Magnetic Resonance Imaging of Human Brain Anatomy and Function

National Institute Of Neurological Disorders And Stroke

Investigators

Linked publications, trials & patents

Abstract

The overarching goal of this project is to develop an improved understanding of brain anatomy and function through application of state-of-the-art MRI techniques and conjoint analysis of functional MRI (fMRI) and electrophysiological signals (EEG, ECoG). EEG and fMRI activity patterns contain a wealth of information that allows detailed study of the major communication pathways in the brain as well as the major clusters of brain regions that work together to subserve specific brain functions. Over recent years, studies in the Advanced MRI section (AMRI) and other labs presented evidence that EEG and fMRI activity patterns are highly dependent on brain arousal state and may undergo major changes over time scales as short as several seconds. In addition, arousal state changes are accompanied by changes in autonomic physiology that, by themselves, may lead to fMRI signal changes. For example, widespread fMRI signals arise both from autonomic and from neuromodulatory activity, especially in absence of attention-demanding tasks and during states of low arousal such as drowsiness and sleep. A longstanding aim of AMRI has been cataloguing the potential contributors to the fMRI signal, with the goal of improving interpretation and to better understand the nature and function arousal variations. To study this in detail, AMRI has been performing overnight sleep studies in which EEG, fMRI and autonomic signals are acquired concurrently. Data from a pilot study acquired in prior years has been made available on openneuro (“ds005127”). Analysis of the pilot study data is continuing and so far has predominantly been performed in a collaborative manner with the groups of Giulio Bernardi (University of Lucca), Zhiwei Ma (Shanghai Tech), and Xiao Liu (Penn State University). Recent research has suggested that spontaneous brain activity has wave-like characteristics, with both fMRI and ECOG showing sequential activation of regions along the cortical functional hierarchy, predominantly around arousal transitions. To investigate how these waves may depend on arousal state and more specifically sleep stage, with Xiao Liu of Penn State University we analyzed data from our pilot sleep study. We found that waves were prevalent during all sleep stages and that rapid eye movement (REM) sleep was characterized by a relatively high level of bottom-up versus top-down propagations (low-to-high versus high-to-low in the functional hierarchy). Earliest activation occurred in thalamus, pons, and visual cortex, key regions in the generation of so-called ponto-geniculo-occipital (PGO) waves associated with REMs. Thus, fMRI may complement EEG in detecting PGO waves and studying their (sub-) cortical involvement, which otherwise requires intracranial recording. Although still only partially understood, PGO waves may have relevance for memory consolidation. This study was published in NPJ Biological Timing and Sleep (Liu et al., 2025). Previous work on mice has shown that adrenergic mediated arousal increases as observed from pupil dilations lead to cascades of intense neuronal spiking activity. Together with the group of Xiao Liu, we investigated the role of these cascades for memory processing in mice. During repeated visual tasks, we observed a temporal order in the firing of a subset of neurons in the hippocampus. During rest, this temporally ordered firing in hippocampus continued, and often in reversed order (backward replay), centered around whole brain cascade centers. This backward replay co-occurred with SPW-ripple complexes. Direct relevance to memory encoding was subsequently confirmed in mice, where high efficiency encoding in a visual task occurred during peak arousal (dilated pupil), alternating with a phase of lower encoding, low arousal, and peak hippocampal ripple activity. This study was published in Advanced Science (Yang et al., 2025). Our main sleep study of 43 subjects was completed in 2024, and we have started data analysis. During 2025, we have completed the pre-processing of the fMRI data to remove contributions from cardiac and respiratory cycles, and from head motion. We have started the pre-processing of the EEG signals, including gradient and cardio-ballistic artifact correction, and sleep-staging. This pre-analysis is expected to continue during the remainder of 2025. AMRI also has continued to make progress in improving anatomical MRI contrast and understanding its contributors. Over the years, AMRI has investigated ways to distinguish between the effects of changes in iron and myelin on MRI contrast, with the goal of characterizing pathology in neurodegenerative disorders such as Multiple Sclerosis, Amyotropic Lateral Sclerosis, and Parkinson’s Disease. At high magnetic field, magnetic susceptibility contrast is exquisitely sensitive to these changes, however this contrast type is difficult to interpret. Because of the limited ability of SW MRI to distinguish between contributions of iron and myelin, we investigated an alternative approach that involves the collection of quantitative T2 contrast. The rationale was to combine quantitative T1 and T2 contrast at high field, where T2 may become increasingly sensitive to iron and T1 becomes increasingly sensitive to myelin due to the longer myelin proton T1. Potential confounds are previously reported sensitivity to the orientation of myelinated fibers for T1 and T2. To overcome difficulties in generation of T2 contrast at high field (7T) associated with RF transmission field inhomogeneities and RF-induced tissue heating, we implemented the GESSE (Gradient echo sampling of the spin echo) technique. We then investigated to what extent field strength affects the sensitivity of T2 to iron by performing GESSE brain scans at 3T and 7T. We found that the contribution of putative brain iron content to R2 (=1/T2) increases linear with field with a proportionality constant of 0.020/s per Tesla per ppm iron. In contrast with previous work, iron was not the only field-dependent contributor to R2: putative myelin/lipid content also contributed to field dependence in addition to a tissue-type independent factor. Combined with the previously established dependencies of R1 (=1/T1) on iron and myelin, these finding for R2 suggest a way to quantify regional variations in iron and myelin by joint analysis of quantitative R1 and R2 measurements. As the fractional contribution of iron and myelin to both R1 and R2 increases with field strength, high field measurements offer improved quantification. These results were published in Magnetic Resonance in Medicine (van Gelderen et al., 2025) Various studies on the origins of T2* and T1 contrast at 7T and a T2 study at 3T have reported a contribution to relaxation time of white matter fiber orientation relative to the MRI static magnetic field. For T2, this contribution may relate to diffusion through extra-axonal susceptibility effects and at 3T may be as high as ±10%. To investigate orientation dependence of R2 at 7T, we performed whole brain GESSE and analyzed extracted R2 maps in detail in both grey and white matter. We found that within the major white matter fiber bundles, there was not only an apparent dependence of ±13% on fiber orientation, but also an apparent dependence on putative axonal diameter index. These contributions to R2 need to be considered when combining with R1 data to infer iron and myelin content. These results were published in Magnetic Resonance in Medicine (Wang et al., 2025)