[1]郇　战,张玉龙,陈　瑛,等.基于二分假设端到端深度学习的ADHD辅助诊断[J].郑州大学学报(工学版),2026,47(01):81-87.[doi:10.13705/j.issn.1671-6833.2025.04.008]
　HUAN Zhan,ZHANG Yulong,CHEN Ying,et al.Auxiliary Diagnosis of ADHD Based on Binary Hypothesis End-to-end Deep Learning[J].Journal of Zhengzhou University (Engineering Science),2026,47(01):81-87.[doi:10.13705/j.issn.1671-6833.2025.04.008]
点击复制

基于二分假设端到端深度学习的ADHD辅助诊断()
分享到：

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:
47
期数:
2026年01期
页码:
81-87
栏目:
出版日期:
2026-01-06

文章信息/Info

Title:
Auxiliary Diagnosis of ADHD Based on Binary Hypothesis End-to-end Deep Learning
文章编号:
1671-6833(2026)01-0081-07
作者:
郇　战1;  张玉龙2;  陈　瑛1;  王乐乐1
1.常州大学 微电子控制与工程学院,江苏 常州 213164;2.常州大学 计算机与人工智能学院,江苏 常州 213164
Author(s):
HUAN Zhan1;  ZHANG Yulong2;  CHEN Ying1;  WANG Lele1
1.School of Microelectronics and Control Engineering, Changzhou University, Changzhou 213164, China; 2.School of Computer Science and Artificial Intelligence, Changzhou University, Changzhou 213164, China
关键词:
注意缺陷多动障碍;  二分假设;  低频波动振幅;  端到端结构;  生物标识
Keywords:
attention deficit hyperactivity disorder;  binary hypothesis;  amplitude of low-frequency fluctuation;  endto-end;  biomarker
分类号:
TP391.41R749.93
DOI:
10.13705/j.issn.1671-6833.2025.04.008
文献标志码:
A
摘要:
在注意缺陷多动障碍(ADHD)的辅助诊断研究中,很多ADHD分类方法存在模型不能一体化或缺乏生物学解释的问题。为此,提出一种基于二分假设端到端深度学习的ADHD分类模型。在二分假设框架下,选择边缘系统相关的低频波动振幅数据作为输入特征,并设置注意力模块,使得网络重点关注分类贡献高的特征。模型整体形成端到端结构,而不是传统的深度学习和机器学习结合的结构。此外,完成检测生物标识的任务,提供生物学解释。在ADHD-200数据库的留一交叉验证实验中,4个子数据库上的平均准确率达到98.1%。随后,在边缘系统上进行ADHD生物标识的统计与分析,得到的ADHD生物标识分别为前扣带与旁扣带脑回、右杏仁核、嗅皮质和左杏仁核,这些结果证实了基于二分假设端到端深度学习模型的合理性。
Abstract:
In the auxiliary diagnosis studies of attention deficit hyperactivity disorder (ADHD), many ADHD classification methods suffer from the problem of model integration or lack of biological explanation. To address this, an ADHD classification model based on the binary hypothesis end-to-end deep learning was proposed. Within the binary hypothesis, amplitude of low-frequency fluctuation related to the limbic system was selected as input features. An attention module was incorporated to enable the network to focus on features with high classification contribution. The model adopted an end-to-end architecture, rather than the traditional deep learning and machine learning combined structure, and accomplished the task of detecting biomarkers, thus providing biological explanations. In leave-one-out cross-validation experiments on the ADHD-200 database, the average accuracy across four sub-databases reached 98.1%. Subsequently, statistical and analytical of ADHD biomarkers on the limbic system revealed ADHD biomarkers including the anterior cingulate and paracingulate gyri, right amygdala, olfactory cortex, and left amygdala. These results proved the rationality of the binary hypothesis end-to-end deep learning model.

参考文献/References:

[1]LEFFA D T, CAYE A, ROHDE L A. ADHD in children and adults: diagnosis and prognosis[J]Current Topics in Behavioral Neurosciences, 2022,57: 1-18.

[2]POLANCZYK G, JENSEN P. Epidemiologic considerations in attention deficit hyperactivity disorder: a review and update[J]. Child and Adolescent Psychiatric Clinics of North America, 2008, 17(2): 245-260.
[3]BARUA P D, DOGAN S, ACHARYA U R. EPSPatNet86: eight-pointed star pattern learning network for detection ADHD disorder using EEG signals[J]. Physiological Measurement, 2022, 43(3): 035002.
[4]QUAAK M, VAN DE MORTEL L, THOMAS R M, et al. Deep learning applications for the classification of psychiatric disorders using neuroimaging data: systematic review and meta-analysis [J]. NeuroImage Clinical, 2021, 30: 102584.
[5]TANG Y B, SUN J, WANG C, et al. ADHD classification using auto-encoding neural network and binary hypothesis testing[J]. Artificial Intelligence in Medicine, 2022, 123: 102209.
[6]SU L H, KAMATA S I. ADHD classification with lowfrequency fluctuation feature map based on 3D CBAMe[C]∥2022 7th International Conference on Biomedical Signal and Image Processing (ICBIP). New York: ACM, 2022: 74-79.
[7]COLBY J B, RUDIE J D, BROWN J A, et al. Insights into multimodal imaging classification of ADHD[J]. Frontiers in Systems Neuroscience, 2012, 6: 59.
[8]RIAZ A, ASAD M, ALONSO E, et al. DeepFMRI: endto-end deep learning for functional connectivity and classification of ADHD using fMRI[J]. Journal of Neuroscience Methods, 2020, 335: 108506.
[9]LIU S Q, ZHAO L, ZHAO J, et al. Attention deficit/hyperactivity disorder classification based on deep spatiotemporal features of functional magnetic resonance imaging[J]. Biomedical Signal Processing and Control, 2022, 71: 103239.
[10] QIANG N, DONG Q L, GE F F, et al. Deep variational autoencoder for mapping functional brain networks[J]. IEEE Transactions on Cognitive and Developmental Systems, 2021, 13(4): 841-852.
[11] QIANG N, DONG Q L, LIANG H T, et al. A novel ADHD classification method based on resting state temporal templates (RSTT) using spatiotemporal attention autoencoder[J]. Neural Computing and Applications, 2022, 34(10): 7815-7833.
[12] CHEN Y, GAO Y, JIANG A M, et al. ADHD classification combining biomarker detection with attention auto-encoding neural network[J]. Biomedical Signal Processing and Control, 2023, 84: 104733.
[13]李荪, 曹峰. 智能语音技术端到端框架模型分析和趋势研究[J]. 计算机科学, 2022, 49(增刊1): 331-336.
LI S, CAO F. Analysis and trend research of end-to-end framework model of intelligent speech technology[J]. Computer Science, 2022, 49(S1): 331-336.
[14] BONATH B, TEGELBECKERS J, WILKE M, et al. Regional gray matter volume differences between adolescents with ADHD and typically developing controls: further evidence for anterior cingulate involvement[J]. Journal of Attention Disorders, 2018, 22(7): 627-638.
[15] UGARTE G, PIÑA R, CONTRERAS D, et al. Attention deficit-hyperactivity disorder (ADHD): from abnormal behavior to impairment in synaptic plasticity[J]. Biology, 2023, 12(9): 1241.
[16] TANG Y B, WANG C, CHEN Y, et al. Identifying ADHD individuals from resting-state functional connectivity using subspace clustering and binary hypothesis testing[J]. Journal of Attention Disorders, 2021, 25(5): 736-748.
[17] TANG Y B, LI X F, CHEN Y, et al. High-accuracy classification of attention deficit hyperactivity disorder with l2, 1-norm linear discriminant analysis and binary hypothesis testing[J]. IEEE Access, 2020, 8: 56228-56237.
[18] SMITH S M, JENKINSON M, WOOLRICH M W, et al. Advances in functional and structural MR image analysis and implementation as FSL[J]. NeuroImage, 2004, 23(S1): S208-S219.
[19] BELLEC P, CHU C, CHOUINARD-DECORTE F, et al. The Neuro Bureau ADHD-200 Preprocessed repository[J]. NeuroImage, 2017, 144(Pt B): 275-286.
[20] LI X, YU R Q, HUANG Q, et al. Alteration of whole brain ALFF/ fALFF and degree centrality in adolescents with depression and suicidal ideation after electroconvulsive therapy: a resting-state fMRI study[J]. Frontiers in Human Neuroscience, 2021, 15: 762343.
[21] HU B, YU Y, WANG W, et al. MICA: a toolkit for multimodal image coupling analysis[J]. Journal of Neuroscience Methods, 2021, 347: 108962.

更新日期/Last Update: 2026-01-17