Predicting behavior through dynamic modes in resting-state fMRI data

Machine learning on interictal intracranial EEG predicts surgical outcome in drug resistant epilepsy

Surgical success for patients with focal drug resistant epilepsy (DRE) relies on accurate localization of the epileptogenic zone (EZ). Currently, no exam delineates this zone unambiguously. Instead, the EZ is approximated by the area where seizures begin, which is identified manually through a tedious process that is prone to errors and biases. More importantly, resection of this area does not always predict good surgical outcome. Here, we propose an artificially intelligent, patient-specific framework that automatically identifies the EZ requiring little to no input from clinicians, without having to wait for a seizure to occur. The framework transforms interictal intracranial electroencephalography data into spatiotemporal representations of brain activity discriminating the interictal epileptogenic network from background activity. The epileptogenic network delineates the EZ with high precision and predicts surgical outcome. Our framework eliminates the need for manual data inspection, reduces prolonged monitoring, and enhances surgical planning for DRE patients.

Generative dynamical models for classification of rsfMRI data

The growing availability of large-scale neuroimaging datasets and user-friendly machine learning tools has led to a recent surge in studies that use fMRI data to predict psychological or behavioral variables. Many such studies classify fMRI data on the basis of static features, but fewer try to leverage brain dynamics for classification. Here, we pilot a generative, dynamical approach for classifying resting-state fMRI (rsfMRI) data. By fitting separate hidden Markov models to the classes in our training data and assigning class labels to test data based on their likelihood under those models, we are able to take advantage of dynamical patterns in the data without confronting the statistical limitations of some other dynamical approaches. Moreover, we demonstrate that hidden Markov models are able to successfully perform within-subject classification on the MyConnectome dataset solely on the basis of transition probabilities among their hidden states. On the other hand, individual Human Connectome Project subjects cannot be identified on the basis of hidden state transition probabilities alone-although a vector autoregressive model does achieve high performance. These results demonstrate a dynamical classification approach for rsfMRI data that shows promising performance, particularly for within-subject classification, and has the potential to afford greater interpretability than other approaches.

Childhood adversities characterize the heterogeneity in the brain pattern of individuals during neurodevelopment

BACKGROUND

Several factors shape the neurodevelopmental trajectory. A key area of focus in neurodevelopmental research is to estimate the factors that have maximal influence on the brain and can tip the balance from typical to atypical development.

METHODS

Utilizing a dissimilarity maximization algorithm on the dynamic mode decomposition (DMD) of the resting state functional MRI data, we classified subjects from the cVEDA neurodevelopmental cohort (n = 987, aged 6-23 years) into homogeneously patterned DMD (representing typical development in 809 subjects) and heterogeneously patterned DMD (indicative of atypical development in 178 subjects).

RESULTS

Significant DMD differences were primarily identified in the default mode network (DMN) regions across these groups (p < 0.05, Bonferroni corrected). While the groups were comparable in cognitive performance, the atypical group had more frequent exposure to adversities and faced higher abuses (p < 0.05, Bonferroni corrected). Upon evaluating brain-behavior correlations, we found that correlation patterns between adversity and DMN dynamic modes exhibited age-dependent variations for atypical subjects, hinting at differential utilization of the DMN due to chronic adversities.

CONCLUSION

Adversities (particularly abuse) maximally influence the DMN during neurodevelopment and lead to the failure in the development of a coherent DMN system. While DMN's integrity is preserved in typical development, the age-dependent variability in atypically developing individuals is contrasting. The flexibility of DMN might be a compensatory mechanism to protect an individual in an abusive environment. However, such adaptability might deprive the neural system of the faculties of normal functioning and may incur long-term effects on the psyche.

Neural Underpinnings of Learning in Dementia Populations: A Review of Motor Learning Studies Combined with Neuroimaging

The intent of this review article is to serve as an overview of current research regarding the neural characteristics of motor learning in Alzheimer disease (AD) as well as prodromal phases of AD: at-risk populations, and mild cognitive impairment. This review seeks to provide a cognitive framework to compare various motor tasks. We will highlight the neural characteristics related to cognitive domains that, through imaging, display functional or structural changes because of AD progression. In turn, this motivates the use of motor learning paradigms as possible screening techniques for AD and will build upon our current understanding of learning abilities in AD populations.

Manifold alteration between major depressive disorder and healthy control subjects using dynamic mode decomposition in resting-state fMRI data

Background

The World Health Organization has reported that approximately 300 million individuals suffer from the mood disorder known as MDD. Non-invasive measurement techniques have been utilized to reveal the mechanism of MDD, with rsfMRI being the predominant method. The previous functional connectivity and energy landscape studies have shown the difference in the coactivation patterns between MDD and HCs. However, these studies did not consider oscillatory temporal dynamics.

Methods

In this study, the dynamic mode decomposition, a method to compute a set of coherent spatial patterns associated with the oscillation frequency and temporal decay rate, was employed to investigate the alteration of the occurrence of dynamic modes between MDD and HCs. Specifically, The BOLD signals of each subject were transformed into dynamic modes representing coherent spatial patterns and discrete-time eigenvalues to capture temporal variations using dynamic mode decomposition. All the dynamic modes were disentangled into a two-dimensional manifold using t-SNE. Density estimation and density ratio estimation were applied to the two-dimensional manifolds after the two-dimensional manifold was split based on HCs and MDD.

Results

The dynamic modes that uniquely emerged in the MDD were not observed. Instead, we have found some dynamic modes that have shown increased or reduced occurrence in MDD compared with HCs. The reduced dynamic modes were associated with the visual and saliency networks while the increased dynamic modes were associated with the default mode and sensory-motor networks.

Conclusion

To the best of our knowledge, this study showed initial evidence of the alteration of occurrence of the dynamic modes between MDD and HCs. To deepen understanding of how the alteration of the dynamic modes emerges from the structure, it is vital to investigate the relationship between the dynamic modes, cortical thickness, and surface areas.

Comparison Between Gradients and Parcellations for Functional Connectivity Prediction of Behavior

Resting-state functional connectivity (RSFC) is widely used to predict behavioral measures. To predict behavioral measures, representing RSFC with parcellations and gradients are the two most popular approaches. Here, we compare parcellation and gradient approaches for RSFC-based prediction of a broad range of behavioral measures in the Human Connectome Project (HCP) and Adolescent Brain Cognitive Development (ABCD) datasets. Among the parcellation approaches, we consider group-average "hard" parcellations (Schaefer et al., 2018), individual-specific "hard" parcellations (Kong et al., 2021a), and an individual-specific "soft" parcellation (spatial independent component analysis with dual regression; Beckmann et al., 2009). For gradient approaches, we consider the well-known principal gradients (Margulies et al., 2016) and the local gradient approach that detects local RSFC changes (Laumann et al., 2015). Across two regression algorithms, individual-specific hard-parcellation performs the best in the HCP dataset, while the principal gradients, spatial independent component analysis and group-average "hard" parcellations exhibit similar performance. On the other hand, principal gradients and all parcellation approaches perform similarly in the ABCD dataset. Across both datasets, local gradients perform the worst. Finally, we find that the principal gradient approach requires at least 40 to 60 gradients to perform as well as parcellation approaches. While most principal gradient studies utilize a single gradient, our results suggest that incorporating higher order gradients can provide significant behaviorally relevant information. Future work will consider the inclusion of additional parcellation and gradient approaches for comparison.