Abstract:Introduction: Uncovering the lifespan changes of the human brain is definitely thrilling and a fundamental goal in discovery neuroscience. The lifespan trajectory coding both normal developing and aging stages of the brain has reflected dramatic changes both its structure [1-3] and function [4-6]. It has been commonly accepted that human brain is structurally and functionally organized into a complex network allowing the segregation and integration of information processing [7,8]. In particular, the brain network, describing the interaction between different brain regions, has demonstrated consistent non-trivial properties and reliable changes throughout normal development, aging and in various pathological conditions. Beyond these advances, how the brain network is reorganized through the whole lifespan or its lifespan trajectory has been rarely studied. Here, armed with the graph theory, we aim to chart the lifespan trajectory of human whole-brain functional networks based on resting-state functional MRI (R-fMRI) datasets from 150 subjects collected from the same scanner. Materials and Methods: Participants and Data acquisition: A public R-fMRI dataset including total 150 right-handed healthy people (males, 87; females, 63; age range, 7-85 years) from the NKI/Rockland sample (http://fcon_1000.projects.nitrc.org/indi/pro/nki.html) was employed. All subjects have no history of neurological and psychiatric disorders. The resting state fMRI scans were performed on the Siemens Trio 3.0 Tesla MRI scanner using an echo-planar imaging (EPI) sequence (TR/TE=2500/30 ms, FA=80°, FOV=216 mm, matrix=256×256, slices=38, thickness=3.0 mm, 260 volumes). High-resolutionT1-weighted images were also collected by magnetization-prepared rapid gradient echo (MPRAGE) sequence (TR/TE=2500/3.5 ms, FA=8°, thickness=1.0 mm, slices=192, matrix=256×256, FOV=256mm). Functional data preprocessing: Data preprocessing was similar to as implemented in [6], producing individual residual timeseries in the MNI152 standard space. Network Construction: A group mask in standard space was made which includes all voxels with non-zeros standard deviation of bold time series and gray matter tissue probability greater than 20%. A high-resolution parcellation template was generated by randomly partitioning the group mask into 1024 contiguous and uniform regions using an algorithm described in [9]. Resting state functional connectivity (RSFC) for each pair of 1024 regions of interest was obtained by computing Pearson correlation coefficients of their representative (i.e., mean across all voxels) time series. Each subject’s weighted functional connectivity network was then constructed based on a Bonforroni correction procedure. Network properties analysis: We studied the age-related variation trends of the following regional characteristic and connection property: 1) the regional correlation strength which measures the average correlation extent by which a given region, i, is connected to the rest of the network. Regions with correlation strength greater than the mean and two standard deviations of the global network correlation strength were defined as hubs in every person. 2) The anatomical distance of the connection between region i and region j is defined as the Euclidean distance between centroids of the two regions. The percentages of long-distance (>75mm) and short-distance (≤75mm) connection of the functional connectome were computed for each subject. Results: The probability map of hubs in population was gotten and several specific regions including default-mode network regions, attention network regions, motor-related regions and visual cortex were detected to be functional hubs consistently (Figure 1.A). Besides, we observed significantly linear decreases of RSFC strength with age primarily located in several default-mode (e.g., medial frontal and parietal cortex, lateral temporal and parietal cortex) and attention network regions (e.g., insula, dorsal anterior cingulate cortex, lateral frontal cortex and temporal-parietal conjunction) while few regions including SMA and temporal pole showed significantly linear increases (p<0.05, FDR corrected) (Figure 1.B). Meanwhile, significant quadratic trajectories of RSFC strength on age were also detected within the left hippocampus and bilateral superior temporal gyrus (p<0.01, uncorrected) (Figure 1.C). The network sparsity (i.e., number of graph edges or network connections showing significant RSFC) between brain regions demonstrated a linear trend of decrease with age (not significant) through the lifespan (p=0.10). The percentage of long-distance connections was found increase up to age 42 and then decrease while the short-distance connections showed an inverse trajectory (p=0.01) (Figure 2).

Norm Atlas of Lifespan for Age and Cognitive Ability Prediction Based on Dual-Paradigm Fmri

Age Prediction Using Resting-State Functional MRI

Connectome-based Predictive Modelling of Ageing, Overall Cognitive Functioning and Memory Performance.

Estimating brain age based on a healthy population with deep learning and structural MRI

Predicting human age using regional morphometry and inter-regional morphological similarity

Decoding Lifespan Changes of the Human Brain Using Resting-State Functional Connectivity MRI.

Predicting brain age using partition modeling strategy and atlas-based attentional enhancement in the Chinese population

Predicting Brain Age of Healthy Adults Based on Structural MRI Parcellation Using Convolutional Neural Networks

Brain Age Prediction Based on Resting-State Functional MRI Using Similarity Metric Convolutional Neural Network.

Brain-age prediction: Systematic evaluation of site effects, and sample age range and size

Mapping Functional Connectome Changes of the Human Brain across the Life Span

Brain age prediction across the human lifespan using multimodal MRI data

Brain age prediction via cross-stratified ensemble learning

Fusing multi-scale functional connectivity patterns via Multi-Branch Vision Transformer (MB-ViT) for macaque brain age prediction

Predicting brain age across the adult lifespan with spontaneous oscillations and functional coupling in resting brain networks captured with magnetoencephalography

Brain Age Prediction in Developing Childhood with Multimodal Magnetic Resonance Images.

Detecting the Information of Functional Connectivity Networks in Normal Aging Using Deep Learning from a Big Data Perspective.

Multiscale Functional Connectivity Patterns of the Aging Brain Learned from Harmonized Rsfmri Data of the Multi-Cohort Istaging Study

Parallel Pathway Convolutional Neural Network with Low-rank Fusion for Brain Age Prediction.

[18F]FDG PET integrated with structural MRI for accurate brain age prediction

Estimating Brain Age with Global and Local Dependencies