Empirical Study of Autism Spectrum Disorder Diagnosis Using Facial Images by Improved Transfer Learning Approach

Abstract

Autism spectrum disorder (ASD) is a neurological illness characterized by deficits in cognition, physical activities, and social skills. There is no specific medication to treat this illness; only early intervention can improve brain functionality. Since there is no medical test to identify ASD, a diagnosis might be challenging. In order to determine a diagnosis, doctors consider the child's behavior and developmental history. The human face can be used as a biomarker as it is one of the potential reflections of the brain and thus can be used as a simple and handy tool for early diagnosis. This study uses several deep convolutional neural network (CNN)-based transfer learning approaches to detect autistic children using the facial image. An empirical study is conducted to select the best optimizer and set of hyperparameters to achieve better prediction accuracy using the CNN model. After training and validating with the optimized setting, the modified Xception model demonstrates the best performance by achieving an accuracy of 95% on the test set, whereas the VGG19, ResNet50V2, MobileNetV2, and EfficientNetB0 achieved 86.5%, 94%, 92%, and 85.8%, accuracy, respectively. Our preliminary computational results demonstrate that our transfer learning approaches outperformed existing methods. Our modified model can be employed to assist doctors and practitioners in validating their initial screening to detect children with ASD disease.

Keywords: ASD diagnosis; convolutional neural network (CNN); deep learning; facial image; transfer learning.

Figure 1

Flow diagram of the ASD article selection procedure for ablation study.

Figure 2

Proposed framework of the transfer learning approaches used in this study.

Figure 3

Modified Xception model structure (number of layers height, width, and filter size).

Figure 4

Demonstration of (a) max-pooling and average-pooling, and (b) classification layers.

Figure 5

Plots of model accuracy of (a) Xception, (b) VGG19, (c) EfficientNetB0, (d) MobileNetV2, and (e) Resnet50V2 following each epoch applied to both training and validation set.

Figure 6

Plots of model loss of (a) Xception, (b) VGG19, (c) EfficientNetB0, (d) MobileNetV2, and (e) Resnet50V2 following each epoch applied to both the training and validation set.

Figure 7

ROC curve of the models.

Figure 8

Confusion matrix of the five models (a) VGG19, (b) EfficientNetB0, (c) Xception, (d) MobileNetV2, and (e) ResNet50V2.

Figure 9

The 12 images in the top row are of (a) autistic children, and those misclassified as normal and vice-versa for the images are shown in (b) the lower row, which belong to normal control child (prediction of Xception).

Figure 10

ROC curves for all models using the dataset introduced by Ref. [28].

Figure 11

Confusion matrix of model (a) Xception (b) ResNet50V2 (c) MobileNetV2 on second test dataset.

Figure 12

Graph illustrating the best accuracy on second dataset and compared with referenced literature [28,44,45].