Detection of autism spectrum disorder from changing of pupil diameter using multi-modal feature fusion based hybrid CNN model

This paper presents an multimodal feature fusion hybrid-CNN method to discriminate between typically developing (TD) and patients having autism spectrum disorder (ASD). Furthermore, it finds levels of those with ASD. The model, which uses the right and left pupil diameter variation of individuals, consists of six stages: augmentation, spectrogram, image fusion, feature extraction, selection, and classifier. ASD and TD discrimination and severity levels of patients having ASD are classified by SVM and k-NN algorithm, and 95.33% and 93.33% accuracy values are obtained respectively. The results are as follows: The proposed method is effective in determining the severity levels of TD patients and ASD patients. Pupil diameter, which is one of the gaze characteristics, can be used to detect ASD patients.