成功大學博碩士論文系統

簡易檢索 / 詳目顯示

回結果列表

研究生：	蔡文賓 Cai, Wun-Bin
論文名稱：	基於功能性磁振造影的疼痛學習任務下經皮耳迷走神經刺激效能個體差異預測與生物標記建立 Predicting Individual Differences in TaVNS Efficacy under Pain Learning Tasks Based on fMRI to Establish Predictive Biomarkers
指導教授：	詹慧伶 Chan, Hui-Ling
學位類別：	碩士 Master
系所名稱：	電機資訊學院 - 資訊工程學系 Department of Computer Science and Information Engineering
論文出版年：	2025
畢業學年度：	114
語文別：	英文
論文頁數：	74
中文關鍵詞：	功能性磁振造影、經皮耳迷走神經刺激、機器學習、深度學習、資料擴增
外文關鍵詞：	fMRI, taVNS, Machine Learning, Deep Learning, Data Augmentation
相關次數：	點閱：23 下載：1
分享至:	分享至facebook 分享至twitter

查詢本校圖書館目錄查詢臺灣博碩士論文知識加值系統勘誤回報

本研究探討如何利用事件相關功能性磁振造影（event-related fMRI），結合機器學習與深度學習方法，預測經皮耳迷走神經刺激（transcutaneous auricular vagus nerve stimulation, taVNS）的效能差異，並尋找具生物學意義的候選生物標記。臨床上，taVNS 的療效存在個體差異，但相關的神經機制與可預測的生物標記仍不明確。

在方法上，本研究的資料基於疼痛學習實驗，受試者在實驗進行中學習疼痛強度與視覺提示的關聯，並同時施予 taVNS。首先使用了傳統功能連結特徵與事件相關時序特徵於機器學習模型上進行 taVNS 的效能預測。其次，研究提出一種空間資料增強策略，透過調整既有方法，使其適用於三維 fMRI 體素，在不破壞結構完整性的前提下提升模型效能。最後進一步結合兩者的特徵分析方法建立一個具可解釋性的模型訓練框架，不僅能提高預測準確性，亦能呈現特徵對模型預測效能的貢獻。

結果顯示，時序特徵在區分反應者與非反應者上具有更佳表現，凸顯在事件相關資料分析中保留時間動態的重要性。再者，本研究提出的空間擴增方法對比不使用空間擴增方法的表現有所提升。最後，機器學習的特徵選取與深度學習的可解釋性方法皆捕捉一致的腦區。小腦、insula 與 ventromedial prefrontal cortex (vmPFC) 在兩種分析方法中皆被選為有助於模型預測的特徵。統計檢驗與相關分析進一步佐證這些腦區在預測 taVNS 效能的重要性，強化其作為 taVNS 效能潛在生物標記的可能。

整體而言，本研究使用空間資料擴增有效地提升了小規模 fMRI 資料集在深度學習模型上的表現，並且，提出的可解釋性模型訓練框架用以建立預測 taVNS效能差異的 fMRI 生物標記，為推動個人化神經調控策略奠定基礎。

This study investigates the use of event-related functional magnetic resonance imaging (fMRI), combined with machine learning and deep learning approaches, to predict variability in the efficacy of transcutaneous auricular vagus nerve stimulation (taVNS) and to identify biologically meaningful candidate biomarkers. Clinically, the therapeutic effects of taVNS vary across individuals, yet the underlying neural mechanisms and reliable predictive biomarkers remain unclear.

The data of this study were derived from a pain-learning experiment in which participants learned the association between pain intensity and visual cues while receiving taVNS. First, both conventional functional connectivity features and event-related temporal features were applied to machine learning models to predict taVNS efficacy. Second, a spatial data augmentation strategy was proposed, adapting existing methods for three-dimensional fMRI voxels to enhance model performance while preserving structural integrity. Finally, by integrating feature analysis from both approaches, an explainable model training framework was established, improving predictive accuracy while also providing transparent insights into the contributions of features to model performance.

The results demonstrated that temporal features achieved superior performance in distinguishing responders from non-responders, underscoring the importance of preserving temporal dynamics in event-related data analysis. Moreover, the proposed spatial augmentation method improved performance compared to models without augmentation. Importantly, convergent findings from machine learning feature selection and deep learning interpretability analyses identified consistent regions, including the cerebellum, insula, and ventromedial prefrontal cortex (vmPFC). Statistical testing and correlation analyses further confirmed the relevance of these regions, supporting their potential as candidate biomarkers for taVNS efficacy.

In summary, this study shows that spatial data augmentation can effectively enhance the performance of deep learning models on small fMRI datasets. Furthermore, the proposed explainable model training framework provides a pathway for establishing fMRI-based biomarkers of taVNS efficacy, laying the foundation for the development of personalized neuromodulation strategies.

摘要 i
Abstract iii
Acknowledgements v
Table of Contents vii
List of Tables ix
List of Figures x
List of Symbols xi
Chapter 1. Introduction 1
1. Background 1
1.1. Transcutaneous Auricular Vagus Nerve Stimulation (taVNS) 1
1.2. Importance of Individual Differences on taVNS Efficacy 1
1.3. taVNS and Interoceptive Regulation 3
1.4. Functional Magnetic Resonance Imaging (fMRI) in taVNS Research 4
2. Research Motivation 5
2.1. Clinical Necessity of Predicting Responders 5
2.2. Lack and Demand for Biomarkers 6
3. Literature Review 6
3.1. Machine Learning Applications in fMRI Classification 6
3.2. The Prediction of Individual Differences on taVNS Efficacy 7
3.3. Deep Learning Approaches in fMRI Analysis 10
3.4. Data Augmentation Strategies for fMRI 11
3.5. Challenges in Current Methods 14
4. Research Objectives 16
4.1. Predict Individual Differences of taVNS Efficacy Using Event-related fMRI Features 16
4.2. Enhance Model Robustness on Small fMRI Datasets 16
4.3. Develop an Explainable Model Training Framework 16
4.4. Identify Preliminary Predictive Biomarkers 17
5. Thesis Scope 17
6. Thesis Organization 18
Chapter 2. Materials and Methods 19
1. Participants 19
2. Experimental Design 19
3. Analysis of Behavioral Data 21
4. fMRI Data Acquisition and Preprocessing 23
4.1. Acquisition 23
4.2. Preprocessing 23
4.3. Trial-wise Data Extraction 24
5. Feature Extraction 24
5.1. ROI-Based Features 24
5.2. Voxel-wise Features 27
6. Machine Learning Models 27
7. Deep Learning Models 28
8. Data Augmentation Strategies 30
8.1. Spatial Augmentation 30
8.2. Temporal Augmentation 31
8.3. Spatial-Temporal Augmentation 32
9. Evaluation Metrics 33
10. Feature Analysis 35
10.1. Feature Stability in SVM Models 35
10.2. Grad-CAM Analysis in 3D CNN Models 35
11. Use of AI-assisted Technologies 36
Chapter 3. Results 37
1. Behavioral Data Results 37
2. Model Performance 40
3. Comparison between Original Data and Augmented Data 40
4. Feature Importance Analysis and Activation Heatmaps 41
4.1. Feature Importance from SVM 42
4.2. Statistical Analysis of Key Regions 42
4.3. Grad-CAM Heatmaps from 3D CNN 45
Chapter 4. Discussion 49
1. Development of Deep Learning Methods for Limited fMRI Data 49
2. Effectiveness of Temporal Features of Event-related fMRI in Efficacy Prediction 50
3. Evaluation of the Explainable Model Training Framework 51
4. Identification of Potential Biomarkers for Predicting taVNS Responders 51
5. Limitations of the Methods 53
6. Recommendations for Future Research 53
Chapter 5. Conclusion 55
References 57
                                    

[1] P. Rong, J. Liu, L. Wang, R. Liu, J. Fang, J. Zhao, Y. Zhao, H. Wang, M. Vangel, S. Sun, H. Ben, J. Park, S. Li, H. Meng, B. Zhu, and J. Kong, “Effect of transcutaneous auricular vagus nerve stimulation on major depressive disorder: A nonrandomized controlled pilot study,” Journal of Affective Disorders, vol. 195, pp. 172–179, 2016.
[2] E. Hein, M. Nowak, O. Kiess, T. Biermann, K. Bayerlein, J. Kornhuber, and T. Kraus, “Auricular transcutaneous electrical nerve stimulation in depressed patients: a randomized controlled pilot study,” Journal of Neural Transmission, vol. 120, pp. 821–827, May 2013.
[3] J. Sun, K. Sun, L. Chen, X. Li, K. Xu, C. Guo, Y. Ma, J. Cao, G. Zhang, Y. Hong, Z. Wang, S. Gao, Y. Luo, Q. Chen, W. Ye, X. Yu, X. Xiao, P. Rong, C. Yu, and J. Fang, “A predictive study of the efficacy of transcutaneous auricular vagus nerve stimulation in the treatment of major depressive disorder: An fMRI-based machine learning analysis,” Asian Journal of Psychiatry, vol. 98, p. 104079, 2024.
[4] C. Fu, Y. Zhang, Y. Ye, X. Hou, Z. Wen, Z. Yan, W. Luo, M. Feng, and B. Liu, “Predicting response to tVNS in patients with migraine using functional MRI: A voxel-based machine learning analysis,” Frontiers in Neuroscience, vol. 16, 2022.
[5] B. J. Jongkees, M. A. Immink, A. Finisguerra, and L. S. Colzato, “Transcutaneous vagus nerve stimulation (tVNS) enhances response selection during sequential action,” Frontiers in Psychology, vol. 9, 2018.
[6] D. H. Lench, T. H. Turner, C. McLeod, H. A. Boger, L. Lovera, L. Heidelberg, J. Elm, A. Phan, B. W. Badran, and V. K. Hinson, “Multi-session transcutaneous auricular vagus nerve stimulation for Parkinson’s disease: evaluating feasibility, safety, and preliminary efficacy,” Frontiers in Neurology, vol. 14, 2023.
[7] V. Villani, M. Tsakiris, and R. Azevedo, “Transcutaneous vagus nerve stimulation improves interoceptive accuracy,” Neuropsychologia, vol. 134, p. 107201, 2019.
[8] C. Ventura-Bort and M. Weymar, “Transcutaneous auricular vagus nerve stimulation modulates the processing of interoceptive prediction error signals and their role in allostatic regulation,” Human Brain Mapping, vol. 45, no. 3, p. e26613, 2024.
[9] A. D. Craig, “How do you feel? Interoception: the sense of the physiological condition of the body,” Nature Reviews Neuroscience, vol. 3, no. 8, pp. 655–666, 2002.
[10] A. V. Apkarian, M. C. Bushnell, R.-D. Treede, and J.-K. Zubieta, “Human brain mechanisms of pain perception and regulation in health and disease,” European Journal of Pain, vol. 9, no. 4, pp. 463–484, 2005.
[11] V. Legrain, G. D. Iannetti, L. Plaghki, and A. Mouraux, “The pain matrix reloaded: a salience detection system for the body,” Progress in Neurobiology, vol. 93, no. 1, pp. 111–124, 2011.
[12] H. D. Critchley, S. Wiens, P. Rotshtein, A. Öhman, and R. J. Dolan, “Neural systems supporting interoceptive awareness,” Nature Neuroscience, vol. 7, no. 2, pp. 189–195, 2004.
[13] A. Horsburgh, S. J. Summers, A. Lewis, R. J. Keegan, and A. Flood, “The relationship between pain and interoception: A systematic review and meta-analysis,” The Journal of Pain, vol. 25, no. 7, p. 104476, 2024.
[14] X. Wu, Y. Zhang, W.-t. Luo, R.-r. Mai, X.-y. Hou, Z.-q. Xia, B.-y. Xu, and B. Liu, “Brain functional mechanisms determining the efficacy of transcutaneous auricular vagus nerve stimulation in primary insomnia,” Frontiers in Neuroscience, vol. 15, 2021.
[15] J. Fang, N. Egorova, P. Rong, J. Liu, Y. Hong, Y. Fan, X. Wang, H. Wang, Y. Yu, Y. Ma, C. Xu, S. Li, J. Zhao, M. Luo, B. Zhu, and J. Kong, “Early cortical biomarkers of longitudinal transcutaneous vagus nerve stimulation treatment success in depression,” NeuroImage: Clinical, vol. 14, pp. 105–111, 2017.
[16] C. Fu, X. Hou, C. Zheng, Y. Zhang, Z. Gao, Z. Yan, Y. Ye, and B. Liu, “Immediate modulatory effects of transcutaneous vagus nerve stimulation on patients with Parkinson’s disease: a crossover self-controlled fMRI study,” Frontiers in Aging Neuroscience, vol. 16, 2024.
[17] G. Chen, B. D. Ward, C. Xie, W. Li, Z. Wu, J. L. Jones, M. Franczak, P. Antuono, and S.-J. Li, “Classification of Alzheimer disease, mild cognitive impairment, and normal cognitive status with large-scale network analysis based on resting-state functional MRI,” Radiology, vol. 259, no. 1, pp. 213–221, 2011.
[18] A. Khazaee, A. Ebrahimzadeh, and A. Babajani-Feremi, “Identifying patients with Alzheimer’s disease using resting-state fMRI and graph theory,” Clinical Neurophysiology, vol. 126, no. 11, pp. 2132–2141, 2015.
[19] D. Long, J. Wang, M. Xuan, Q. Gu, X. Xu, D. Kong, and M. Zhang, “Automatic classification of early Parkinson’s disease with multi-modal MR imaging,” PLOS ONE, vol. 7, no. 11, p. e47714, 2012.
[20] R. C. Welsh, L. M. Jelsone-Swain, and B. R. Foerster, “The utility of independent component analysis and machine learning in the identification of the amyotrophic lateral sclerosis diseased brain,” Frontiers in Human Neuroscience, vol. 7, p. 251, 2013.
[21] A. Lord, D. Horn, M. Breakspear, and M. Walter, “Changes in community structure of resting state functional connectivity in unipolar depression,” PLOS ONE, 2012.
[22] L.-L. Zeng, H. Shen, L. Liu, L. Wang, B. Li, P. Fang, Z. Zhou, Y. Li, and D. Hu, “Identifying major depression using whole-brain functional connectivity: a multivariate pattern analysis,” Brain, vol. 135, no. 5, pp. 1498–1507, 2012.
[23] F. Zhao, Z. Chen, I. Rekik, S.-W. Lee, and D. Shen, “Diagnosis of autism spectrum disorder using central-moment features from low- and high-order dynamic resting-state functional connectivity networks,” Frontiers in Neuroscience, vol. 14, 2020.
[24] T. Price, C.-Y. Wee, W. Gao, and D. Shen, “Multiple-network classification of childhood autism using functional connectivity dynamics,” International Conference on Medical Image Computing and Computer-Assisted Intervention, pp. 177–184, Springer, 2014.
[25] A. Eloyan, J. Muschelli, M. B. Nebel, H. Liu, F. Han, T. Zhao, A. D. Barber, S. Joel, J. J. Pekar, S. H. Mostofsky, et al., “Automated diagnoses of attention deficit hyperactive disorder using magnetic resonance imaging,” Frontiers in Systems Neuroscience, vol. 6, p. 61, 2012.
[26] D.-E. Meskaldji, M. G. Preti, T. A. Bolton, M.-L. Montandon, C. Rodriguez, S. Morgenthaler, P. Giannakopoulos, S. Haller, and D. Van De Ville, “Prediction of long-term memory scores in MCI based on resting-state fMRI,” NeuroImage: Clinical, vol. 12, pp. 785–795, 2016.
[27] H. Zhang, R. Song, L. Wang, L. Zhang, D. Wang, C. Wang, and W. Zhang, “Classification of brain disorders in rs-fMRI via local-to-global graph neural networks,” IEEE Transactions on Medical Imaging, vol. 42, no. 2, pp. 444–455, 2023.
[28] J. Chen and C. Park, “A deep learning paradigm for medical imaging data,” Expert Systems with Applications, vol. 255, p. 124480, 2024.
[29] Y. Wang, H. Long, T. Bo, and J. Zheng, “Residual graph transformer for autism spectrum disorder prediction,” Computer Methods and Programs in Biomedicine, vol. 247, p. 108065, 2024.
[30] N. U. Islam, R. Khanam, and A. Kumar, “Using 3D CNN for classification of Parkinson’s disease from resting-state fMRI data,” Journal of Engineering and Applied Science, vol. 70, no. 1, p. 89, 2023.
[31] X. Wang, X. Liang, Z. Jiang, B. A. Nguchu, Y. Zhou, Y. Wang, H. Wang, Y. Li, Y. Zhu, F. Wu, J.-H. Gao, and B. Qiu, “Decoding and mapping task states of the human brain via deep learning,” Human Brain Mapping, vol. 41, no. 6, pp. 1505–1519, 2020.
[32] Y.-Y. Wu, Y.-S. Hu, J. Wang, Y.-F. Zang, and Y. Zhang, “Toward precise localization of abnormal brain activity: 1D CNN on single voxel fMRI time-series,” Frontiers in Computational Neuroscience, vol. 16, 2022.
[33] A. Di Martino, C.-G. Yan, Q. Li, E. Denio, F. Castellanos, K. Alaerts, J. Anderson, M. Assaf, S. Bookheimer, M. Dapretto, B. Deen, S. Delmonte, I. Dinstein, E.-W. Birgit, D. Fair, L. Gallagher, D. Kennedy, C. Keown, C. Keysers, and M. Milham, “The autism brain imaging data exchange: Towards large-scale evaluation of the intrinsic brain architecture in autism,” Molecular Psychiatry, vol. 19, 2013.
[34] R. C. Petersen, P. S. Aisen, L. A. Beckett, M. C. Donohue, A. C. Gamst, D. J. Harvey, C. R. Jack, W. J. Jagust, L. M. Shaw, A. W. Toga, J. Q. Trojanowski, and M. W. Weiner, “Alzheimer’s disease neuroimaging initiative (ADNI),” Neurology, vol. 74, no. 3, pp. 201–209, 2010.
[35] Y. Fang, J. Zhang, L. Wang, Q. Wang, and M. Liu, “ACTION: Augmentation and computation toolbox for brain network analysis with functional MRI,” NeuroImage, vol. 305, p. 120967, 2025.
[36] B. Efron, The Jackknife, the Bootstrap and Other Resampling Plans. SIAM, 1982.
[37] M. F. Glasser, S. N. Sotiropoulos, J. A. Wilson, T. S. Coalson, B. Fischl, J. L. Andersson, J. Xu, S. Jbabdi, M. Webster, J. R. Polimeni, D. C. Van Essen, and M. Jenkinson, “The minimal preprocessing pipelines for the Human Connectome Project,” NeuroImage, vol. 80, pp. 105–124, 2013.
[38] G. Chen, P. A. Taylor, R. C. Reynolds, E. Leibenluft, D. S. Pine, M. A. Brotman, D. Pagliaccio, and S. P. Haller, “BOLD response is more than just magnitude: Improving detection sensitivity through capturing hemodynamic profiles,” NeuroImage, vol. 277, p. 120224, 2023.
[39] L. Fan, H. Li, J. Zhuo, Y. Zhang, J. Wang, L. Chen, Z. Yang, C. Chu, S. Xie, A. R. Laird, P. T. Fox, S. B. Eickhoff, C. Yu, and T. Jiang, “The human Brainnetome atlas: A new brain atlas based on connectional architecture,” Cerebral Cortex, vol. 26, pp. 3508–3526, Aug. 2016.
[40] J. Diedrichsen, “A spatially unbiased atlas template of the human cerebellum,” NeuroImage, vol. 33, pp. 127–138, Oct. 2006.
[41] A. Fermin, K. Friston, and S. Yamawaki, “An insula hierarchical network architecture for active interoceptive inference,” Royal Society Open Science, vol. 9, 2022.
[42] C. H. Lubba, S. Sethi, P. Knaute, S. Schultz, B. Fulcher, and N. Jones, “catch22: Canonical time-series characteristics selected through highly comparative time-series analysis,” Data Mining and Knowledge Discovery, vol. 33, 2019.
[43] F. Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V. Dubourg, J. Vanderplas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot, E. Duchesnay, and G. Louppe, “Scikit-learn: Machine learning in Python,” Journal of Machine Learning Research, vol. 12, 2012.
[44] C. Cortes and V. Vapnik, “Support-vector networks,” Machine Learning, vol. 20, no. 3, pp. 273–297, 1995.
[45] A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, A. Desmaison, A. Köpf, E. Yang, Z. DeVito, M. Raison, A. Tejani, S. Chilamkurthy, B. Steiner, L. Fang, J. Bai, and S. Chintala, “PyTorch: An imperative style, high-performance deep learning library,” 2019.
[46] B. Xin, L. Hu, Y. Wang, and W. Gao, “Stable feature selection from brain sMRI,” Proceedings of the AAAI Conference on Artificial Intelligence, vol. 29, 2015.
[47] R. R. Selvaraju, A. Das, R. Vedantam, M. Cogswell, D. Parikh, and D. Batra, “Grad-CAM: Why did you say that? Visual explanations from deep networks via gradient-based localization,” CoRR, vol. abs/1610.02391, 2016.
[48] A. Ploghaus, I. Tracey, J. S. Gati, S. Clare, R. S. Menon, P. M. Matthews, and J. N. P. Rawlins, “Dissociating pain from its anticipation in the human brain,” Science, vol. 284, no. 5422, pp. 1979–1981, 1999.
[49] D. Restuccia, G. D. Marca, M. Valeriani, M. G. Leggio, and M. Molinari, “Cerebellar damage impairs detection of somatosensory input changes: A somatosensory mismatch negativity study,” Brain, vol. 130, no. 1, pp. 276–287, 2007.
[50] H. Bastuji, M. Frot, C. Perchet, M. Magnin, and L. Garcia-Larrea, “Pain networks from the inside: Spatiotemporal analysis of brain responses leading from nociception to conscious perception,” Human Brain Mapping, vol. 37, no. 12, pp. 4301–4315, 2016.

校內：立即公開
校外：立即公開

簡易檢索 / 詳目顯示

相關論文