ASD-SAENet: A Sparse Autoencoder, and Deep-Neural Network Model for Detecting Autism Spectrum Disorder (ASD) Using fMRI Data

Abstract

Autism spectrum disorder (ASD) is a heterogenous neurodevelopmental disorder which is characterized by impaired communication, and limited social interactions. The shortcomings of current clinical approaches which are based exclusively on behavioral observation of symptomology, and poor understanding of the neurological mechanisms underlying ASD necessitates the identification of new biomarkers that can aid in study of brain development, and functioning, and can lead to accurate and early detection of ASD. In this paper, we developed a deep-learning model called ASD-SAENet for classifying patients with ASD from typical control subjects using fMRI data. We designed and implemented a sparse autoencoder (SAE) which results in optimized extraction of features that can be used for classification. These features are then fed into a deep neural network (DNN) which results in superior classification of fMRI brain scans more prone to ASD. Our proposed model is trained to optimize the classifier while improving extracted features based on both reconstructed data error and the classifier error. We evaluated our proposed deep-learning model using publicly available Autism Brain Imaging Data Exchange (ABIDE) dataset collected from 17 different research centers, and include more than 1,035 subjects. Our extensive experimentation demonstrate that ASD-SAENet exhibits comparable accuracy (70.8%), and superior specificity (79.1%) for the whole dataset as compared to other methods. Further, our experiments demonstrate superior results as compared to other state-of-the-art methods on 12 out of the 17 imaging centers exhibiting superior generalizability across different data acquisition sites and protocols. The implemented code is available on GitHub portal of our lab at: https://github.com/pcdslab/ASD-SAENet.

Keywords: ABIDE; ASD; autoencoder; classification; deep-learning; diagnosis; fMRI; sparse autoencoder.

Figure 1

An overview of our proposed model and how the sparse autoencoder is used as feature selection to the deep neural network. The limitation of the feature lead to better generalizability of the model across various data-acquisition sites, and may lead to better interpretability of the models.

Figure 2

(A) Shows the preprocessing steps which include extract time series from fMRI scans, calculate the Pearson's correlations, and then 1/4 smallest and largest average pairwise correlations were selected for feature vectors. (B) Explain how our model is trained at the same time to improve feature selection while obtaining optimal classification model. The DNN classifier input is the bottleneck of the SAE. (C) Shows the testing process where the input subject is fed into the trained SAE, and then the DNN will take the bottleneck to make the classification using softmax layer.