Deep Learning-Based Response-to-Name Detection: Empirical Study on Early Screening of Autism Spectrum Disorder in Children
Early screening and intervention for children with autism spectrum disorder (ASD) is crucial for their long-term outcomes and quality of life, and response-to-name (RTN) tests have shown promise in aiding early detection. Leveraging computer vision (CV) and deep learning techniques, this study explo...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10988848/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Early screening and intervention for children with autism spectrum disorder (ASD) is crucial for their long-term outcomes and quality of life, and response-to-name (RTN) tests have shown promise in aiding early detection. Leveraging computer vision (CV) and deep learning techniques, this study explores an automatic recognition algorithm for RTN tests based on Residual Network 18 (ResNet18) and Bidirectional Long Short-Term Memory (Bi-LSTM) for enhanced early screening of ASD in children. Using a dataset of 45 RTN cases, including 22 children with ASD and 23 typically developing (TD) children, various deep learning architectures such as Visual Geometry Group 16 (VGG16) and ResNet18 combined with Convolutional Neural Networks (CNN), Gated Recurrent Unit (GRU), and Bi-LSTM were employed to automate RTN test recognition and evaluate algorithm performance. The results revealed that the ResNet18-Bi-LSTM combination achieved the highest efficacy, with a remarkable accuracy of 100%, demonstrating robust stability and accuracy even with small datasets, and holding potential to assist in early ASD screening for children. |
|---|---|
| ISSN: | 2169-3536 |