrum disorders (SSD) and psychotic bipolar disorder share a number of genetic and neurobiological features, despite a divergence in clinical course and outcome trajectories. We studied the diagnostic classification potential that can be achieved on the basis of the structure and connectivity within a triple network system (the default mode, salience and central executive network) in patients with SSD and psychotic bipolar disorder. METHODS: Directed static connectivity and its dynamic variance was estimated among 8 nodes of the three large-scale networks. Multivariate autoregressive models of deconvolved resting state functional magnetic resonance imaging time series were obtained from 57 patients (38 with SSD and 19 with bipolar disorder and psychosis). We used 2/3 of the patients for training and validation of the classifier and the remaining 1/3 as an independent hold-out test data for performance estimation. RESULTS: A high level of discrimination between bipolar disorder with psychosis and SSD (combined balanced accuracy = 96.2%; class accuracies 100% for bipolar and 92.3% for SSD) was achieved when effective connectivity and morphometry of the triple network nodes was combined with symptom scores. Patients with SSD were discriminated from patients with bipolar disorder and psychosis as showing higher clinical severity of disorganization and higher variability in the effective connectivity between salience and executive networks. CONCLUSIONS: Our results support the view that the study of network-level connectivity patterns can not only clarify the pathophysiology of SSD but also provide a measure of excellent clinical utility to identify discrete diagnostic/prognostic groups among individuals with psychosis.

Resting state fMRI data from subjects scanned with the EPI-PACE (Echoplanar Imaging - Prospective Acquisition CorrEction) sequence.

Due to the confounding effects of head motion artifacts on resting-state functional connectivity (RSFC), there has been a growing interest in both acquisition and preprocessing strategies for removing motion-related artifacts from resting state functional Magnetic Resonance Imaging (Rs-fMRI) data. Prospective motion correction by the Siemens' EPI-PACE sequence could offer new insights on the effectiveness of this sequence to correct head motion artifacts in Rs-fMRI data. The head motion parameters along with Rs-fMRI data obtained from 47 healthy individuals scanned with the EPI-PACE sequence is presented in this article. This data is useful to understand the effectiveness of prospective motion correction strategies such as 3D PACE for reducing head motion artifacts in Rs-fMRI data and help devise effective motion correction strategies. The utility of the EPI-PACE sequence in reducing motion correction artifacts in healthy controls can be found in our research article on the topic [1].

Dynamic brain connectivity is a better predictor of PTSD than static connectivity.

Using resting-state functional magnetic resonance imaging, we test the hypothesis that subjects with post-traumatic stress disorder (PTSD) are characterized by reduced temporal variability of brain connectivity compared to matched healthy controls. Specifically, we test whether PTSD is characterized by elevated static connectivity, coupled with decreased temporal variability of those connections, with the latter providing greater sensitivity toward the pathology than the former. Static functional connectivity (FC; nondirectional zero-lag correlation) and static effective connectivity (EC; directional time-lagged relationships) were obtained over the entire brain using conventional models. Dynamic FC and dynamic EC were estimated by letting the conventional models to vary as a function of time. Statistical separation and discriminability of these metrics between the groups and their ability to accurately predict the diagnostic label of a novel subject were ascertained using separate support vector machine classifiers. Our findings support our hypothesis that PTSD subjects have stronger static connectivity, but reduced temporal variability of connectivity. Further, machine learning classification accuracy obtained with dynamic FC and dynamic EC was significantly higher than that obtained with static FC and static EC, respectively. Furthermore, results also indicate that the ease with which brain regions engage or disengage with other regions may be more sensitive to underlying pathology than the strength with which they are engaged. Future studies must examine whether this is true only in the case of PTSD or is a general organizing principle in the human brain. Hum Brain Mapp 38:4479-4496, 2017. © 2017 Wiley Periodicals, Inc.