Scholarlink

Early diagnosis of neurological diseases is needed in order to enable early intervention and improved patient outcomes. In this paper, we propose a new method using Spatio-Temporal Graph Neural Networks (ST-GNNs) to diagnose functional brain networks from fMRI scans. Our method can assess spatial connectivity of networks and temporal trends in brain regions to identify early biomarkers of Alzheimer's and Parkinson's disease conditions. Comparing on the ADNI and PPMI data sets, our model achieved a mean classification accuracy of 93.2% on average, which is 7.8% better than that of CNN and RNN-based models. The model also identified 15% sensitivity gain on early- stage cases. Such performance indicates the potential of ST-GNNs to support clinical diagnosis from machine-decoded, explainable neuroimaging analysis.

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

Early Detection of Neurological Disorders Through Functional Brain Network Analysis Using Spatio-Temporal Graph Neural Networks

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Someswari Perla,
  • Bura Vijay Kumar,
  • Balajee Maram,
  • Alok Misra,
  • Maram Sai Shashank Raj,
  • U. D. Prasan

摘要

Early diagnosis of neurological diseases is needed in order to enable early intervention and improved patient outcomes. In this paper, we propose a new method using Spatio-Temporal Graph Neural Networks (ST-GNNs) to diagnose functional brain networks from fMRI scans. Our method can assess spatial connectivity of networks and temporal trends in brain regions to identify early biomarkers of Alzheimer's and Parkinson's disease conditions. Comparing on the ADNI and PPMI data sets, our model achieved a mean classification accuracy of 93.2% on average, which is 7.8% better than that of CNN and RNN-based models. The model also identified 15% sensitivity gain on early- stage cases. Such performance indicates the potential of ST-GNNs to support clinical diagnosis from machine-decoded, explainable neuroimaging analysis.