Regression Analysis
Research Papers
Diffusion tensor imaging of the corpus callosum in healthy aging: Investigating higher order polynomial regression modelling
Previous diffusion tensor imaging (DTI) studies confirmed the vulnerability of corpus callosum (CC) fibers to aging. However, most studies employed lower order regressions to study the relationship between age and white matter microstructure. The present study investigated whether higher order polynomial regression modelling can better describe the relationship between age and CC DTI metrics compared to lower order models in 140 healthy participants (ages 18-85). The CC was found to be non-uniformly affected by aging, with accelerated and earlier degradation occurring in anterior portion; callosal volume, fiber count, fiber length, mean fibers per voxel, and FA decreased with age while mean, axial, and radial diffusivities increased. Half of the parameters studied also displayed significant age-sex interaction or intracranial volume effects. Higher order models were chosen as the best fit, based on Bayesian Information Criterion minimization, in 16 out of 23 significant cases when describing the relationship between DTI measurements and age. Higher order model fits provided different estimations of aging trajectory peaks and decline onsets than lower order models; however, a likelihood ratio test found that higher order regressions generally did not fit the data significantly better than lower order polynomial or linear models. The results contrast the modelling approaches and highlight the importance of using higher order polynomial regression modelling when investigating associations between age and CC white matter microstructure.
View Full Paper →Predictors of Neurofeedback Outcomes Following qEEG Individualized Protocols for Anxiety
In this retrospective study, researchers examined effects of quantitative electroencephalography (qEEG), individualized neurofeedback treatment protocols for anxiety. The present study includes 52 clients with 53.8% (n = 28) self-reporting as male and included two time points (pre and post). Secondary analyses utilized a subset of client data (n = 21) with measurements from three time points (pre, post, and follow-up). All clients completed qEEG and self-report assessments. Clients agreed to attend a minimum of 15 biweekly sessions, for one academic semester. Findings from regression analyses revealed three predictors of posttreatment outcomes. In addition, analysis of a subsample of data assessed at three time points revealed statistically significant improvement from pre to post and sustained outcomes from post to follow-up. We discuss limitations and implications for future research.
View Full Paper →Towards Using Microstate-Neurofeedback for the Treatment of Psychotic Symptoms in Schizophrenia. A Feasibility Study in Healthy Participants
Spontaneous EEG signal can be parsed into sub-second periods of stable functional states (microstates) that assumingly correspond to brief large scale synchronization events. In schizophrenia, a specific class of microstate (class "D") has been found to be shorter than in healthy controls and to be correlated with positive symptoms. To explore potential new treatment options in schizophrenia, we tested in healthy controls if neurofeedback training to self-regulate microstate D presence is feasible and what learning patterns are observed. Twenty subjects underwent EEG-neurofeedback training to up-regulate microstate D presence. The protocol included 20 training sessions, consisting of baseline trials (resting state), regulation trials with auditory feedback contingent on microstate D presence, and a transfer trial. Response to neurofeedback was assessed with mixed effects modelling. All participants increased the percentage of time spent producing microstate D in at least one of the three conditions (p < 0.05). Significant between-subjects across-sessions results showed an increase of 0.42 % of time spent producing microstate D in baseline (reflecting a sustained change in the resting state), 1.93 % of increase during regulation and 1.83 % during transfer. Within-session analysis (performed in baseline and regulation trials only) showed a significant 1.65 % increase in baseline and 0.53 % increase in regulation. These values are in a range that is expected to have an impact upon psychotic experiences. Additionally, we found a negative correlation between alpha power and microstate D contribution during neurofeedback training. Given that microstate D has been related to attentional processes, this result provides further evidence that the training was to some degree specific for the attentional network. We conclude that microstate-neurofeedback training proved feasible in healthy subjects. The implementation of the same protocol in schizophrenia patients may promote skills useful to reduce positive symptoms by means of EEG-neurofeedback.
View Full Paper →Spatially aggregated multiclass pattern classification in functional MRI using optimally selected functional brain areas
In previous works, boosting aggregation of classifier outputs from discrete brain areas has been demonstrated to reduce dimensionality and improve the robustness and accuracy of functional magnetic resonance imaging (fMRI) classification. However, dimensionality reduction and classification of mixed activation patterns of multiple classes remain challenging. In the present study, the goals were (a) to reduce dimensionality by combining feature reduction at the voxel level and backward elimination of optimally aggregated classifiers at the region level, (b) to compare region selection for spatially aggregated classification using boosting and partial least squares regression methods and (c) to resolve mixed activation patterns using probabilistic prediction of individual tasks. Brain activation maps from interleaved visual, motor, auditory and cognitive tasks were segmented into 144 functional regions. Feature selection reduced the number of feature voxels by more than 50%, leaving 95 regions. The two aggregation approaches further reduced the number of regions to 30, resulting in more than 75% reduction of classification time and misclassification rates of less than 3%. Boosting and partial least squares (PLS) were compared to select the most discriminative and the most task correlated regions, respectively. Successful task prediction in mixed activation patterns was feasible within the first block of task activation in real-time fMRI experiments. This methodology is suitable for sparsifying activation patterns in real-time fMRI and for neurofeedback from distributed networks of brain activation.
View Full Paper →Common carotid flow velocity is associated with cognition in older adults
AIMS: To assess the relationship between carotid flow velocity and cognitive impairment in patients with mild-moderate (<50%) carotid artery disease. METHODS: We studied 407 participants with available carotid ultrasound and cognitive measures. We related peak systolic velocity (PSV) and end diastolic velocity (EDV) of internal carotid artery (ICA) and common carotid artery (CCA) and intimal medial thickness (IMT) to Mini Mental State Examination (MMSE), Clock Draw Test (CDT), Activities of Daily Living Scale (ADL)and Montreal Cognitive Assessment (MoCA). RESULTS: EDV of CCA was significantly different in higher and lower MoCA (MMSE) groups. Multiple regression analysis demonstrated that lower EDV was significantly associated with lower MoCA (+0.459 per standard deviation (SD), p<0. 01 for the left; +0.539 per SD, p<0. 01 for the right) and CDT (odds ratio (OR) 0.093, p< 0.05 for the left; OR) 0.120, p<0. 01 for the right) scores. PSV of left CCA (-0.205 per SD, p<0.05) and IMT (+42.536 per SD, p< 0.001) were associated with ADL. PSV of right CCA was associated with MMSE (+0.081 per SD, p<0.001). No significant relationship between ICA flow velocity and cognitive performance was observed. CONCLUSIONS: Our preliminary data show that common carotid artery flow velocity was associated with cognitive performance.
View Full Paper →Ready to Optimize Your Brain?
Schedule a free consultation to discuss regression analysis and how neurofeedback training can help
Or call us directly at 855-88-BRAIN
View Programs & Pricing →