Autism Spectrum Disorder Diagnosis Based on Attentional Feature Fusion Using NasNetMobile and DeiT Networks DOI Open Access

Zainab A. Altomi,

Yasmin M. Alsakar,

M. M. El-Gayar

et al.

Electronics, Journal Year: 2025, Volume and Issue: 14(9), P. 1822 - 1822

Published: April 29, 2025

Autism spectrum disorder (ASD) is a neurodevelopmental condition that affects social interactions, communication, and behavior. Prompt precise diagnosis essential for prompt support intervention. In this study, deep learning-based framework diagnosing ASD using facial images has been proposed. The methodology begins with logarithmic transformation image pre-processing, enhancing contrast making subtle features more distinguishable. Next, feature extraction performed NasNetMobile DeiT networks, where captures high-level abstract patterns, the network focuses on fine-grained characteristics relevant to identification. extracted are then fused attentional fusion, which adaptively assigns importance most discriminative features, ensuring an optimal representation. Finally, classification conducted bagging vector machine (SVM) classifier employing polynomial kernel, generalization robustness. Experimental results validate effectiveness of proposed approach, achieving 95.77% recall, 95.67% precision, 95.66% F1-score, accuracy, demonstrating its strong potential assisting in through analysis.

Language: Английский

Autism Spectrum Disorder Diagnosis Based on Attentional Feature Fusion Using NasNetMobile and DeiT Networks DOI Open Access

Zainab A. Altomi,

Yasmin M. Alsakar,

M. M. El-Gayar

et al.

Electronics, Journal Year: 2025, Volume and Issue: 14(9), P. 1822 - 1822

Published: April 29, 2025

Autism spectrum disorder (ASD) is a neurodevelopmental condition that affects social interactions, communication, and behavior. Prompt precise diagnosis essential for prompt support intervention. In this study, deep learning-based framework diagnosing ASD using facial images has been proposed. The methodology begins with logarithmic transformation image pre-processing, enhancing contrast making subtle features more distinguishable. Next, feature extraction performed NasNetMobile DeiT networks, where captures high-level abstract patterns, the network focuses on fine-grained characteristics relevant to identification. extracted are then fused attentional fusion, which adaptively assigns importance most discriminative features, ensuring an optimal representation. Finally, classification conducted bagging vector machine (SVM) classifier employing polynomial kernel, generalization robustness. Experimental results validate effectiveness of proposed approach, achieving 95.77% recall, 95.67% precision, 95.66% F1-score, accuracy, demonstrating its strong potential assisting in through analysis.

Language: Английский

Citations

0