其他摘要 | The brain is the most important and complex structure in the nervous system. Decoding how the brain works is the ultimate goal of neuroscientists. Neurons in the brain transmit and process information by generating action potentials. These microscopic electrical signals can be measured directly or indirectly by different techniques, generating neural oscillations at different levels. Single-unit recording, for example, can directly measure the electrical fluctuations of neurons at the microscopic level. Electroencephalography (EEG) measures brain oscillations generated by the electrical activity of millions of neurons at a macro level through extracranial electrodes. Functional magnetic resonance imaging (fMRI) can indirectly measure neural oscillations by measuring changes of the blood-oxygenation-level-dependent (BOLD) oscillations in cerebral blood flow. Over the past century, researches on human brain function using techniques such as EEG have found that different brain functions and physiological or psychological states prefer different oscillation frequencies. However, due to the limitations of the signal-to-noise ratio and sampling rate in the early studies, most fMRI studies focused on a composite band. And few studies explored the functional differences of BOLD oscillations at different frequencies. Although single-band fMRI research has made great progress in the localization of brain function and recognition of brain spontaneous networks, the frequency information contained in the brain oscillations has been ignored in most traditional fMRI studies. In the last decade, more and more scientists began to pay attention to the activation patterns of BOLD oscillations in different frequency bands. This is a step closer to unraveiling the mystery of brain function by using fMR.
In order to fully understand the progress of multi-band fMRI research, we conducted a bibliometric analysis of an early study on multi-band fMRI analysis using theory-based frequency decomosition method, and systematically reviewed the results of multiband fMRI research in the past decade. There are three limitations in the existing multi-band fMRI research: First, different research adopted different frequency decomposition methods, which includes experience-based method, theory-based method and data-driven method. It is difficult to directly compare the results of studies using different methods. Second, the existing multi-band studies are mainly resting-state studies and clinical studies, so there is a lack of direct evidence of the functions of different frequency bands. Third, most resting-state studies focused on the comparison between slow-4 and slow-5 bands, thus lacking systematic research on the organization of spontaneous activity across the whole frequency bands. Based on the existing results, we summarized the spontaneous activity patterns of BOLD oscillations in different frequency bands, hypothesized the functions of different frequency bands, and proposed a theoretical model to describe the function and interaction of BOLD oscillations in different frequency bands – the three-level interaction model. The main scientific question of this study is to investigate the function of BOLD oscillations in different frequency bands, and preliminarily verify the theoretical model. Based on the limitations of existing studies, the following three aspects were carried out: (1) We determined the frequency decomposition method applicable to BOLD oscillations, and clarified its calculation method in practical application; (2) Multi-band analysis of task-state fMRI research was carried out to directly locate different functional activated bands; (3) The functional connectome gradient analysis method was used to study the spontaneous activity organization of the large-scale network of BOLD oscillations in different frequency bands and the development pattern at school age. The main research questions and results include:
Study 1: Calculation and application of the natural logarithm linear law (N3L). Firstly, we determined the calculation of utilizing the N3L on frequency decompositino for discrete data. And we developed a toolbox for frequency decomposition based on the N3L. Then, taking multi-band analysis of head motion signals as an example, we explored the general application of the N3L-based method in physiological oscillations. Based on the head motion data of 84 healthy children aged 3 to 16 years, we revealed the developmental effects and gender differences in head motion oscillations at different frequency bands.
Study 2: The function of multi-band BOLD oscillations. In this study, task-state fMRI data with high sampling rate from the Human Connectome Project (HCP) S1200 database were used to study the activation patterns of BOLD oscillations in different frequency bands under VI different tasks. The results are cosistent with the assumptions in the three-level interaction model. That is, oscillations of slow-1 to slow-3 bands are mainly involved in the detection and preliminary processing of visual and auditory stimuli. Slow-4 band is involved in motor function and various cognitive functions. Slow-5 band plays an important role in complex cognitive functions and memory.
Study 3: Functional connectome gradient analysis of multi-band BOLD oscillations. Based on the resting-state fMRI data with high sampling rate from HCP S1200 dataset, we performed gradient analysis to study the spontaneous organization of the large-scale network in different frequency bands. The results show that, the first and second gradients have similar distribution patterns across frequency bands as in the traditional resting-state interval. The first gradient shows gradient changes from transmodal brain region to unimodal brain region, and the second gradient shows gradient changes between different modules of the primary sensory cortex. However, the gradient distribution of slow-1 and slow-2 bands differs in details with lower bands, which may reflect the difference between functional segregation and functional integration. Based on the development data collected by Chinese Color Nest Project (CCNP) in Chongqing, we investigated the development pattern of gradients in slow-3 to slow-5 bands. The results show that: (1) The developmental process shows three distinct patterns, which reflects that there are stages in the development of brain function and possible critical stages of development; (2) The development patterns of different frequency bands differentiate greatly, which is also a reflection of the functional differences of different frequency bands.
This study explored the function of multi-frequency BOLD oscillations from theory to demonstration. The functional organization pattern of BOLD oscillations is proposed theoretically, and the frequency decomposition method suitable for BOLD oscillations is determined. Multi-band brain activation analysis of task-state fMRI data with high sampling rates provided direct evidence of different band functions. We studied the organization patterns of spontaneous brain networks in different frequency bands and revealed the development patterns of functional connectome gradients in different frequency bands. |
修改评论