Scholarlink

<p>Attention deficit hyperactivity disorder (ADHD) is a neurodevelopmental disorder characterized by genetic predisposition and alterations in brain structural connectivity. While existing studies have established associations between genetic variants and neuroanatomical features, the specific relationships in ADHD remained poorly understood. To address this gap, we developed adversarial deep canonical correlation analysis models (A-DCCA) to disentangle ADHD-specific and non-specific “gene-white matter” association patterns. Utilizing diffusion tensor imaging and genotype data from six-hundred ADHD and typically developed children in a Chiese cohort, the current study revealed ADHD-specific correlations between the right cerebral peduncle, right posterior limb of the internal capsule, and genes regulating neural apoptotic processes (CAMK1D, METTL15, and MAP2K4). In contrast, associations involving the left cerebral peduncle, left posterior limb of the internal capsule, right superior longitudinal fasciculus, and right posterior thalamic radiation with genes related to early neural development (FYN, PHF2, ZSCAN31, and CD82) presented associations shared by ADHD and non-ADHD groups. Incorporating interpretable deep learning models, the current study unveiled white matter regions vulnerable to genetic influences in ADHD-specific and non-specific ways, shedding light on the understanding of biological substrates of ADHD.</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

Deep adversarial learning identifies ADHD-specific associations between apoptotic genes and white matter microstructure in frontal-striatum-cerebellum circuit

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年8月26日

查看全文
  • Yilu Zhao,
  • Xiangyu Zheng,
  • Xuping Gao,
  • Ning Wang,
  • Zhao Fu,
  • Junbin Tian,
  • Kangfuxi Zhang,
  • Peng Wang,
  • ShaoXian Li,
  • Jichang Zhang,
  • XueTong Ding,
  • Hui Zhang,
  • Li Sun,
  • Binrang Yang,
  • Shuyu Li,
  • Suhua Chang,
  • Qingjiu Cao,
  • Yufeng Wang,
  • Li Yang

摘要

Attention deficit hyperactivity disorder (ADHD) is a neurodevelopmental disorder characterized by genetic predisposition and alterations in brain structural connectivity. While existing studies have established associations between genetic variants and neuroanatomical features, the specific relationships in ADHD remained poorly understood. To address this gap, we developed adversarial deep canonical correlation analysis models (A-DCCA) to disentangle ADHD-specific and non-specific “gene-white matter” association patterns. Utilizing diffusion tensor imaging and genotype data from six-hundred ADHD and typically developed children in a Chiese cohort, the current study revealed ADHD-specific correlations between the right cerebral peduncle, right posterior limb of the internal capsule, and genes regulating neural apoptotic processes (CAMK1D, METTL15, and MAP2K4). In contrast, associations involving the left cerebral peduncle, left posterior limb of the internal capsule, right superior longitudinal fasciculus, and right posterior thalamic radiation with genes related to early neural development (FYN, PHF2, ZSCAN31, and CD82) presented associations shared by ADHD and non-ADHD groups. Incorporating interpretable deep learning models, the current study unveiled white matter regions vulnerable to genetic influences in ADHD-specific and non-specific ways, shedding light on the understanding of biological substrates of ADHD.