Detecting the Information of Functional Connectivity Networks in Normal Aging Using Deep Learning From a Big Data
Xin Wen1, Li Dong2,3, Junjie Chen1
1College of Information and Computer, Taiyuan University of Technology, Taiyuan, China.
Frontiers in Neuroscience
|February 4, 2020
Summary
This study introduces a new deep learning method, DAFA, to analyze brain function changes in aging using fMRI data. DAFA effectively identifies age-related functional connectivity alterations, outperforming traditional methods.
Area of Science:
- Neuroscience
- Artificial Intelligence
- Big Data Analytics
Background:
- Aging research increasingly utilizes functional magnetic resonance imaging (fMRI) to understand brain function changes.
- Deep learning and big data approaches are emerging as powerful tools for analyzing complex fMRI datasets in aging studies.
- Identifying age-related alterations in functional brain networks is crucial for understanding cognitive decline and developing interventions.
Purpose of the Study:
- To propose and validate a novel deep learning method, Deep neural network (DNN) with Autoencoder (AE) pretrained Functional connectivity Analysis (DAFA), for analyzing brain aging using fMRI.
- To identify significant functional connectivity (FC) changes associated with aging by leveraging big data from the CamCAN fMRI dataset.
- To compare the efficacy of DAFA against traditional methods in detecting age-related brain function alterations.
Main Methods:
- Utilized resting-state fMRI data from 421 subjects from the CamCAN dataset.
- Calculated functional connectivities using a sliding window method and employed an Autoencoder (AE) for complex pattern mining.
- Applied an AE-pretrained Deep Neural Network (DNN) for subject classification (young vs. old) and search-back analysis to identify age-related FC changes.
Main Results:
- DAFA identified significant age-related functional connectivity changes, including within and between various brain networks (DMN, sensorimotor, frontoparietal, etc.).
- These identified FC alterations were correlated with behavioral data, specifically fluid intelligence and response time.
- The DAFA method demonstrated superior performance in discovering age-related brain functional connectivity changes compared to traditional FC analysis techniques.
Conclusions:
- The proposed DAFA method is effective and superior to traditional approaches for detecting age-related functional connectivity changes in fMRI data.
- DAFA offers a promising avenue for exploring critical information in brain aging and other fMRI research areas.
- This deep learning-based approach enhances the understanding of brain function alterations during the aging process.
