Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 13, 1435
eCollection

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

Affiliations

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

Xin Wen et al. Front Neurosci.

Abstract

A resting-state functional connectivity (rsFC)-constructed functional network (FN) derived from functional magnetic resonance imaging (fMRI) data can effectively mine alterations in brain function during aging due to the non-invasive and effective advantages of fMRI. With global health research focusing on aging, several open fMRI datasets have been made available that combine deep learning with big data and are a new, promising trend and open issue for brain information detection in fMRI studies of brain aging. In this study, we proposed a new method based on deep learning from the perspective of big data, named Deep neural network (DNN) with Autoencoder (AE) pretrained Functional connectivity Analysis (DAFA), to deeply mine the important functional connectivity changes in fMRI during brain aging. First, using resting-state fMRI data from 421 subjects from the CamCAN dataset, functional connectivities were calculated using sliding window method, and the complex functional patterns were mined by an AE. Then, to increase the statistical power and reliability of the results, we used an AE-pretrained DNN to relabel the functional connectivities of each subject to classify them as belonging to the attributes of young or old individuals. A method called search-back analysis was performed to find alterations in brain function during aging according to the relabeled functional connectivities. Finally, behavioral data regarding fluid intelligence and response time were used to verify the revealed functional changes. Compared to traditional methods, DAFA revealed additional, important aged-related changes in FC patterns [e.g., FC connections within the default mode (DMN) and the sensorimotor and cingulo-opercular networks, as well as connections between the frontoparietal and cingulo-opercular networks, between the DMN and the frontoparietal/cingulo-opercular/sensorimotor/occipital/cerebellum networks, and between the sensorimotor and frontoparietal/cingulo-opercular networks], which were correlated to behavioral data. These findings demonstrated that the proposed DAFA method was superior to traditional FC-determining methods in discovering changes in brain functional connectivity during aging. In addition, it may be a promising method for exploring important information in other fMRI studies.

Keywords: DNN; aging; big data; deep learning; functional connectivity.

Figures

FIGURE 1
FIGURE 1
Workflow of the DAFA method. (A) Constructing FCs using sliding windows and Pearson’s correlation methods, all the FCs from a subject contains most of “young FCs”/“old FCs” and mix FCs. (B) Relabeling all FCs via a DNN based on AE pretraining, and then calculating the average FC of “young FCs” in young subject and “old FCs” in old subject as this subject’s FC. (C) Performing a search-back analysis to compare the percentages of “young FCs” and functional alterations between the young and old groups.
FIGURE 2
FIGURE 2
Learnt parameters. (A) As the batch size increases, loss increases; (B) when the number of epochs = 20, loss reached its first minimum value and then changed periodically with increasing number of epochs.
FIGURE 3
FIGURE 3
The first column shows the average FC patterns of the young group and old group on the left and right, respectively; the second column shows the distribution of “young FCs” (samples in blue indicate individuals in the young group whose percentage of “young FCs” is greater than 50%; samples in orange indicate individuals in the young group whose percentage of “young FCs” is equal to or below 50%; samples in green indicate individuals in the old group whose percentage of “young FCs” is above 50%; and samples in yellow indicate individuals in the old group whose percentage of “young FCs” is equal to or less than 50%); the third column shows the mean and standard deviation of “young FCs” in the two groups. YY indicates samples in the young group for whom more than 50% of the FCs were relabeled to young; YO indicates samples in the young group for whom more than 50% of the FCs were relabeled to old; OO indicates samples in the old group for whom more than 50% of the FCs were relabeled to old; and OY indicates samples in the old group for whom more than 50% of the FCs were relabeled to young.
FIGURE 4
FIGURE 4
Figures in the left column show significant changes in FCs (T-map, p < 0.01, FWE corrected) between the young and old groups via the static, sliding windows and DAFA methods. Blue indicates that the FC from the old group is stronger than that from the young group (i.e., old > young), and red indicates that the FC from the young group is stronger than that from old group (i.e., young > old). The figures in the right column show additional FCs revealed by the sliding windows and DAFA methods compared with the static method. Areas marked by circle indicated the additional altered FCs found by DAFA, compared with the sliding windows method.
FIGURE 5
FIGURE 5
Significant relationships between the additional altered FCs (revealed by the DAFA method) and behavioral performance (p < 0.01). Figures on the top were additional FCs revealed by DAFA. Red lines indicate young > old; blue lines indicate old > young. The first column (from second row to the bottom row) shows the correlations between FCs (young > old) and the Cattell score/M-RT/SD-RT, while the second column (from second row to the bottom row) shows the correlations between FCs (young < old) and the Cattell score/M-RT/SD-RT.

Similar articles

See all similar articles

References

    1. Andrew Z., Michael B. (2015). Towards a statistical test for functional connectivity dynamics. Neuroimage 114 466–470. 10.1016/j.neuroimage.2015.03.047 - DOI - PubMed
    1. Avelar-Pereira B., Backman L., Wahlin A., Nyberg L., Salami A. (2017). Age-related differences in dynamic interactions among default mode, frontoparietal control, and dorsal attention networks during resting-state and interference resolution. Front. Aging Neurosci. 9:152. 10.3389/fnagi.2017.00152 - DOI - PMC - PubMed
    1. Biswal B.B., Mennes M., Zuo X., Gohel S., Kelly C., Smith S.M., et al. (2010). Toward discovery science of human brain function. Proc. Natl. Acad. Sci. U.S.A. 107 4734–4739. 10.1073/pnas.0911855107 - DOI - PMC - PubMed
    1. Biswal B.B., Yetkin F.Z., Haughton V.M., Hyde J.S. (1995). Functional connectivity in the motor cortex of resting human brain using echo-planar mri. Magn. Reson. Med. 34 537–541. 10.1002/mrm.1910340409 - DOI - PubMed
    1. Calhoun V.D., Sui J. (2016). Multimodal fusion of brain imaging data: a key to finding the missing link(s) in complex mental illness. Biol. Psychiatry Cogn. Neurosci. Neuroimaging 1 230–244. 10.1016/j.bpsc.2015.12.005 - DOI - PMC - PubMed
Feedback