Autism Spectrum Disorder Studies Using fMRI Data and Machine Learning: A Review

Meijie Liu,Baojuan Li,Dewen Hu

doi:10.3389/fnins.2021.697870

Abstract

Machine learning methods have been frequently applied in the field of cognitive neuroscience in the last decade. A great deal of attention has been attracted to introduce machine learning methods to study the autism spectrum disorder (ASD) in order to find out its neurophysiological underpinnings. In this paper, we presented a comprehensive review about the previous studies since 2011, which applied machine learning methods to analyze the functional magnetic resonance imaging (fMRI) data of autistic individuals and the typical controls (TCs). The all-round process was covered, including feature construction from raw fMRI data, feature selection methods, machine learning methods, factors for high classification accuracy, and critical conclusions. Applying different machine learning methods and fMRI data acquired from different sites, classification accuracies were obtained ranging from 48.3% up to 97%, and informative brain regions and networks were located. Through thorough analysis, high classification accuracies were found to usually occur in the studies which involved task-based fMRI data, single dataset for some selection principle, effective feature selection methods, or advanced machine learning methods. Advanced deep learning together with the multi-site Autism Brain Imaging Data Exchange (ABIDE) dataset became research trends especially in the recent 4 years. In the future, advanced feature selection and machine learning methods combined with multi-site dataset or easily operated task-based fMRI data may appear to have the potentiality to serve as a promising diagnostic tool for ASD.

Highlights

As a pervasive neurodevelopmental disorder, autism spectrum disorder (ASD) is characterized by deficits in social communication and interaction and restricted and repetitive behaviors (Hull et al, 2017), which was known to be an urgent public health concern that could benefit from enhanced strategies to help identify ASD earlier (Jon et al, 2018)
It is well known that one major goal of the classifications between ASD and TC is to obtain high accuracies, while another is to determine the informative brain regions or networks contributing to the classifications with the ultimate goal of yielding a possible biomarker of ASD
As an anatomically defined atlas, the Automated Anatomical Labeling (AAL) atlas was frequently adopted in ASD classifications (Murdaugh et al, 2012; Wang et al, 2012; Iidaka, 2015; Guo et al, 2017; Bi et al, 2018; Liu J. et al, 2020; Tang et al, 2020; Zhao et al, 2020)

Summary

Introduction

As a pervasive neurodevelopmental disorder, autism spectrum disorder (ASD) is characterized by deficits in social communication and interaction and restricted and repetitive behaviors (Hull et al, 2017), which was known to be an urgent public health concern that could benefit from enhanced strategies to help identify ASD earlier (Jon et al, 2018). Autism Spectrum Disorder Review different advanced neuroimaging tools have been applied for ASD research, including structural and functional magnetic resonance imaging (MRI), positron emission tomography (PET), electroencephalography (EEG), magnetoencephalography (MEG), and novel protocols (Alessandro et al, 2017; Du et al, 2018). Functional MRI (fMRI) studies involving task-based and resting-state fMRI (rs-fMRI) data occupy a large proportion. With the appearance and development of freely available rs-fMRI databases, such as the Autism Brain Imaging Data Exchange (Martino et al, 2014), which provides functional and structural brain imaging datasets collected from more than 24 different independent sites, researchers from different countries have expanded a series of studies based on it. The review about ASD classification is restricted to fMRI data for more specific analysis. The aforementioned fMRI data consist of rs-fMRI data and task-based fMRI data, which are collected from scanning the brain using fMRI technology while the subject is resting and performing a special task, respectively

Objectives

Methods

Results

Conclusion