Electroencephalogram (EEG)-based automated depression diagnosis systems have been suggested for early and accurate detection of mood disorders. EEG signals are highly irregular, nonlinear, and nonstationary in nature and are traditionally studied from a linear viewpoint by means of statistical and frequency features. Since, linear metrics present certain limitations and nonlinear methods have proven to be an efficient tool in understanding the complexities of the brain in the identification of underlying behavior of biological signals, such as electrocardiogram, EEG and magnetoencephalogram and thus, can be applied to all nonstationary signals. Various nonlinear algorithms can be used in the analysis of EEG signals. In this research paper, we aim to develop a novel methodology for EEG-based depression diagnosis utilizing 2 advanced computational techniques: frequency-domain extended multivariate autoregressive (eMVAR) and deep learning (DL). We proposed a hybrid method comprising a pretrained ResNet-50 and long-short term memory (LSTM) to capture depression-specific information and compared with a strong conventional machine learning (ML) framework having eMVAR connectivity features. The following 8 causality measures, which interpret the interaction mechanisms among spectrally decomposed oscillations, were used to extract features from multivariate EEG time series: directed coherence (DC), directed transfer function (DTF), partial DC (PDC), generalized PDC (gPDC), extended DC (eDC), delayed DC (dDC), extended PDC (ePDC), and delayed PDC (dPDC). The classification accuracies were 84% with DC, 85% with DTF, 95.3% with PDC, 95.1% with gPDC, 84.8% with eDC, 84.6% with dDC, 84.2% with ePDC, and 95.9% with dPDC for the eMVAR framework. Through a DL framework (ResNet-50 + LSTM), the classification accuracy was achieved as 90.22%. The results demonstrate that our DL methodology is a competitive alternative to the strong feature extraction-based ML methods in depression classification.
AcharyaURSudarshanVKAdeliHSanthoshJKohJEAdeliA. Computer-aided diagnosis of depression using EEG signals. Eur Neurol. 2015;73(5-6):329-336.
2.
BashashatiAFatourechiMWardRKBirchGE. A survey of signal processing algorithms in brain–computer interfaces based on electrical brain signals. J Neural Eng. 2007;4(2):32-57. doi:10.1088/1741-2560/4/2/r03
3.
KappenBGielenS. Neural networks: artificial intelligence and industrial applications. Proceedings of the Third Annual SNN Symposium on Neural Networks; September 14-15, 1995; Nijmegen, The Netherlands.
4.
BersonASmithSThearlingK. Building Data Mining Applications for CRM. McGraw-Hill; 2000.
5.
HanJKamberMPeiJ. Data Mining: Concepts and Techniques. Morgan Kaufmann; 2012.
6.
AcharyaURSudarshanVKAdeliH,et al.A novel depression diagnosis index using nonlinear features in EEG signals. Eur Neurol. 2015;74(1-2):79-83.
7.
XingfangDXinxinYRuiZBiCLiDYaoG. Classifying major depression patients and healthy controls using EEG, eye tracking and galvanic skin response data. J Affect Disord. 2019;251:156-161.
8.
CaiHChenYHanJZhangXHuB. Study on feature selection methods for depression detection using three-electrode EEG data. Interdiscip Sci: Comput Life Sci. 2018;10(3):558-565.
9.
MumtazWAliSSAYasinMAMMalikAS. A machine learning framework involving EEG-based functional connectivity to diagnose major depressive disorder (MDD). Med Biol Eng Comput. 2018;56(2):233-246.
10.
MumtazWMalikAS. A comparative study of different EEG reference choices for diagnosing unipolar depression. Brain Topogr. 2018;31(5):875-885.
11.
BachmannMPäeskeLKalevK,et al.Methods for classifying depression in single channel EEG using linear and nonlinear signal analysis. Comput Methods Programs Biomed. 2018;155:11-17.
12.
SharmaMAchuthPVDebDPuthankattilSDAcharyaUR. An automated diagnosis of depression using three-channel bandwidth-duration localized wavelet filter bank with EEG signals. Cogn Syst Res. 2018;52:508-520.
13.
BiKChattunMRLiuX,et al.Abnormal early dynamic individual patterns of functional networks in low gamma band for depression recognition. J Affect Disord. 2018;238:366-374.
14.
LiaoSCWuCTHuangHCChengWTLiuYH. Major depression detection from EEG signals using kernel eigen-filter-bank common spatial patterns. Sensors. 2017;17(6):1385.
15.
BachmannMLassJHinrikusH. Single channel EEG analysis for detection of depression. Biomed Signal Process Control. 2017;31:391-397.
16.
MumtazWXiaLAliSSAYasinMAMHussainMMalikAS. Electroencephalogram (EEG)-based computer-aided technique to diagnose major depressive disorder (MDD). Biomed Signal Process Control. 2017;31:108-115.
17.
MohammadiMAl-AzabFRaahemiB, et al.Data mining EEG signals in depression for their diagnostic value. BMC Med Inform Decis Mak. 2015;15(1):108.
18.
ErguzelTTTasCCebiM. A wrapper-based approach for feature selection and classification of major depressive disorder–bipolar disorders. Comput Biol Med. 2015;64:127-137.
19.
Khodayari-RostamabadAReillyJPHaseyGdeBruinHMacCrimmonD. Diagnosis of psychiatric disorders using EEG data and employing a statistical decision model. In: 2010 Annual International Conference of the IEEE Engineering in Medicine and Biology; August 2010:4006-4009. IEEE.
20.
AcharyaUROhSLHagiwaraYTanJHAdeliHSubhaDP. Automated EEG-based screening of depression using deep convolutional neural network. Comput Methods Programs Biomed. 2018;161:103-113.
21.
IercitanoCMammoneNHussainAMorabitoFC. A novel multi-modal machine learning based approach for automatic classification of EEG recordings in dementia. Neural Netw. 2020;123:176-190. doi:10.1016/j.neunet.2019.12.006
22.
MammoneNIercitanoCMorabitoFC. A deep CNN approach to decode motor preparation of upper limbs from time–frequency maps of EEG signals at source level. Neural Netw. 2020;124:357-372–12, doi:10.1016/j.neunet.2020.01.027
23.
GrangerCWJ. Investigating causal relations by econometric models and cross-spectral methods. Econometrica. 1969;37:424-438. doi:10.2307/1912791
24.
FaesLErlaSPortaANolloG. A framework for assessing frequency domain causality in physiological time series with instantaneous effects. Philos Trans R Soc A: Math Phys Eng Sci. 2013;371(1997):20110618-20110618. doi:10.1098/rsta.2011.0618
25.
FaesLNolloG. Multivariate frequency domain analysis of causal interactions in physiological time series. Biomed Eng Trends Electron Commun Softw. 2011:403-428. doi:10.5772/13065
26.
FaesLNolloGErlaSPapadelisCBraunCPortaA. Detecting nonlinear causal interactions between dynamical systems by non-uniform embedding of multiple time series. In: 2010 Annual International Conference of the IEEE Engineering in Medicine and Biology; 2010. doi:10.1109/iembs.2010.5626127
27.
FaesLErlaSNolloG. Measuring connectivity in linear multivariate processes: definitions, interpretation, and practical analysis. Comput Math Methods Med. 2012;(Special Issue):1-18. doi:10.1155/2012/140513
28.
CheungBLRiednerBATononiGVeenBV. Estimation of cortical connectivity from EEG using state-space models. IEEE Trans Biomed Eng. 2010;57(9):2122-2134. doi:10.1109/tbme.2010.2050319
29.
SongsiriJDahlJVandenbergheL. Graphical models of autoregressive processes. In: Palomar DP, Eldar YC, eds. Convex Optimization in Signal Processing and Communications. Cambridge University Press; 2009:89-116. doi:10.1017/cbo9780511804458.004
30.
FaesL. Assessing connectivity in the presence of instantaneous causality. In: Sameshima K, Baccala LA, eds. Methods in Brain Connectivity Inference through Multivariate Time Series Analysis Frontiers in Neuroengineering Series. CRC Press; 2014:87-112. doi:10.1201/b16550-8
31.
BlinowskaKJ. Review of the methods of determination of directed connectivity from multichannel data. Med Biol Eng Comput. 2011;49(5):521-529. doi:10.1007/s11517-011-0739-x
32.
BaccaláLASameshimaK. Partial directed coherence: a new concept in neural structure determination. Biol Cybern. 2001;84(6):463-474. doi:10.1007/pl00007990
33.
BastosAMSchoffelenJ. A tutorial review of functional connectivity analysis methods and their interpretational pitfalls. Front Syst Neurosci. 2016;9(84):1-23. doi:10.3389/fnsys.2015.00175
34.
WittenIHFrankEHallMAPalCJ. Deep learning. Data Min. 2017:417-466. doi:10.1016/b978-0-12-804291-5.00010-6
35.
WengYXiaC. A new deep learning-based handwritten character recognition system on mobile computing devices. Mobile Netw Appl. 2019;25:402-411. doi:10.1007/s11036-019-01243-5
36.
TisseraMDMcdonnellMD. Enhancing deep extreme learning machines by error backpropagation. In: 2016 International Joint Conference on Neural Networks (IJCNN); 2016. doi:10.1109/ijcnn.2016.7727273
37.
KanjoEYounisEMAngCS. Deep learning analysis of mobile physiological, environmental and location sensor data for emotion detection. Inf Fusion. 2019;49:46-56. doi:10.1016/j.inffus.2018.09.001
38.
CholletF. Deep Learning in Python: A Introduction to Convolutional Neural Networks. Manning; 2019. doi:10.4135/9781526493477
39.
IoffeSSzegedyC. Batch normalization: accelerating deep network training by reducing internal covariate shift. In: International Conference on Machine Learning; 2015:448-456.
40.
AyBYildirimOTaloM,et al.Automated depression detection using deep representation and sequence learning with EEG signals. J Med Syst. 2019;43(7):1-12. doi:10.1007/s10916-019-1345-y
41.
SzegedyCToshevAErhanD. Deep neural networks for object detection. In: NIPS; 2013.
42.
HeKZhangXRenSSunJ. Deep residual learning for image recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 2016. doi:10.1109/cvpr.2016.90
43.
SharmaPGuptaUKOzaMSharmaS. Reception – a deep learning based hybrid residual network. 2019 6th Swiss Conference on Data Science (SDS); 2019. doi:10.1109/sds.2019.00012
44.
WenLZhaoYLiSChengHZhangC. MST-ResNet: a multiscale spatial temporal ResNet for steering prediction. 2019 2nd China Symposium on Cognitive Computing and Hybrid Intelligence (CCHI); 2019. doi:10.1109/cchi.2019.8901931
45.
LimBLeeKM. Deep recurrent ResNet for video super-resolution. 2017 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC); 2017. doi:10.1109/apsipa.2017.8282261
46.
GuoHChenJ. Dynamic facial expression recognition based on ResNet and LSTM. IOP Conf Ser: Mater Sci Eng. 2020;790:012145. doi:10.1088/1757-899x/790/1/012145
47.
ZhouYZhangHLiYNingG. ECG Heartbeat classification based on ResNet and Bi-LSTM. IOP Conf Ser: Earth Environ Sci. 2020;428:012014. doi:10.1088/1755-1315/428/1/012014
48.
ChenZXieZZhangWXuX. ResNet and model fusion for automatic spoofing detection. Interspeech. 2017:102-106. doi:10.21437/interspeech.2017-1085
49.
Nagendra KumarM. Breast cancer classification of image using modified -ResNet. J Adv Res Dyn Control Syst. 2020;12(3):134-140. doi:10.5373/jardcs/v12i3/20201175
50.
TurkogluMHanbayDSengurA. Multi-model LSTM-based convolutional neural networks for detection of apple diseases and pests. J Ambient Intell Humaniz Comput. 2019:1-11. doi:10.1007/s12652-019-01591-w
51.
KarimFMajumdarSDarabiHChenS. LSTM Fully convolutional networks for time series classification. IEEE Access. 2018;6:1662-1669. doi:10.1109/access.2017.2779939
52.
SongJGaoLGuoZLiuWZhangDShenHT. Hierarchical LSTM with adjusted temporal attention for video captioning. Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence; 2017. doi:10.24963/ijcai.2017/381
53.
YouYHseuJYingCDemmelJKeutzerKHsiehC. Large-batch training for LSTM and beyond. Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis; 2019. doi:10.1145/3295500.3356137
54.
CourtneyLSreenivasR. Using deep convolutional LSTM networks for learning spatiotemporal features. Lect Notes Comput Sci Pattern Recognit. 2020:307-320. doi:10.1007/978-3-030-41299-9_24
FirstMSpitzerRGibbonMWilliamsJ. Structured Clinical Interview for DSM-IV Axis 1 Disorders, Clinical Version (SCID-CV). American Psychiatric Association; 1997.
57.
MaxwellME. Family Interview for Genetic Studies (FIGS): A Manual for FIGS. Clinical Neurogenetics Branch, Intramural Research Program, National Institute of Mental Health; 1992.
58.
MillerGAGrationGYeeCM. Generalized implementation of an eye movement correction procedure. Psychophysiology. 1988;25(2):241-243.
59.
UlrichGHaugHJFähndrichE. Acute vs. Chronic EEG effects in maprotiline- and in clomipramine-treated depressive inpatients and the prediction of therapeutic outcome. J Affect Disord. 1994;32(3):213-217.
60.
YangBLiHWangQZhangY. Subject-based feature extraction by using fisher WPD-CSP in brain–computer interfaces. Comput Methods Programs Biomed. 2016;129:21-28. doi:10.1016/j.cmpb.2016.02.020
61.
YanTJingtianTAndongG. Multi-class EEG classification for brain computer interface based on CSP. In: 2008 International Conference on BioMedical Engineering and Informatics; 2008. doi:10.1109/bmei.2008.72
62.
SongXYoonS. Improving brain–computer interface classification using adaptive common spatial patterns. Comput Biol Med. 2015;61:150-160. doi:10.1016/j.compbiomed.2015.03.023
63.
RamoserHMuller-GerkingJPfurtschellerG. Optimal spatial filtering of single trial EEG during imagined hand movement. IEEE Trans Rehabil Eng. 2000;8(4):441-446. doi:10.1109/86.895946
64.
BlankertzBTomiokaRLemmSKawanabeMMullerK. Optimizing spatial filters for robust EEG single-trial analysis. IEEE Signal Process Mag. 2008;25(1):41-56. doi:10.1109/msp.2008.4408441
65.
WuWChenZGaoSBrownEN.Hierarchical Bayesian modeling of inter-trial variability and variational Bayesian learning of common spatial patterns from multichannel EEG. 2010 IEEE International Conference on Acoustics, Speech and Signal Processing; 2010. doi:10.1109/icassp.2010.5495663
66.
BlankertzBMullerKCurioG,et al.The BCI competition 2003: progress and perspectives in detection and discrimination of EEG single trials. IEEE Trans Biomed Eng. 2004;51(6):1044-1051. doi:10.1109/tbme.2004.826692
67.
AngKKChinZYWangCGuanCZhangH. Filter bank common spatial pattern algorithm on BCI competition IV datasets 2a and 2b. Front Neurosci. 2012;6:1-9. doi:10.3389/fnins.2012.00039
68.
WuWChenZGaoXLiYBrownENGaoS. Probabilistic common spatial patterns for multichannel EEG analysis. IEEE Trans Pattern Anal Mach Intell. 2015;37(3):639-653. doi:10.1109/tpami.2014.2330598
69.
ChiangJWangZJMckeownMJ. A generalized multivariate autoregressive (GmAR)-based approach for EEG source connectivity analysis. IEEE Trans Signal Process. 2012;60(1):453-465. doi:10.1109/tsp.2011.2166392
70.
GutniskyDAJosićK. Generation of spatiotemporally correlated spike trains and local field potentials using a multivariate autoregressive process. J Neurophysiol. 2010;103(5):2912-2930. doi:10.1152/jn.00518.2009
71.
WangYKockelmanKM. A Poisson-lognormal conditional-autoregressive model for multivariate spatial analysis of pedestrian crash counts across neighborhoods. Accid Anal Prev. 2013;60:71-84. doi:10.1016/j.aap.2013.07.030
72.
ZhangRXuPLiuT,et al.Local temporal correlation common spatial patterns for single trial EEG classification during motor imagery. Comput Math Methods Med. 2013;2013:1-7. doi:10.1155/2013/591216
73.
ChiangJWangZMckeownM. A hidden Markov, multivariate autoregressive (HMM-mAR) network framework for analysis of surface EMG (sEMG) data. IEEE Trans Signal Process. 2008;56(8):4069-4081. doi:10.1109/tsp.2008.925246
74.
EEG Signal analysis via a cleaning procedure based on multivariate empirical mode decomposition. Proceedings of the 4th International Joint Conference on Computational Intelligence; 2012. doi:10.5220/0004182206700676
75.
SmithWAPHancockER.Modelling surface normal distribution using the azimuthal equidistant projection. In: MartinRBezHSabinM, eds. Mathematics of Surfaces XI. Lecture Notes in Computer Science. Vol 3604. Springer; 2005:381-394. doi:10.1007/11537908_23
76.
GokmenTOnenMHaenschW. Training deep convolutional neural networks with resistive cross-point devices. Front Syst Neurosci. 2017;11:1-13. doi:10.3389/fnins.2017.00538
77.
AcharyaUROhSLHagiwaraYTanJHAdeliHSubhaDP. Automated EEG-based screening of depression using deep convolutional neural network. Comput Methods Programs Biomed. 2018;161:103-113. doi:10.1016/j.cmpb.2018.04.012
78.
HuangGLiuZMaatenLVDWeinbergerKQ. Densely connected convolutional networks. In: 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 2017. doi:10.1109/cvpr.2017.243
79.
Efficacy detection of type-II diabetes mellitus with support vector machine and artificial neural network. In: Proceedings Of The International Conference On Advanced Theoretical Computer Applications & Proceedings Of The International Conference On Contemporary Management Practices; 2016. doi:10.18797/caasr/2ndicatca/iccmp/2016/05/05/04
80.
ErcelebiESubasiA. Classification of EEG for epilepsy diagnosis in wavelet domain using artificial neural network and multilinear regression. In: 2006 IEEE 14th Signal Processing and Communications Applications. doi:10.1109/siu.2006.1659852
81.
TunMT. Alphabet recognition system based on artificial neural network. Int J Trend Sci Res Dev. 2018;2(5):469-473. doi:10.31142/ijtsrd15854
82.
HeKZhangXRenSSunJ. Deep residual learning for image recognition. In: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 2016. 10.1109/cvpr.2016.90
83.
DuYLiX. Recursive deep residual learning for single image dehazing. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW); 2018. doi:10.1109/cvprw.2018.00116
84.
AcharyaURSudarshanVKAdeliHSanthoshJKohJEAdeliA. Computer-aided diagnosis of depression using EEG signals. Eur Neurol. 2015;73(5-6):329-336.
85.
Fernández-PalleiroPRivera-BaltanásTRodrigues-AmorimD, et al.Brainwaves oscillations as a potential biomarker for major depression disorder risk. Clin EEG Neurosci. 2019;51(1):1-7. doi:1550059419876807
86.
SubhaDPJosephPKAcharyaRLimCM. EEG Signal analysis: a survey. J Med Syst. 2010;34(2):195-212.
87.
LakshmiMRPrasadTVPrakashDVC. Survey on EEG signal processing methods. Int J Adv Res Comput Sci Softw Eng. 2014;4(1):84-91.
88.
KhoshnevisSASankarR. Applications of higher order statistics in electroencephalography signal processing: a comprehensive survey. IEEE Rev Biomed Eng. 2019;13:169-183. doi:10.1109/rbme.2019.2951328
89.
A framework for designing the architectures of deep convolutional neural networks. Entropy. 2017;19(6):242. doi:10.3390/e19060242
90.
MumtazWAliSSAYasinMAMMalikAS. A machine learning framework involving EEG-based functional connectivity to diagnose major depressive disorder (MDD). Med Biol Eng Comput. 2018;56(2):233-246.
91.
DunlopKTalishinskyA, Liston C intrinsic brain network biomarkers of antidepressant response: a review. Curr Psychiatry Rep. 2019;21(9):87.
92.
HelmKViolKWeigerTM,et al.Neuronal connectivity in major depressive disorder: a systematic review. Neuropsychiatr Dis Treat. 2018;2018(14):2715-2737.
93.
MumtazWQayyumA. A deep learning framework for automatic diagnosis of unipolar depression. Int J Med Inform. 2019;132:103983.
94.
GoodfellowIBengioYCourvilleA. Deep Learning. Cambridge (EE. UU.). MIT Press; 2016.
95.
GregorKDanihelkaIMnihABlundellCWierstraD. Deep autoregressive networks. arXiv preprint arXiv:1310.8499. 2013;2.
96.
ZhengYZhangYJLarochelleH. A deep and autoregressive approach for topic modeling of multimodal data. IEEE Trans Pattern Anal Mach Intell. 2016;38(6):10561069. doi:10.1109/tpami.2015.2476802
97.
KnottVMahoneyCKennedySEvansK. EEG Correlates of acute and chronic paroxetine treatment in depression. J Affect Disord. 2002;69(1-3), 241-249.
98.
NeckelmannDBjorvatnBBjørkumAAUrsinR. Citalopram: differential sleep/wake and EEG power spectrum effects after single dose and chronic administration. Behav Brain Res. 1996;79(1-2), 183-192.
99.
MohammadiMAl-AzabFRaahemiB, et al.Data mining EEG signals in depression for their diagnostic value. BMC Med Inform Decis Mak. 2015;15:108.
100.
ZeleninaMPrataDMP. Machine learning with electroencephalography features for precise diagnosis of depression subtypes. Paper presented at: MIDL 2019 Conference Abstract Submission; July 8-10, 2019; London, United Kingdom.
101.
UyulanCErgüzelTTUnubolH,et al.Major depressive disorder classification based on different convolutional neural network models: deep learning approach. Clin EEG Neurosci. 2020;52(1):38-51. doi:10.1177/1550059420916634
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
