ORIGINAL RESEARCH article
Front. Psychiatry
Sec. Computational Psychiatry
Volume 16 - 2025 | doi: 10.3389/fpsyt.2025.1652891
This article is part of the Research TopicMedical Image Reconstruction and Big Data Analysis for Neurological DisordersView all articles
Object Detection in Neuroimaging: Enhancing Early Diagnosis of Neurological Disorders through Big Data Analysis
Provisionally accepted- Xinjiang University of Finance and Economics, Ürümqi, China
Select one of your emails
You have multiple emails registered with Frontiers:
Notify me on publication
Please enter your email address:
If you already have an account, please login
You don't have a Frontiers account ? You can register here
The accurate interpretation of neuroimaging data is critical for the early diagnosis of neurological disorders, which is increasingly feasible due to advancements in brain scanning technologies. Neurological diseases often involve subtle and widespread alterations in brain connectivity. These patterns require robust methodologies capable of extracting clinically relevant features from high-dimensional data. Traditional approaches, including graph theory metrics and convolutional neural networks, face challenges in simultaneously capturing localized disruptions and global structural changes, while also struggling to provide statistically interpretable outputs under varying data quality. To tackle the aforementioned problems, we design a hybrid framework that merges deep neural graph representations with statistical analysis, specifically designed for brain network analysis. The NeuroGraph Anomaly Detector (N-GAD) employs a multi-scale graph encoder that incorporates motif-augmented message passing, jump-aware feature aggregation, and spectral cross-convolution to capture both localized and distributed anomalies. This encoding is coupled with Contrastive Structural Differentiation (CSD), an inference strategy that formulates anomaly detection as a topological hypothesis testing problem using graphlet statistics, spectral, and Wasserstein-based distributional metrics. Experimental evaluations demonstrate that this framework improves sensitivity to pathological alterations while maintaining interpretability and statistical rigor, making it particularly suitable for large-scale neuroimaging applications aimed at early disorder diagnosis. This approach contributes to computational neurodiagnostics by enhancing both accuracy and transparency in clinical decision-making.
Keywords: Neuroimaging, Graph neural network, anomaly detection, structural inference, neurological disorders
Received: 24 Jun 2025; Accepted: 22 Sep 2025.
Copyright: © 2025 Pan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
* Correspondence: Yuting Pan, seziultu@hotmail.com
Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.