Neural Dynamics during Resting State
Abstract
As per Dr Jean Chen, the brain is a complex system with high order. Both bodily movements and mental activities are dependent on the information transmitted by billions to millions of neurons. The high dimension of neural signals makes potential patterns difficult to discover. Researches in the past have found rotary paths in rhythmic and nonrhythm movement when transforming neural electrical signals into a 2-dimensional space. The validity of this analogy at rest is unknown. Because of the low-frequency fluctuations seen during spontaneous neural activity using functional magnetic resonance imaging, (fMRI), one can easily hypothesize that the neural reaction at resting condition also exhibits a periodic trajectory. The potential patterns in resting fMRI data were explored on two cohorts. One consisted of older adults and the second of patients suffering from Alzheimer’s disease. The other was made up of normal controls and patients with Alzheimer’s. To visualize the trajectory, the jPCA method was applied. It reduced the BOLD signal which was high-dimensional to a two-dimensional space. The results revealed that the “resting condition” is a basic state that shows an inherent dynamic pattern with low frequency and a long period during normal ageing. There are changes in the rotary time at the slow4 frequency spectrum (0.027 to 0.073 Hz) during Alzheimer’s disease (AD). These findings add to the existing understanding of how neural signals can self-rotate and that motor executive signals are made up of neurosignals. The rotary period at frequency band slow4 might be a physiological marker of AD. Further studies on this frequency band could help us understand the potential pathophysiology, as well as aid in diagnosing AD.
1. Introduction
The brain is a complex, high-order system. Information is transmitted between billions and billions to enable body movements as well as mental activities to be possible [ 1]. There has been evidence that neural activities can be used to determine the response to a specific movement or task [ 1- 5]. To decode the neural information and identify possible patterns, scientists have tried to determine whether body movement or consciousness can be inferred from it [ 6 – ]
Because it is noninvasive and has a mature method for data acquisition (9), functional magnetic resonance imaging, or fMRI, has been an important technique to study brain function since 1995. One approach to imaging is to record the blood oxygenation level dependent (BOLD), fMRI signal during imaging to measure metabolic fluctuations in different brain regions [ 0_]. These early studies were based on task-based assessments, which sought to establish the relationship between neural signs and task paradigms [ 11]. A growing number of studies has begun to explore BOLD fluctuations at resting status to capture intrinsic activities not easily measured by task-based studies [ 11 ] Previous studies showed that there were low-frequency (0.1Hz) fluctuations in spontaneous neuro activities [ 14], 15], which may have been related to human physiological motions, such as respiration and heart rate [ 16– 18]. The default mode (DMN), one the resting states networks, was activated as a result periodic introspection [ 19]. These patterns, or intrinsic activities, are naturally hidden in BOLD signal.
The problem is that neural data can be difficult to decode. First, the decoding process is complicated by the presence of noise and intrinsic fluctuations. A second problem is the difficulty of assessing neural signals that are high in dimensionality. This prevents statistical analysis as well as observation. Tidal fluctuations and complex data make it difficult to see potential patterns in the neuron population. Furthermore, the analysis model can be complicated due to the high dimensionality in neural signals. This may lead to poor generalization performance.
Recent research has shown that it is possible for high-dimension neural data to be reduced into a lower-dimension space (two or three dimensional) for visualization. Kristan & Calabrese looked into the swimming of leeches. Their findings revealed that single neurons have firing rate oscillations around 1.5 Hz [ 20]. Churchland, et al. The neural population responses were projected to a 2-dimensional space using the jPCA algorithm. The responses showed rotation trajectories of this rhythm movement [ 21]. They also recorded electrical impulses in monkeys during reaching and discovered rotary trajectory even though reaching does not constitute a rhythm motion [ 21]. Meanwhile, Hao et al. Similar results were also obtained with the Laplace Eigenmaps (22). So it is feasible and reasonable to see if there is a consistent trajectory for normal and older subjects at rest. And if these periodic trajectory phenomena are affected by damage to the brain such as in Alzheimer’s disease pathophysiological processes (AD).
We can actually use dimensionality to look at the potential patterns of high dimension neural signals. There have been very few studies to date on visualization of fMRI datasets. Our study sought to understand the intrinsic dynamics and interactions between brain systems. We studied potential patterns in the rs–fMRI data of the elderly and the young. The jPCA algorithm helped to reduce high-dimension BOLD signals into a two-dimensional space, allowing visualization of its dynamic trajectory. To investigate the dynamic patterns of the trajectory and to understand the brain’s working mechanism, we quantitively analysed it. Finally, the same experiments were used to analyze the rsfMRI results of AD and normal controls (NC), in order to investigate any changes in the brain’s patterns in a disordered state.