Scholarlink

<p>Autism Spectrum Disorder (ASD) is a complex neurodevelopmental condition influenced by both environmental and hereditary factors, making its diagnosis particularly challenging. Traditional detection methods typically rely on single-modal data, which often fail to provide satisfactory results due to the heterogeneous nature of ASD. To address this gap, this paper proposes a novel approach, named Secretary Bird Jellyfish Search Optimization algorithm with Convolutional LeNet Forward Taylor Network (SBJSO_CLeFTNet) that integrates multiple modalities, both image and clinical data, for ASD detection. The significance of this work lies in its multimodal fusion strategy and biologically inspired optimization to improve diagnostic accuracy. Image preprocessing is performed using the Adaptive Wiener Filter, followed by the extraction of the Region of Interest (ROI). Key features, including the Local Gradient Pattern (LGP) and statistical features, are extracted and considered as output-1. Simultaneously, autism-related clinical data are normalized using the Min–max approach, with relevant features selected through Matusita and chord distances. To address class imbalance, the Synthetic Minority Oversampling Technique (SMOTE) is applied to augment the clinical data, which forms output-2. These outputs are concatenated to create a combined feature set, which is then fed into the SBJSO_CLeFTNet model for ASD detection. The proposed model achieved impressive results, with an accuracy of 92.00%, sensitivity of 91.52%, and specificity of 91.67%, demonstrating its superior performance and strong potential for reliable ASD diagnosis using multimodal data.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Hybrid Secretary Bird Jellyfish Search Optimization Trained Deep Learning for Autism Spectrum Disorder Detection Using Multimodal Data

  • 期刊论文

  • 发表日期：

    2025年8月12日

查看全文
  • Aswathy Wilson,
  • J. Anitha

摘要

Autism Spectrum Disorder (ASD) is a complex neurodevelopmental condition influenced by both environmental and hereditary factors, making its diagnosis particularly challenging. Traditional detection methods typically rely on single-modal data, which often fail to provide satisfactory results due to the heterogeneous nature of ASD. To address this gap, this paper proposes a novel approach, named Secretary Bird Jellyfish Search Optimization algorithm with Convolutional LeNet Forward Taylor Network (SBJSO_CLeFTNet) that integrates multiple modalities, both image and clinical data, for ASD detection. The significance of this work lies in its multimodal fusion strategy and biologically inspired optimization to improve diagnostic accuracy. Image preprocessing is performed using the Adaptive Wiener Filter, followed by the extraction of the Region of Interest (ROI). Key features, including the Local Gradient Pattern (LGP) and statistical features, are extracted and considered as output-1. Simultaneously, autism-related clinical data are normalized using the Min–max approach, with relevant features selected through Matusita and chord distances. To address class imbalance, the Synthetic Minority Oversampling Technique (SMOTE) is applied to augment the clinical data, which forms output-2. These outputs are concatenated to create a combined feature set, which is then fed into the SBJSO_CLeFTNet model for ASD detection. The proposed model achieved impressive results, with an accuracy of 92.00%, sensitivity of 91.52%, and specificity of 91.67%, demonstrating its superior performance and strong potential for reliable ASD diagnosis using multimodal data.