Scholarlink

<p>Attention deficit hyperactivity disorder (ADHD) is a neurodevelopmental disorder that has become increasingly prevalent among children of school age in recent years. An accurate diagnosis of ADHD and its subtypes is critical for effective intervention and management. Misdiagnosis can lead to inappropriate treatment and support, potentially worsening the child’s difficulties. This study developed an effective deep learning classification model to accurately classify children into typically developing (TD) and ADHD subtypes (ADHD-Inattentive and ADHD-Combined). We proposed a multiclass conditional oversampling convolutional neural network classification model using the middle slices of grey matter from T1-weighted magnetic resonance imaging (MRI). The proposed classification model uses five middle T1-weighted MRI slices of grey matter to achieve a maximum accuracy of 95.5275<InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12652_2024_4950_Article_IEq2.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="15" /> </InlineMediaObject> <EquationSource Format="TEX">\(\%\)</EquationSource> </InlineEquation>. This classification model uses the local interpretable model-agnostic explanations method to identify the regions responsible for TD and ADHD subtypes. The brain regions responsible with ADHD-Inattentive are identified as the anterior superior frontal gyrus, middle superior frontal gyrus, and posterior superior frontal gyrus. ADHD-Combined, on the other hand, is associated with additional regions such as the body of corpus callosum, posterior cingulate, precuneus, paracentral gyrus, and cerebellum. This study may help healthcare providers diagnose ADHD subtypes accurately.</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

Local interpretable model-agnostic explanations guided brain magnetic resonance imaging classification for identifying attention deficit hyperactivity disorder subtypes

  • 期刊论文

  • 发表日期：

    2025年2月24日

查看全文
  • K. Usha Rupni,
  • P. Aruna Priya

摘要

Attention deficit hyperactivity disorder (ADHD) is a neurodevelopmental disorder that has become increasingly prevalent among children of school age in recent years. An accurate diagnosis of ADHD and its subtypes is critical for effective intervention and management. Misdiagnosis can lead to inappropriate treatment and support, potentially worsening the child’s difficulties. This study developed an effective deep learning classification model to accurately classify children into typically developing (TD) and ADHD subtypes (ADHD-Inattentive and ADHD-Combined). We proposed a multiclass conditional oversampling convolutional neural network classification model using the middle slices of grey matter from T1-weighted magnetic resonance imaging (MRI). The proposed classification model uses five middle T1-weighted MRI slices of grey matter to achieve a maximum accuracy of 95.5275 \(\%\) . This classification model uses the local interpretable model-agnostic explanations method to identify the regions responsible for TD and ADHD subtypes. The brain regions responsible with ADHD-Inattentive are identified as the anterior superior frontal gyrus, middle superior frontal gyrus, and posterior superior frontal gyrus. ADHD-Combined, on the other hand, is associated with additional regions such as the body of corpus callosum, posterior cingulate, precuneus, paracentral gyrus, and cerebellum. This study may help healthcare providers diagnose ADHD subtypes accurately.