Motor Imagery EEG Classification for Patients with Amyotrophic Lateral Sclerosis Using Fractal Dimension and Fisher’s Criterion-Based Channel Selection
<p>Electrode placement. Thirty-three channels are used, among which two are references (A1, A2), one is the ground (GND), and the remaining are used for EEG recording. The positions of the electrodes follow the international 10–20 system.</p> "> Figure 2
<p>The motor imagery tasks include left hand grasping, right hand grasping, and stepping across a black line in front of the patient with left foot. The three pictures were presented during the preparation stage so that the patients could practice the designated motor imagery following the movements shown in the pictures.</p> "> Figure 3
<p>Resting and motor imagery EEG data collection. The motor imagery data collection consisted of three runs separated by 1- to 3-min breaks, and each run had 40 trials. One trial started with a 5-s ready-start period followed by a 3-s motor imagery period during which the participants performed the designated motor imagery task according to both visual and auditory cues presented in the ready-to-start period.</p> "> Figure 4
<p>Plots of <math display="inline"> <semantics> <mrow> <msub> <mi>d</mi> <mi>c</mi> </msub> </mrow> </semantics> </math> against <span class="html-italic">M</span> for four 3-s EEG signals recorded under resting, left hand imagery, right hand imagery, and left foot imagery, respectively, and were all recorded at F7 from the patient participant Sub1. The correlation dimension <math display="inline"> <semantics> <mrow> <msub> <mi>d</mi> <mi>c</mi> </msub> </mrow> </semantics> </math> value marked in black in each curve is the FD estimated by the GP method. In this example, the <math display="inline"> <semantics> <mrow> <msub> <mi>d</mi> <mi>c</mi> </msub> </mrow> </semantics> </math> values for resting, left hand, right hand, and left foot conditions reached the plateau before <math display="inline"> <semantics> <mrow> <mi>M</mi> <mo>=</mo> <mn>30</mn> </mrow> </semantics> </math> (i.e., <math display="inline"> <semantics> <mrow> <mi>M</mi> </mrow> </semantics> </math> = 23, 20, 21, and 13, respectively). The FDs of the four signals are 4.2479, 3.4062, 3.9438, and 3.4577, respectively. The time delay was set as 50 for the four cases.</p> "> Figure 5
<p>Log-Log plots of <math display="inline"> <semantics> <mrow> <mi>L</mi> <mrow> <mo>(</mo> <mi>k</mi> <mo>)</mo> </mrow> <mo> </mo> </mrow> </semantics> </math> versus <math display="inline"> <semantics> <mrow> <mn>1</mn> <mo>/</mo> <mi>k</mi> </mrow> </semantics> </math> based on the 3-s EEG signals from (<b>a</b>) resting, (<b>b</b>) left hand (LH) imagery, (<b>c</b>) right hand (RH) imagery, and (<b>d</b>) left foot (LF) imagery, respectively. The four signals were recorded at the same channel F7, and were collected from the patient participant Sub1. In this example, the fractal dimensions (FDs) estimated by the Higuchi method for the four tasks are 1.8202, 1.7359, 1.8004, and 1.8004, respectively.</p> "> Figure 6
<p>Cross validation accuracies based on 30 channels versus different values of time delay. The error bars are the standard error of the mean.</p> "> Figure 7
<p>Comparison of the average <span class="html-italic">K-</span>NN classification accuracies among different feature extraction methods using (<b>a</b>) 30 EEG channels, (<b>b</b>) 5 EEG channels, and (<b>c</b>) single EEG channel. The five channels are the ones have the top five Fisher scores among the 30, and the channel used in (<b>c</b>) is the best among the five in (<b>b</b>). <b>*</b> indicates significant difference between methods at <span class="html-italic">p</span> < 0.05.</p> "> Figure 8
<p>Average GPFD topoplots, where (<b>a</b>,<b>b</b>) display the resting and right hand (RH) imagery topoplots of patient Sub1, respectively, (<b>c</b>) shows the top five channels and their Fisher scores, (<b>d</b>,<b>e</b>) display the resting and RH imagery topoplots of patient Sub3, respectively, and (<b>f</b>) shows the patient Sub3’s top five channels and their Fisher scores. The values shown in (<b>a</b>,<b>b,d,e</b>) are average GPFD values. The top five channels in (<b>c</b>,<b>f</b>) are marked in yellow. The value of each best channel is the Fisher score. For example, the Fisher score of the patient Sub1’s channel Cz is 0.3829. The best channels for patients Sub1 and Sub3 are TP7 and C4, respectively.</p> "> Figure 9
<p>Comparison of the accuracies between the top-five-channel configuration and the five channels (C3, Cz, C4, CP3, CP4) within the sensorimotor area. <b>*</b> indicates significant difference between EEG configurations (top five channels and the five channels on the sensorimotor area) for each method at <span class="html-italic">p</span> < 0.05.</p> ">
Abstract
:1. Introduction
1.1. Related Works
1.2. Proposed Work
2. Materials and Methods
2.1. Participants
2.2. EEG Recording and Setting
2.3. Preparation
2.4. EEG Data Collection
2.5. Feature Extraction
2.5.1. GPFD
- Step 1.
- initialize ;
- Step 2.
- ;
- Step 3.
- calculate the value at the given ;
- Step 4.
- repeat step 2 and step 3 until ;
- Step 5.
- .
2.5.2. HFD
2.6. Channel Selection Based on Fisher’s Criterion
2.7. Classification and Parameter Optimization
3. Results and Discussion
3.1. Results
3.2. Discussion
4. Conclusions
Acknowledgments
Author Contributions
Conflicts of Interest
References
- Lotte, F.; Congedo, M.; Lécuyer, A.; Lamarche, F.; Arnaldi, B. A review of classification algorithms for EEG-based brain–computer interfaces. J. Neural Eng. 2007, 4, R1–R13. [Google Scholar] [CrossRef] [PubMed]
- Porro, C.A.; Francescato, M.P.; Cettolo, V.; Diamond, M.E.; Baraldi, P.; Zuiani, C.; Bazzocchi, M.; diPrampero, P.E. Primary motor and sensory cortex activation during motor performance and motor imagery: A functional magnetic resonance imaging study. J. Neurosci. 1996, 16, 7688–7698. [Google Scholar] [PubMed]
- Khan, M.J.; Hong, M.J.; Hong, K.-S. Decoding of four movement directions using hybrid NIRS-EEG brain-computer interface. Front. Hum. Neurosci. 2014, 8, 1–10. [Google Scholar] [CrossRef] [PubMed]
- Naseer, N.; Hong, M.J.; Hong, K.-S. Online binary decision decoding using functional near-infrared spectroscopy for development of a brain-computer interface. Exp. Brain Res. 2014, 232, 555–564. [Google Scholar] [CrossRef] [PubMed]
- Naseer, N.; Hong, K.-S. fNIRS-based brain-computer interfaces: A review. Front. Hum. Neurosci. 2015, 9, 1–15. [Google Scholar] [CrossRef] [PubMed]
- Liu, X.; Hong, K.-S. Detection of primary RGB colors projected on a screen using fNIRS. J. Innov. Opt. Health Sci. 2017, 10, 1750006. [Google Scholar] [CrossRef]
- Marchetti, M.; Priftis, K. Brain–computer interfaces in amyotrophic lateral sclerosis: A metanalysis. Clin. Neurophysiol. 2015, 126, 1255–1263. [Google Scholar] [CrossRef] [PubMed]
- Strong, M.J.; Grace, G.M.; Freedman, M.; Lomen-Hoerth, C.; Woolley, S.; Goldstein, L.H.; Murphy, J.; Shoesmith, C.; Rosenfeld, J.; Leigh, P.N.; et al. Consensus criteria for the diagnosis of frontotemporal cognitive and behavioural syndromes in amyotrophic lateral sclerosis. Amyotroph. Later. Scler. 2009, 10, 131–146. [Google Scholar] [CrossRef]
- Geronimo, A.; Simmons, Z.; Schiff, S.J. Performance predictors of brain–computer interfaces in patients with amyotrophic lateral sclerosis. J. Neural Eng. 2016, 13, 026002. [Google Scholar] [CrossRef] [PubMed]
- Moghimi, S.; Kushki, A.; Guerguerian, A.M.; Chau, T. A review of EEG-based brain-computer interfaces as access pathways for individuals with severe disabilities. Assist. Technol. 2012, 25, 99–110. [Google Scholar] [CrossRef] [PubMed]
- Kübler, A.K.; Nijboer, F.; Mellingerp, J.; Vaughan, T.M.; Pawelzik, H.; Schalk, G.; McFarland, D.J.; Birbaumer, N.; Wolpaw, J.R. Patients with ALS can use sensorimotor rhythms to operate a brain-computer interface. Neurology 2005, 64, 1775–1777. [Google Scholar] [CrossRef] [PubMed]
- Bai, O.; Lin, P.; Vorbach, S.; Floeter, M.; Hattori, N.; Hallett, M. A high performance sensorimotor beta rhythm-based brain–computer interface associated with human natural motor behavior. J. Neural Eng. 2008, 5, 24–35. [Google Scholar] [CrossRef] [PubMed]
- Bai, O.; Lin, P.; Huang, D.; Fei, D.; Floeter, M. Towards a user friendly brain-computer interface: Initial tests in als and pls patients. Clin. Neurophysiol. 2010, 121, 1293–1303. [Google Scholar] [CrossRef] [PubMed]
- Jacobs, L.; Bozian, D.; Heffner, R.R.; Barron, S.A. An eye movement disorder in amyotrophic lateral sclerosis. Neurology 1981, 31, 1282–1287. [Google Scholar] [CrossRef] [PubMed]
- Pfurtscheller, G.; Neuper, C. Motor imagery and direct brain-computer communication. Proc. IEEE 2001, 89, 1123–1134. [Google Scholar] [CrossRef]
- Pfurtscheller, G. Functional brain imaging based on ERD/ERS. Vis. Res. 2001, 41, 1257–1260. [Google Scholar] [CrossRef]
- Liao, K.; Xiao, R.; Conzalez, J.; Ding, L. Decoding individuals finger movements from one hand using human EEG signals. PLoS ONE 2014, 9, e85192. [Google Scholar] [CrossRef] [PubMed]
- Yong, X.; Menon, C. EEG classification of different imaginary movements with the same limb. PLoS ONE 2015, 10, e0121896. [Google Scholar] [CrossRef] [PubMed]
- Yang, B.; Li, H.; Wang, Q.; Zhang, Y. Subject-based feature extraction by using fisher WPD-CSP in brain-computer interfaces. Comput. Methods Progr. Biomed. 2016, 129, 21–28. [Google Scholar] [CrossRef] [PubMed]
- Nihei, K.; McKee, A.C.; Kowall, N.W. Patterns of neuronal degeneration in the motor cortex of amyotrophic lateral sclerosis patients. Acta Neuropathol. 1993, 86, 55–64. [Google Scholar] [CrossRef] [PubMed]
- Santhosh, J.; Bhatia, M.; Sahu, S.; Anand, S. Quantitative EEG analysis for assessment to ‘plan’ a task in amyotrophic lateral sclerosis patients: A study of executive functions (planning) in ALS patients. Cogn. Brain Res. 2004, 22, 59–66. [Google Scholar] [CrossRef] [PubMed]
- Kasahara, T.; Terasaki, K.; Ogawa, Y.; Ushiba, J.; Aramaki, H.; Masakado, Y. The correlation between motor impairments and event-related desynchronization during motor imagery in ALS patients. BMC Neurosci. 2012, 13, 66. [Google Scholar] [CrossRef] [PubMed]
- Hohmann, M.R.; Fomina, T.; Jayaram, V.; Widmann, N.; Förster, C.; Müller, V.; Hagen, J.; Synofzik, M.; Schölkopf, B.; Schöls, L.; et al. A cognitive brain-computer interface for patients with amyotrophic lateral sclerosis. In Proceedings of the International Joint Conference on IEEE Systems, Man, and Cybernetics, Hong Kong, China, 9–12 October 2015. [Google Scholar]
- Pfurtscheller, G.; Lopes da Silva, F.H. Event-related EEG/MEG synchronization and desynchronization: Basic principles. Clin. Neurophysiol. 1999, 110, 1842–1857. [Google Scholar] [CrossRef]
- Hashimoto, Y.; Ushiba, J. EEG-based classification of imaginary left and right foot movements using beta rebound. Clin. Neurophysiol. 2013, 124, 2153–2160. [Google Scholar] [CrossRef] [PubMed]
- Lulé, D.; Ludolph, A.C.; Kassubek, J. MRI-based functional neuroimaging in ALS: An update. Amyotrop. Lateral Scler. 2009, 10, 258–268. [Google Scholar] [CrossRef] [PubMed]
- Mandelbrot, B.B. The Fractal Geometry of Nature; Freeman Press: New York, NY, USA, 1982. [Google Scholar]
- Preissl, H.; Lutzenberger, W.; Pulvermüller, F.; Birbaumer, N. Fractal dimensions of short EEG time series in humans. Neurosci. Lett. 1997, 225, 77–80. [Google Scholar] [CrossRef]
- Hosseinifard, B.; Moradi, M.H.; Rostami, R. Classifying depression patients and normal subjects using machine learning techniques and nonlinear features from EEG signal. Comput. Methods Progr. Biomed. 2013, 109, 339–345. [Google Scholar] [CrossRef] [PubMed]
- Zappasodi, F.; Olejarczyk, E.; Marzetti, L.; Assenza, G.; Pizzella, V.; Tecchio, F. Fractal dimension of EEG activity senses neuronal impairment in acute stroke. PLoS ONE 2014, 9, e100199. [Google Scholar] [CrossRef] [PubMed]
- Boostani, R.; Moradi, M.H. A new approach in the BCI research based on fractal dimension as feature and Adaboost as classifier. J. Neural Eng. 2004, 1, 212–217. [Google Scholar] [CrossRef] [PubMed]
- Phothisonothai, M.; Nakagawa, M. Fractal-Based EEG Data Analysis of Body Parts Movement imagery tasks. J. Physiol. Sci. 2007, 57, 217–226. [Google Scholar] [CrossRef] [PubMed]
- Phothisonothai, M.; Nakagawa, M. EEG-based classification of motor imagery tasks using fractal dimension and neural network for brain-computer interface. IEICE Trans. Inf. Syst. 2008, E91-D, 44–53. [Google Scholar] [CrossRef]
- Güclü, U.; Güclütürk, Y.; Loo, C.K. Evaluation of fractal dimension estimation methods for feature extraction in motor imagery based brain computer interface. Proced. Comput. Sci. 2011, 3, 590–594. [Google Scholar] [CrossRef]
- Hsu, W.Y. Fuzzy Hopfield neural network clustering for single-trial motor imagery EEG classification. Expert Syst. Appl. 2012, 39, 1055–1061. [Google Scholar] [CrossRef]
- Aguilar, J.M.; Castillo, J.; Elias, D. EEG signals processing based on fractal dimension features and classified by neural network and support vector machine in motor imagery for a BCI. In Proceedings of the VI Latin American Congress on Biomedical Engineering CLAIB, Paraná, Argentina, 29–31 October 2014. [Google Scholar] [CrossRef]
- Peng, C.K.; Mietus, J.; Hausdorff, J.M.; Havlin, S.; Stanley, H.E.; Goldberger, A.L. Long-range anticorrelations and non-Gaussian behavior of the heartbeat. Phys. Rev. Lett. 1993, 70, 1343–1346. [Google Scholar] [CrossRef] [PubMed]
- Katz, M.J. Fractals and the analysis of waveforms. Comput. Biol. Med. 1988, 18, 145–156. [Google Scholar] [CrossRef]
- Higuchi, T. Approach to an irregular time series on the basis of the fractal theory. Phys. D Nonlinear Phenom. 1988, 31, 277–283. [Google Scholar] [CrossRef]
- Hurst, H.E. Long-term storage capacity of reservoirs. Am. Soc. Civil Eng. 1951, 116, 770–799. [Google Scholar]
- Grassberger, P.; Procaccia, I. Measuring the strangeness of attractors. Phys. D Nonlinear Phenom. 1983, 9, 189–208. [Google Scholar] [CrossRef]
- Wang, S.; Liu, C.L.; Zheng, L. Feature selection by combing Fisher criterion and principal feature analysis. In Proceedings of the International Joint Conference on Machine Learning and Cybernetics, Hong Kong, China, 19–22 August 2007; pp. 1149–1154. [Google Scholar]
- Fang, L.; Zhao, H.; Wang, P.; Yu, M.; Yan, J.; Cheng, W.; Chen, P. Feature selection method based on mutual information and class separability for dimension reduction in multidimensional time series for clinical data. Biomed. Signal Process. Control 2015, 21, 82–89. [Google Scholar] [CrossRef]
- Naseer, N.; Hong, K.-S. Classification of functional near-infrared spectroscopy signals corresponding to right- and left-wrist motor imagery for development of a brain-computer interface. Neurosci. Lett. 2013, 553, 84–89. [Google Scholar] [CrossRef] [PubMed]
- Guo, L.; Rivero, D.; Dorado, J.; Munteanu, C.R.; Pazos, A. Automatic feature extraction using genetic programming: An application to epileptic EEG classification. Expert Syst. Appl. 2011, 38, 10425–10436. [Google Scholar] [CrossRef]
- Naseer, N.; Noori, F.M.; Qureshi, N.K.; Hong, K.-S. Determining optimal feature-combination for LDA classification of functional near-infrared spectroscopy signals in brain-computer interface application. Front. Hum. Neurosci. 2016, 10, 237. [Google Scholar] [CrossRef] [PubMed]
- Cedarbaum, J.M.; Stambler, N.; Malta, E.; Fuller, C.; Hilt, D.; Thurmond, B.; Nakanishi, A. The ALSFRS-R: A revised ALS functional rating scale that incorporates assessments of respiratory function. J. Neurol. Sci 1999, 169, 13–21. [Google Scholar] [CrossRef]
- Roche, J.C.; Rojas-Garcia, R.; Scott, K.M.; Scotton, W.; Ellis, C.E.; Burman, R.; Wijesekera, L.; Turner, M.R.; Leigh, P.N.; Shaw, C.E.; et al. A proposed staging system for amyotrophic lateral sclerosis. Brain 2012, 135, 847–852. [Google Scholar] [CrossRef] [PubMed]
- Balendra, R.; Jones, A.; Jivraj, N.; Steen, I.N.; Young, C.A.; Shaw, P.J.; Turner, M.R.; Leigh, P.N.; Al-Chalabi, A. Use of clinical staging in amyotrophic lateral sclerosis for phase 3 clinical trials. J. Neurol. Neurosurg. Psychiatry 2015, 86, 45–49. [Google Scholar] [CrossRef] [PubMed]
- Hsu, W.C.; Lin, L.F.; Chou, C.W.; Hsiao, Y.T.; Liu, Y.H. EEG classification of imaginary lower limb stepping movements based on fuzzy support vector machine with kernel-induced membership function. Int. J. Fuzzy Syst. 2017, 19, 566–579. [Google Scholar] [CrossRef]
- Esteller, R.; Vachtsevanos, G.; Echauz, J. A comparison of waveform fractal dimension. IEEE. Trans. Circuits Syst. 2001, 48, 177–183. [Google Scholar] [CrossRef]
- Ramoser, H.; Müller-Gerking, J.; Pfurtscheller, G. Optimal spatial filtering of single trial EEG during imagined hand movement. IEEE Trans. Rehab. Eng. 2000, 8, 441–446. [Google Scholar] [CrossRef]
- Tang, Z.; Sun, S.; Zhang, S.; Chen, Y.; Li, C.; Chen, S. A Brain-Machine Interface Based on ERD/ERS for an Upper-Limb Exoskeleton Control. Sensors 2016, 16, 2050. [Google Scholar] [CrossRef] [PubMed]
- Zhang, Y.; Liu, B.; Ji, X.; Huang, D. Classification of EEG Signals Based on Autoregressive Model and Wavelet Packet Decomposition. Neural Process. Lett. 2017, 45, 365. [Google Scholar] [CrossRef]
- Blankertz, B.; Tomioka, R.; Lemm, S.; Kawanabe, M.; Müller, K.-R. Optimizing spatial filters for robust EEG single-trial analysis. IEEE Sign. Process. Mag. 2008, 25, 41–56. [Google Scholar] [CrossRef]
- Cecotti, H.; Gräser, A. Convolutional Neural Networks for P300 Detection with Application to Brain-Computer Interfaces. IEEE Trans. Pattern Anal. Mach. Intell. 2011, 33, 433–445. [Google Scholar] [CrossRef] [PubMed]
- Qian, K.; Nikolov, P.; Huang, D.; Fei, D.Y.; Chen, X.; Bai, O. A motor imagery-based online interactive brain-controlled switch: Paradigm development and preliminary test. Clin. Neurophysiol. 2010, 121, 1304–1313. [Google Scholar] [CrossRef] [PubMed]
- Bennasar, M.; Hicks, Y.; Setchi, R. Feature selection using joint mutual information maximization. Expert Syst. Appl. 2015, 42, 8520–8532. [Google Scholar] [CrossRef]
- Gonzalez, A.; Nambu, I.; Hokari, H.; Wada, Y. EEG channel selection using particle swarm optimization for the classification of auditory event-related potentials. Sci. World J. 2014, 2014, 350270. [Google Scholar] [CrossRef]
Patient | Age | Sex | TSSO 1 | ALSFRS-R 2 | Limb Function | Speech | Stage |
---|---|---|---|---|---|---|---|
Sub1 | 59 | M | 11 y | 16 | No | Yes | 4A & 4B |
Sub2 | 65 | M | 7 y | 7 | Minimal | Slow | 4A & 4B |
Sub3 | 64 | M | 20 mo | 12 | Minimal | Yes | 4B |
Sub4 | 39 | F | 44 mo 3 | 24 | Minimal | Slow | 4A |
Sub5 | 47 | M | 11 y | 5 | No | Slow | 4A & 4B |
sub | 8–12 Hz BP | 12–16 Hz BP | 16–20 Hz BP | 20–24 Hz BP | 24–28 Hz BP | GPFD | HFD |
---|---|---|---|---|---|---|---|
1 | T5 (0.1959) | TP7 (0.4776) | TP7 (0.8866) | TP7 (0.6420) | TP7 (1.0184) | TP7 (2.9226) | TP7 (2.1797) |
CP3 (0.1868) | T3 (0.1518) | T4 (0.7818) | T4 (0.3990) | T4 (0.7183) | CPz (0.4713) | FT8 (0.6036) | |
T3 (0.1326) | Pz (0.1195) | FP2 (0.3299) | FP2 (0.3753) | FP1 (0.2776) | Cz (0.3829) | CPz (0.2593) | |
O2 (0.1236) | FT7 (0.1073) | O2 (0.1559) | T3 (0.1249) | O2 (0.2209) | FT8 (0.3811) | O1 (0.2634) | |
TP7 (0.1117) | O2 (0.0767) | T3 (0.1158) | O2 (0.1170) | FP2 (0.1556) | CP3 (0.2656) | Pz (0.1515) | |
2 | Pz (0.1162) | T6 (0.5264) | FCz (0.4075) | F7 (0.3662) | FT8 (0.4431) | TP8 (0.5208) | O2 (0.1715) |
P3 (0.1137) | FC4 (0.4914) | FC4 (0.3178) | FP1 (0.3457) | C4 (0.2878) | T3 (0.3017) | Oz (0.1489) | |
CP3 (0.0775) | F4 (0.2750) | F7 (0.2691) | FCz (0.3146) | F8 (0.2746) | FCz (0.2071) | FP1 (0.1249) | |
CPz (0.0696) | FC3 (0.1727) | F4 (0.2173) | Fz (0.1943) | FT7 (0.2606) | C3 (0.2057) | O1 (0.1245) | |
P4 (0.0657) | Cz (0.1598) | FC3 (0.2071) | F3 (0.1711) | T4 (0.2472) | FT7 (0.1684) | T5 (0.0864) | |
3 | FT8 (0.1150) | T4 (1.1110) | T4 (1.0073) | T4 (1.8131) | T4 (1.4962) | C4 (22.1579) | C4 (9.5388) |
Pz (0.1117) | TP8 (0.6196) | TP8 (0.9739) | TP8 (1.4694) | TP8 (1.1764) | CP4(17.7842) | P4 (6.5059) | |
Oz (0.0868) | T3 (0.4998) | T3 (0.7371) | F3 (1.1074) | FC3 (1.0906) | P4 (13.0471) | CP4 (5.9313) | |
F8 (0.0713) | TP7 (0.2949) | CP3 (0.7054) | T3 (0.9613) | TP7 (1.0091) | T6 (9.2076) | T6 (4.1303) | |
F7 (0.0495) | FP1 (0.2803) | FP1 (0.6020) | CP4 (0.8266) | F3 (0.9516) | FC4 (6.6343) | FC4 (3.3493) | |
4 | FT7 (0.1673) | F8 (0.1450) | TP7 (0.3527) | F7 (1.3163) | F8 (0.8605) | T5 (2.3012) | T5 (0.7612) |
T3 (0.0570) | Oz (0.1393) | F8 (0.2183) | F8 (0.4023) | F7 (0.1334) | FC4 (1.0024) | FC4 (0.5805) | |
CP3 (0.0554) | F7 (0.1398) | O1 (0.1501) | O1 (0.3579) | O1 (0.0926) | Pz, (0.8049) | P3 (0.5124) | |
C3 (0.0505) | O1 (0.1380) | F7 (0.1210) | Oz (0.3201) | TP7 (0.0856) | P3 (0.7944) | Oz (0.5050) | |
TP7 (0.0458) | O2 (0.1350) | CP3 (0.1209) | TP7 (0.3143) | Oz (0.0824) | Fz (0.7880) | Pz (0.4517) | |
5 | FP2 (1.0668) | FP2 (1.4845) | FP2 (1.5629) | FP2 (1.4995) | T4 (1.4758) | TP8 (4.0383) | TP8 (2.5587) |
FP1 (0.9840) | TP8 (0.9154) | TP8 (1.2318) | F7 (1.4014) | FP2 (1.4747) | T4 (2.5665) | P4 (2.1619) | |
FT7 (0.8580) | FP1 (0.7829) | FP1 (1.1796) | TP8 (1.1532) | FP1 (1.3258) | P4 (2.4334) | T4, (1.9509) | |
F8 (0.8472) | F8 (0.3947) | T4 (1.1297) | FP1 (1.0076) | FT7 (1.3033) | C4 (1.6472) | C4 (1.7615) | |
F3 (0.8374) | F7 (0.3658) | CP4 (0.9431) | FT7 (0.8936) | F7 (1.2617) | CP4 (1.3791) | C3 (1.5695) |
sub | 8–12 Hz BP | 12–16 Hz BP | 16–20 Hz BP | 20–24 Hz BP | 24–28 Hz BP | GPFD | HFD |
---|---|---|---|---|---|---|---|
1 | TP7 (0.1351) | TP7 (0.2906) | TP7 (0.6980) | TP7 (0.3016) | FT8 (0.3469) | TP7 (1.3889) | TP7 (2.1183) |
TP8 (0.0670) | FT8 (0.1265) | TF8 (0.1432) | FT8 (0.1337) | TP7 (0.1313) | CPz (0.6460) | P3 (0.8449) | |
P4 (0.0611) | FT7 (0.0981) | T4 (0.0754) | FT7 (0.0902) | FP7 (0.0779) | O1 (0.5262) | CPz (0.7107) | |
O2 (0.0500) | T5 (0.0529) | O2 (0.0454) | T4 (0.0673) | T4 (0.0747) | P3 (0.5262) | CP3 (0.4847) | |
T4 (0.0467) | T3 (0.0526) | FT7 (0.0333) | T5 (0.0536) | T5 (0.0607) | T5 (0.5088) | Pz (0.3650) | |
2 | P3 (0.1425) | FC3 (0.1658) | FC3 (0.1317) | FC3 (0.1825) | FC3 (0.4775) | TP8 (0.7588) | TP8 (0.2160) |
Pz (0.1316) | C3 (0.1503) | C3 (0.1289) | F7 (0.1088) | FC4 (0.2909) | T6 (0.2034) | FP1 (0.0863) | |
CP3 (0.0870) | T6 (0.1500) | T6 (0.1282) | C3 (0.0737) | F8 (0.2580) | FT8 (0.1806) | P3 (0.0840) | |
T6 (0.0866) | TP8 (0.1451) | F4 (0.1156) | FT7 (0.696) | F4 (0.2351) | F8 (0.1679) | O1 (0.0815) | |
O2 (0.0418) | F4 (0.1300) | FC4 (0.1100) | FCz (0.0657) | FT8 (0.2008) | F4 (0.1531) | T5 (0.0803) | |
3 | Pz (0.0785) | T4 (0.8316) | T4 (1.4278) | C4 (1.5888) | T4 (1.4773) | CP3 (7.5453) | P3 (1.5629) |
CPz (0.0390) | T3 (0.5088) | C4 (0.8031) | T4 (1.1867) | F7 (0.7968) | P3 (7.0727) | TP8 (1.2314) | |
Oz (0.0335) | F7 (0.1977) | F7 (0.1428) | TP8 (0.1362) | TP8 (0.2109) | T5 (4.5642) | CP3 (0.9180) | |
Cz (0.0297) | FP1 (0.1459) | TP8 (0.1123) | CP4 (0.1117) | CP4 (0.1894) | FT8 (2.6046) | F7 (0.5414) | |
T4 (0.0235) | TP8 (0.1381) | CP4 (0.0776) | F7 (0.0907) | C3 (0.1807) | T3 (1.9923) | O1 (0.5411) | |
4 | FT7 (0.0542) | FT8 (0.2108) | T5 (0.0796) | T5 (0.2058) | T5 (0.2268) | Fz (0.4379) | FC4 (0.2708) |
T3 (0.0472) | T4 (0.1761) | O1 (0.0793) | P3 (0.1353) | Fz (0.1363) | FC4 (0.3620) | Pz (0.2141) | |
T5 (0.0353) | Oz (0.1710) | FT8 (0.0515) | Pz (0.1061) | FP1 (0.0745) | Cz (0.3444) | CPz (0.1951) | |
P3 (0.0339) | O1 (0.1487) | Oz (0.0492) | CPz (0.0985) | CP3 (0.0745) | FCz (0.3265) | P3 (0.1744) | |
CP3 (0.0297) | F7 (0.1100) | F8 (0.0412) | Oz (0.0649) | F7 (0.0534) | Oz (0.2655) | CP3 (0.1717) | |
5 | FP2 (0.7118) | FP2 (1.0015) | FP2 (0.9555) | CP3 (1.2064) | F7 (2.2860) | T4 (3.8428) | T4 (2.2796) |
FP1 (0.6990) | FP1 (0.7181) | F3 (0.8529) | C3 (1.0922) | FT7 (1.4914) | TP8 (2.7892) | TP8 (2.0638) | |
F7 (0.6279) | T3 (0.4920) | FP1 (0.8094) | FP2 (1.0032) | FP1 (1.3404) | P4 (1.9385) | C3, (1.8043) | |
F8 (0.5995) | TP8 (0.4740) | T4 (0.8026) | TP8 (0.9364) | T4 (1.2849) | C4 (1.8107) | C4 (1.6101) | |
F3 (0.5892) | FT7 (0.2995) | C3 (0.7651) | FC3 (0.8981) | C3 (1.0573) | CP4 (1.2856) | P4 (1.6054) |
sub | 8–12 Hz BP | 12–16 Hz BP | 16–20 Hz BP | 20–24 Hz BP | 24–28 Hz BP | GPFD | HFD |
---|---|---|---|---|---|---|---|
1 | TP8 (0.0983) | T4 (0.5434) | T4 (0.7272) | T4 (0.7545) | T4 (0.8082) | CP4 (2.7924) | TP7 (2.8712) |
P4 (0.0652) | T3 (0.3338) | F8 (0.6317) | CP3 (0.4574) | T3 (0.6576) | TP7 (1.7434) | CP4 (2.4801) | |
T4 (0.0433) | TP7 (0.2698) | TP7 (0.4421) | T3 (0.4488) | TP7 (0.3488) | C4 (1.5566) | T4 (1.6187) | |
C3 (0.0363) | C4 (0.0824) | FC4 (0.3896) | T6 (0.4156) | FT7 (0.2678) | FC4 (1.3140) | C4 (1.4083) | |
TP7 (0.0355) | FT7 (0.0397) | Fz (0.3703) | FC3 (0.3544) | C4 (0.2244) | T4 (1.2547) | P4 (1.3327) | |
2 | Pz (0.2704) | T6(0.3515) | F4 (0.4160) | FP2 (0.3385) | T3 (0.5032) | FT8 (0.5611) | TP8 (0.1155) |
P3 (0.2291) | T3(0.2691) | F7 (0.3449) | FP1 (0.3038) | FP2 (0.4180) | T3 (0.1328) | T5 (0.0859) | |
CPz (0.1975) | TP8 (0.2269) | T6 (0.3429) | TP8 (0.2709) | FT7 (0.3611) | T5 (0.1255) | O1 (0.0787) | |
CP3 (0.1885) | F4 (0.2152) | FC4 (0.3394) | Cz (0.2687) | T6 (0.3563) | FP2 (0.878) | P3 (0.0784) | |
P4 (0.0657) | O2 (0.2068) | FP2 (0.3149) | FCz (0.0657) | F4 (0.3550) | TP7 (0.0669) | P4 (0.0756) | |
3 | T4 (0.2269) | F7 (2.7868) | F7 (2.8897) | F7 (1.5145) | F7 (2.1683) | C4 (25.2850) | C4 (8.3864) |
TP8 (0.0641) | TP8 (0.2692) | T4 (1.3730) | C4 (1.4156) | C3 (1.6969) | CP4(20.6432) | CP4 (5.1367) | |
F7 (0.0634) | TP7 (0.2405) | C4 (0.8541) | T4 (1.3621) | FC3 (1.4402) | FC3(13.8415) | FC4 (4.9778) | |
FT8 (0.0419) | T4 (0.2150) | C3 (0.8371) | C3 (1.3085) | C4 (1.2559) | C3 (11.2478) | P4 (4.9573) | |
FP1 (0.0378) | F3 (0.1817) | FC3 (0.6426) | FT8 (1.2335) | T4 (1.2525) | F3 (10.6492) | FC3 (4.7763) | |
4 | P3 (0.0664) | Cz (0.0809) | CP3 (0.2001) | Pz (0.3259) | Cz (0.0998) | FCz (0.9769) | Fz (0.4666) |
CP3 (0.0646) | FCz (0.0619) | Cz (0.1675) | CPz (0.3119) | Pz (0.0996) | Fz (0.6215) | Oz (0.3230) | |
T5 (0.0599) | Fz (0.0484) | P3 (0.1618) | P4 (0.2125) | P3 (0.0971) | Pz, (0.4877) | CPz (0.2814) | |
C3 (0.0595) | FT7 (0.0461) | CPz (0.1575) | O1 (0.2026) | CPz (0.0958) | FC4 (0.4635) | Pz (0.2349) | |
O1 (0.0465) | F8 (0.0436) | Pz (0.1416) | Oz (0.1926) | O1 (0.0922) | TP7 (0.4519) | TP7 (0.2339) | |
5 | FP2 (1.0830) | FP2 (0.6187) | FP1 (1.2658) | FT7 (0.9768) | FT7 (1.1204) | P4 (1.8536) | C3 (1.4460) |
FP1 (1.0581) | T3 (0.4771) | F3 (1.0748) | T4 (0.6783) | F7 (0.5620) | T3 (1.0703) | T3 (1.3716) | |
F7 (1.0344) | FP1 (0.4640) | F7 (1.0565) | F3 (0.5796) | T4 (0.4857) | FT8 (0.9700) | T4 (1.3633) | |
FT7 (1.0052) | F8 (0.4436) | FT7 (0.8939) | F7 (0.5622) | T3 (0.4718) | TP7 (0.8080) | FC3 (1.3311) | |
F8 (1.0001) | F7 (0.3675) | T4 (0.8543) | T3 (0.4731) | F3 (0.4100) | FT7 (0.7736) | TP7 (1.2434) |
Sub | Features | Top Five Channels | Worst Five Channels | ||||
---|---|---|---|---|---|---|---|
RH vs. R | LH vs. R | LF vs. R | RH vs. R | LH vs. R | LF vs. R | ||
1 | BP (8–12 Hz) | 61.25 | 68.75 | 56.25 | 55.00 | 53.75 | 57.50 |
BP (12–16 Hz) | 93.75 | 80.00 | 92.50 | 35.00 | 60.00 | 53.75 | |
BP (16–20 Hz) | 93.75 | 83.75 | 87.50 | 58.75 | 51.25 | 56.25 | |
BP (20–24 Hz) | 82.50 | 81.25 | 85.00 | 52.50 | 47.50 | 63.75 | |
BP (24–28 Hz) | 92.50 | 78.75 | 90.00 | 47.50 | 57.50 | 56.25 | |
GPFD | 98.75 | 97.50 | 97.50 | 65.00 | 58.75 | 62.50 | |
HFD | 97.50 | 97.50 | 97.50 | 55.00 | 52.50 | 56.25 | |
2 | BP (8–12 Hz) | 65.00 | 57.50 | 66.25 | 50.00 | 51.25 | 57.50 |
BP (12–16 Hz) | 66.25 | 60.00 | 78.7 | 56.25 | 43.75 | 45.00 | |
BP (16–20 Hz) | 70.00 | 61.25 | 80.00 | 56.25 | 52.50 | 73.75 | |
BP (20–24 Hz) | 66.25 | 57.50 | 62.50 | 80.00 | 51.25 | 67.50 | |
BP (24–28 Hz) | 75.00 | 65.00 | 82.50 | 50.00 | 56.25 | 62.50 | |
GPFD | 81.25 | 77.50 | 72.50 | 60.00 | 48.75 | 60.00 | |
HFD | 66.25 | 75.00 | 77.50 | 60.00 | 63.75 | 68.75 | |
3 | BP (8–12 Hz) | 63.75 | 60.00 | 65.00 | 55.00 | 52.50 | 56.25 |
BP (12–16 Hz) | 90.00 | 86.25 | 97.50 | 47.50 | 50.00 | 48.75 | |
BP (16–20 Hz) | 95.00 | 86.25 | 98.75 | 45.00 | 52.50 | 67.50 | |
BP (20–24 Hz) | 97.50 | 96.25 | 93.75 | 48.75 | 41.25 | 53.75 | |
BP (24–28 Hz) | 97.50 | 97.50 | 100.00 | 51.25 | 48.75 | 55.00 | |
GPFD | 100.00 | 100.00 | 100.00 | 81.25 | 52.50 | 96.25 | |
HFD | 100.00 | 95.00 | 100.00 | 76.25 | 55.00 | 93.75 | |
4 | BP (8–12 Hz) | 55.00 | 58.75 | 56.25 | 37.50 | 51.25 | 42.50 |
BP (12–16 Hz) | 61.25 | 68.75 | 60.00 | 46.25 | 48.75 | 40.00 | |
BP (16–20 Hz) | 81.25 | 68.75 | 56.25 | 53.75 | 58.75 | 51.25 | |
BP (20–24 Hz) | 90.00 | 55.00 | 70.00 | 65.00 | 68.75 | 63.75 | |
BP (24–28 Hz) | 90.00 | 73.75 | 51.25 | 60.00 | 67.50 | 53.75 | |
GPFD | 88.75 | 77.50 | 86.25 | 61.25 | 60.00 | 56.25 | |
HFD | 86.25 | 60.00 | 76.25 | 61.25 | 48.75 | 47.50 | |
5 | BP (8–12 Hz) | 80.00 | 78.75 | 87.50 | 71.25 | 68.75 | 76.25 |
BP (12–16 Hz) | 98.75 | 87.50 | 90.00 | 47.50 | 52.50 | 55.00 | |
BP (16–20 Hz) | 96.25 | 96.25 | 91.25 | 70.00 | 72.50 | 82.50 | |
BP (20–24 Hz) | 96.25 | 97.50 | 95.00 | 66.25 | 77.50 | 71.25 | |
BP (24–28 Hz) | 98.75 | 96.25 | 93.75 | 66.25 | 81.25 | 78.75 | |
GPFD | 97.50 | 98.75 | 96.25 | 60.00 | 78.75 | 53.75 | |
HFD | 98.75 | 96.25 | 93.75 | 77.50 | 87.50 | 71.25 | |
Ave. | BP (8–12 Hz) | 65.00 | 64.75 | 66.25 | 53.75 | 55.50 | 58.00 |
BP (12–16 Hz) | 82.00 | 76.50 | 83.75 | 46.50 | 51.00 | 48.50 | |
BP (16–20 Hz) | 87.25 | 79.25 | 82.75 | 56.75 | 57.50 | 66.25 | |
BP (20–24 Hz) | 86.50 | 77.50 | 81.25 | 62.50 | 57.25 | 64.00 | |
BP (24–28 Hz) | 90.75 | 82.25 | 83.50 | 55.00 | 62.25 | 61.25 | |
GPFD | 93.25 | 90.25 | 90.50 | 65.50 | 59.75 | 65.75 | |
HFD | 89.75 | 84.75 | 89.00 | 66.00 | 61.50 | 67.50 |
Channels (num.) | K-NN Classifier | LDA Classifier | ||||
---|---|---|---|---|---|---|
RH vs. R | LH vs. R | LF vs. R | RH vs. R | LH vs. R | LF vs. R | |
All (30) | 95.25 | 88.25 | 92.25 | 95.25 | 90.50 | 93.50 |
Top five (5) | 93.25 | 90.25 | 90.50 | 93.50 | 88.75 | 91.75 |
The best (1) | 90.50 | 88.00 | 87.25 | 91.00 | 88.50 | 88.50 |
Methods | 30 Channels | Top Five Channels | Top One Channel | ||||||
---|---|---|---|---|---|---|---|---|---|
RH-R | LH-R | LF-R | RH-R | LH-R | LF-R | RH-R | LH-R | LF-R | |
GPFD | 95.25 | 88.25 | 92.25 | 93.25 | 90.25 | 90.50 | 90.50 | 88.00 | 87.25 |
AR | 75.75 | 74.50 | 74.75 | 65.25 | 69.75 | 71.50 | 72.50 | 70.25 | 69.00 |
CSP (mu) | 83.00 | 75.75 | 79.50 | 64.25 | 62.00 | 72.50 | N/A | N/A | N/A |
CSP (beta) | 92.25 | 90.25 | 91.50 | 88.75 | 86.00 | 78.00 | N/A | N/A | N/A |
© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
Share and Cite
Liu, Y.-H.; Huang, S.; Huang, Y.-D. Motor Imagery EEG Classification for Patients with Amyotrophic Lateral Sclerosis Using Fractal Dimension and Fisher’s Criterion-Based Channel Selection. Sensors 2017, 17, 1557. https://doi.org/10.3390/s17071557
Liu Y-H, Huang S, Huang Y-D. Motor Imagery EEG Classification for Patients with Amyotrophic Lateral Sclerosis Using Fractal Dimension and Fisher’s Criterion-Based Channel Selection. Sensors. 2017; 17(7):1557. https://doi.org/10.3390/s17071557
Chicago/Turabian StyleLiu, Yi-Hung, Shiuan Huang, and Yi-De Huang. 2017. "Motor Imagery EEG Classification for Patients with Amyotrophic Lateral Sclerosis Using Fractal Dimension and Fisher’s Criterion-Based Channel Selection" Sensors 17, no. 7: 1557. https://doi.org/10.3390/s17071557